Introduction

Idiopathic membranous nephropathy (iMN) is the most common cause of primary nephrotic syndrome, especially among the elderly [1]. About 30% of patients develop end-stage renal disease (ESRD) within 20 years [2, 3]. There is no standard treatment for iMN. The Kidney Disease Improving Global Outcomes (KDIGO) guidelines recommend that the initial therapy should be supportive and that treatment should only be started once the urinary protein excretion level reaches ≥4 g/gCre. The guidelines also recommend that supportive therapy be performed for at least the first 6 months after diagnosis [4]. The KDIGO guidelines recommend that the first-line immunosuppressive therapy should be cytotoxic drugs, such as cyclophosphamide, plus glucocorticoids, or a calcineurin inhibitor [5, 6]. Recent reports have indicated that rituximab and tacrolimus are effective for achieving a complete remission of proteinuria [7,8,9]. iMN may spontaneously recur in approximately 30% of patients. In Japan, corticosteroids are administered after diagnosis [10,11,12], and steroid monotherapy has been reported to induce remission [2, 13]. However, since iMN is common among the elderly, measures to avoid complications arising from immunosuppression are needed, since steroid therapy can cause adverse effects, such as infection, osteoporosis, and impaired glucose tolerance.

In the present study, we compared patients with and those without initial steroid therapy during the first 6 months after diagnosis, focusing on a 40% reduction of eGFR and rates of complete remission (CR).

Materials and methods

Patients

Data were extracted from our hospital records for patients over 16 years of age who had undergone a renal biopsy between 1993 and 2013. A total of 1970 renal biopsies were performed during this period, and 151 (7.7%) cases were diagnosed as having membranous nephropathy. Seventeen cases that had undergone a re-biopsy and four cases that were diagnosed as having some other form of glomerulonephritis or secondary membranous nephropathy were excluded. Among the remaining 134 cases, a total of 114 patients were registered as having iMN, and 97 of these patients had a urinary protein level of ≥1 g/gCre at the time of their renal biopsy. In addition, cases that received other immunosuppressive drugs (such as cytotoxic drugs or calcineurin inhibitors), within 6 months of biopsy, were excluded. Finally, 83 patients (Group S1, Group H1) were included in the analysis.

Study design

The patients were divided into those who received initial steroid therapy within 6 months after diagnosis (Group S1, n = 52) and those who received supportive therapy (Group H1, n = 31), and the renal outcomes and the rates of remission were retrospectively examined using clinical and pathological data.

In addition, propensity score matching was used to compare two groups of 20 patients each (Group S2, Group H2). Furthermore, we divided 37 cases with a urinary protein level of 1–3.5 g/gCre (Group S3, n = 18; Group H3, n = 19) and 46 cases with a urinary protein level of more than 3.5 g/gCre (Group S4, n = 34; Group H4, n = 12) into two groups according to their and compared their outcomes.

Treatment

In the steroid therapy group, the initial dose of prednisone was 0.8 mg/kg/day, and the doctors in charge of the patient decided when to begin dose reduction.

For cases with a urinary protein level higher than 3.5 g/gCre, or uncontrollable edema, we suggested therapy. The decision to initiate treatment was made by the doctors in charge of the patient decided whether to continue receiving treatment at some point during the first 6 months of treatment.

For the patients in Group S1, whose urinary protein level had increased at 6 months after biopsy and who had not experienced any adverse effects from the initial steroid therapy, the steroid dose was increased for some patients; for other patients who had an increased risk of diabetes and infection, the attending doctors prescribed the use of other immunosuppressive drugs. Supportive therapy included the administration of angiotensin-converting enzyme inhibitors (ACEIs) or angiotensin-receptor blockers (ARBs) and antiplatelet and anticholesterol agents. The treatment used for Group H1 patients after 6 months was selected based on the patients’ background, such as the presence of diabetes and their age. After 6 months, for patients with a urinary protein level over 3.5 g/gCre, we increased the dose of RASS inhibitors, and selected steroid therapy as the first-line treatment; however, two patients were treated with other immunosuppressants because of depression or their refusal to undergo steroid therapy.

Study endpoints

Proteinuria was used to determine the therapeutic outcome. According to the 2011 Guidelines for the Treatment of Nephrotic Syndrome [11], CR is regarded as a proteinuria level of <0.3 g/gCre.

The primary study endpoint was a 40% reduction in the estimated glomerular filtration rate (eGFR), and the secondary endpoint was the achievement of remission. The outcomes of both groups were examined at 5 and 10 years.

Statistical analyses

The statistical analyses were performed using the Chi-square test, the Pearson and Spearman correlation coefficients for analyses of parametric and nonparametric data, and the Kaplan–Meier method. Data were given as proportions, medians, and interquartile ranges. All the statistical analyses were performed using the JMP software package version Pro11.2, and P values of less than 0.05 were considered to indicate a significant linear or nonlinear trend. A Cox regression analysis was used to determine the incidences and hazard ratios (HR) of a 40% reduction in the eGFR and a CR.

To account for differences between patients with and those without steroid therapy, a one-to-one propensity matching analysis was also performed. The initial variables included were proteinuria, serum total protein, albumin, total cholesterol, and immunoglobulin levels. The propensity score was used as the sole criterion for matching pairs of patients. A matched pair was formed when a patient selected from Group S1 had a propensity score that was nearest to that of a patient in Group H1.

Results

The baseline characteristics of the 83 patients with a urinary protein level of ≥1 g/gCre are listed in Table 1. Proteinuria and the serum total protein, albumin, total cholesterol, and IgG levels showed significant differences between the two groups. The incidences of morbidities, such as hypertension and diabetes mellitus, did not differ significantly between the two groups. Histopathologically, the presence of global sclerosis and tubulointerstitial fibrosis and deposition of C3 and IgG4 did not differ significantly between the two groups; however, the incidences of focal segmental sclerosis (FGS) were 29 and 6%, respectively, and these values were significantly different (P = 0.01) (Tables 1, 2, 3, 4).

Table 1 Baseline characteristic (proteinuria ≥1 g/gCre)
Full size table
Table 2 Baseline characteristic, after matching in proteinuria ≥1 g/gCre
Full size table
Table 3 Baseline characteristic (proteinuria 1–3.5 g/gCre)
Full size table
Table 4 Baseline characteristic (proteinuria ≥3.5 g/gCre)
Full size table

Fourteen patients (27%) in Group S1 received immunosuppressive drugs (mostly cyclosporine) within 6 months of biopsy. Seven patients (23%) in Group H1 received steroid therapy at some timepoint at least 6 months after the biopsy, and immunosuppressive drugs were added to the treatments of four patients (13%) in Group H1. The rates of relapse in the two groups were different (P = 0.001). Two patients (3.9%) in Group S1 developed ESRD, and three patients (5.8%) in Group S1 died because of myocardial infarction, alcoholic liver dysfunction, or malignant lymphoma. One patient (3.2%) in Group H1 died because of hepatocellular carcinoma. Six patients (12%) in Group S1 developed diabetes mellitus, but none of the patients in Group H1 developed diabetes (Tables 5, 6, 7, 8).

Table 5 After 6 months from initial therapy (proteinuria ≥1 g/gCre)
Full size table
Table 6 After 6 months from initial therapy, after matching in proteinuria ≥1 g/gCre
Full size table
Table 7 After 6 months from initial therapy (proteinuria 1–3.5 g/gCre)
Full size table
Table 8 After 6 months from initial therapy (proteinuria ≥3.5 g/gCre)
Full size table

Figure 1 shows the Kaplan–Meier plots for the incidences of a 40% reduction in the eGFR and CR. A 40% reduction in the eGFR at the end of 5 years was observed in 18% of the patients in Group S1 and 17% of the patients in Group H1 (P = 0.93), while the incidences at 10 years were 43% for Group S1 and 50% for Group H1 (P = 0.88). A CR was observed at the end of 5 years in 58% of the patients in Group S1 and 32% of the patients in Group H1 (P = 0.02), and at the end of 10 years in 65% of the patients in Group S1 and 39% of the patients in Group H1 (P = 0.02) (Fig. 1a, b). Cox regression analyses for the incidence of a 40% reduction in the eGFR showed significant differences according to the presence of segmental sclerosis (HR 6.21; 95% confidence interval 1.36–30.12; P = 0.02) and diabetes mellitus (HR 4.61; 95% confidence interval 1.01–19.98; P = 0.04) (Table 9). However, the presence of steroid use and patient age (≥65 years), sex, eGFR (<45 mL/min/1.73 m2), proteinuria (≥3.5 g/gCre), serum albumin level (<2.0 g/dL), hypertension, global sclerosis (<20%), and tubulointerstitial fibrosis were not selected as significant factors.

Fig. 1
figure 1

Incidences of complete remission (CR) and a 40% reduction in the estimated glomerular filtration rate (eGFR) in patients receiving steroid therapy (Group S) and patients receiving supportive care (Group H). The Kaplan–Meier plots show the cumulative incidences of iMN. The X-axis represents the follow-up period (months), and the Y-axis represents the incidences of a 40% reduction in the eGFR or CR. a Comparison of a 40% reduction in the eGFR between Group S1 and Group H1 (proteinuria >1 g/gCre; difference not significant, log-rank test); b comparison of CR between Group S1 and Group H1 (proteinuria >1 g/gCre; P = 0.02, log-rank test). c Comparison of a 40% reduction in the eGFR between Group S2 and Group H2 (using propensity score matching; proteinuria >1 g/gCre; difference not significant, log-rank test). d Comparison of CR between Group S2 and Group H2 (using propensity score matching; proteinuria >1 g/gCre; P = 0.03, log-rank test); e comparison of a 40% reduction in the eGFR between Group S3 and Group H3 (proteinuria 1–3.5 g/gCre; difference not significant, log-rank test); f comparison of CR between Group S3 and Group H3 (proteinuria 1–3.5 g/gCre; P = 0.03 after 10 years, log-rank test). g Comparison of a 40% reduction in the eGFR between Group S4 and Group H4 (proteinuria >3.5 g/gCre; difference not significant, log-rank test); h comparison of CR between Group S4 and Group H4 (proteinuria > 3.5 g/gCre; P = 0.02 after 5 years, log-rank test)

Full size image
Table 9 Cox proportional hazards model for incidences of 40% reduction in the estimated glomerular filtration rate (eGFR)
Full size table

Using propensity score matching for patients with a urinary protein level of ≥1 g/gCre (Table 2), we compared 20 cases in each treatment group which were performed using significant differences of the background, serum total protein, albumin, total cholesterol, and IgG. For Group S2 and Group H2, the median ages differed between the two groups (Table 2). No other differences in outcomes were observed between the two groups (Table 6). When compared using the Kaplan–Meier method, a 40% reduction in the eGFR was observed at the end of 5 years in 13 and 18% of the patients, respectively (P = 0.74); at the end of 10 years, these rates had increased to 63 and 56%, respectively (P = 0.56). The CR rates at the end of 5 years were 55 and 25%, respectively (P = 0.05), while the rates at 10 years were 65 and 30%, respectively (P = 0.03) (Fig. 1c, d).

Next, we divided the patients with a urinary protein level of 1.0–3.5 g/gCre into two groups: Group S3 (n = 18) and Group H3 (n = 19). The serum albumin levels and total cholesterol, and immunoglobulin levels differed between the two groups (Table 7). The rate of relapse for the two groups was also different (P = 0.03) (Table 7). Using the Kaplan–Meier method, a 40% reduction in the eGFR was observed at the end of 5 years in 15 and 9% of the patients, respectively (P = 0.64); at the end of 10 years, these rates had increased to 29 and 43%, respectively (P = 0.57). The CR rates at the end of 5 years were 61 and 32%, respectively (P = 0.07), while the rates at 10 years were 72 and 37%, respectively (P = 0.03) (Fig. 1e, f).

Finally, 46 patients with massive proteinuria (≥3.5 g/gCre) were analyzed (Table 8). Since the numbers of patients in Group S4 (n = 34) and Group H4 (n = 12) were relatively small, we could not perform propensity score matching. The median serum albumin levels were significantly different (P = 0.04) (Table 4). No differences in patient outcomes or the incidences of complications were observed between the two groups (Table 8). When compared using the Kaplan–Meier method, a 40% reduction in the eGFR was observed at the end of 5 years in 19 and 29% of the patients, respectively (P = 0.59); at the end of 10 years, these rates had increased to 50 and 60%, respectively (P = 0.72). The CR rates at the end of 5 years were 67 and 25%, respectively (P = 0.04), while the rates at 10 years were 50 and 25%, respectively (P = 0.13) (Fig. 1g, h).

Two patients (6%) in Group H1, 1 patient (5%) in Group H2, 2 patents (11%) in Group H3, and no patient in Group H4 experienced spontaneous remission at 6 months.

Among the cases in Group H1 who did not receive steroid therapy at 6 months after diagnosis, 9 patients (29%) experienced spontaneous remission after a median period of 24 months (6–120 months); 4 patients (40%) in Group H4 experienced spontaneous remission after a median period of 42 months (24–120 months).

Relapses of proteinuria of ≥1 g/gCre occurred in 18 cases (34.6%) in Group S1 after a median period of 66 months (17–144 months) and in 1 case (3.2%) in Group H1 after a median period of 108 months (P = 0.001) (Table 5). Relapses occurred in 5 cases (25%) in Group S2 and in 1 case (5%) in Group H2 (P = 0.08) (Table 6).

After 6 months, five cases (42%) in Group H4 continued to exhibit nephrotic syndrome. After two patients in Group H4 began taking steroids (one at 13 months, the other at 108 months), one attained a CR at 21 months, and the other attained a CR at 132 months (Table 8).

We performed a Cox regression analysis to examine relapses in proteinuria of over 1 g/gCre (Table 10). Using a multivariate analysis, differences in steroid therapy (HR 9.40; 95% confidence interval 1.58–185.63; P = 0.01) and a proteinuria level of over 3.5 g/gCre (HR 9.11; 95% confidence interval 1.74–65.73; P = 0.008) were observed.

Table 10 Cox proportional hazards model for incidences of relapse
Full size table

Discussion

We examined the long-term outcome of iMN according to the initial therapy in patients over 16 years. We found that corticosteroid monotherapy within 6 months showed a tendency to induce early remission but did not have a significant impact on a 40% reduction in eGFR, compared with supportive therapy.

In Japan, supportive therapy is recommended and there is no standard treatment for iMN patients with a proteinuria level of less than 3.5 g/gCre [11, 12]. However, some moderate-quality evidence recommends that corticosteroid monotherapy not be used to induce remission or to delay the onset of progressive renal failure and the kidney disease.

Global Outcomes (KDIGO) guidelines recommend that patients with iMN only receive supportive care for at least the first 6 months after diagnosis [4]. In this study, we used corticosteroid therapy, but studies in other countries have shown that the effect of corticosteroids did not differ from that of a placebo [13,14,15]. If cases likely to experience spontaneous remission could be identified before therapy, we could decrease the rates of excessive treatments and complications [16]. Cattran et al. [14] reported a randomized control trial comparing 6 months of prednisone treatment with a control group. Eighty-one patients received prednisone (45 mg/m2) on alternate days for 6 months. No differences were observed between the two groups. This previous study differed from ours in that the median ages were 46 years for the PSL group (n = 81) and 45 years for the control group (n = 77), which were younger than our cases. The period of therapy was also shorter.

Our study suggested that the presence of segmental sclerosis and the presence of diabetes mellitus were risk factors for renal outcome regardless of steroid therapy. Wakai et al. reported that focal glomerulosclerosis (FGS) in iMN was associated with a poorer outcome than MN without FGS among patients with MN, and patients of MN with FGS had a higher urinary protein level [17].

Concerning relapses, steroid therapy within the first 6 months after diagnosis did not appear to be correlated with relapse, since the total follow-up duration for patients who had initially received steroids was longer than that for those who had received supportive care only. Caro et al. reported that the amount of proteinuria and the withdrawal of drugs were correlated with remission [9].

What we most want to know is how to predict the natural course of remission. Beck et al. [18] reported a correlation between phospholipase A2 antibodies (PLA2R-Ab) and iMN for the first time. PLA2R-Ab are positive in 60–80% of the general population and about 50% of the Japanese population [19, 20]. The PLA2R-Ab level is associated with clinical activity [21,22,23,24]. In Japan, the rate of PLA2R-Ab positivity is relatively low, so it would be difficult to use PLA2R-Ab as a marker of iMN. Blood sampling for anti-PLA2R was not performed in the present study.

A major limitation of this study was the choice of treatment during the first 6 months, since we did not choose the primary treatment according to a precise protocol. Second, we chose patients with a proteinuria level of ≥1 g/gCre, because the number of patients with a level of ≥3.5 g/gCre was relatively small. Third, we could not measure the levels of PLA2R-Ab and similar markers because of the difficulty associated with measuring these markers.

In conclusion, we believe that steroid therapy is not superior to supportive care within the first 6 months after diagnosis in terms of a 40% reduction in eGFR. Thus, we recommended the use of supportive therapy, in agreement with the KDIGO guidelines.