Comparison of Dosage of Glucocorticoid in Idiopathic Membranous Nephropathy: A Systematic Review and Network Meta-Analysis

Purpose: Idiopathic membranous nephropathy (IMN) with moderate risk or above was recommended to receive immunosuppressive therapy. We attempted to evaluate the optimal dose of glucocorticoid when combined with evidence-proven effective immunosuppressants by network meta-analysis. Methods: A systematic review of the literature was conducted in PubMed, Embase, Cochrane Library, and ClinicalTrials.gov from inception until January 2022. Randomized controlled trials (RCTs) in IMN limited to supportive care, glucocorticoids, cyclophosphamide, chlorambucil, calcineurin inhibitors (CNIs), and rituximab were screened. Results: Twenty-eight RCTs of 1,830 patients were included. Therapeutic regimens were divided as follows: moderate- to high-dose glucocorticoids plus CNIs (HMSCn), moderate- to high-dose glucocorticoids plus cyclophosphamide (HMSCt), moderate- to high-dose glucocorticoids plus chlorambucil (HMSCh), zero- to low-dose glucocorticoids plus CNIs (LNSCn), zero- to low-dose glucocorticoids plus cyclophosphamide (LNSCt), rituximab alone (R), glucocorticoids alone (SE), and supportive care alone (SP). Compared with SP, HMSCh (risk ratio [RR]: 1.77, 95% confidence interval [CI]: 1, 3.18), HMSCn (RR: 2.5, 95%CI: 1.25, 5.11), HMSCt (RR: 2.15, 95%CI: 1.29, 3.64), LNSCn (RR: 2.16, 95%CI: 1.25, 3.95), and R (RR: 2.07, 95%CI: 1, 4.39) had a higher probability of total remission rate, while HMSCn represented the highest probability depending on the surface under the cumulative ranking area (SUCRA) ranking values. Regarding infection, no significant difference was found between different doses of glucocorticoids plus the same immunosuppressant. HMSCn and HMSCt showed superiority in reducing 24-hour urine total protein compared with HMSCh, LNSCn, SE, and SP, while HMSCn seemed to be the most effective regimen through the ranking of SUCRA value. Conclusion: Moderate- to high-dose glucocorticoids showed superiority in proteinuria remission when combined with CNIs in IMN, with no increasing risk of infection.


Introduction And Background
Membranous nephropathy (MN) is a kidney-specific autoimmune disease, characterized by the thickening of the glomerular basement membrane due to the deposition of subepithelial immune complexes.About 80% of MN are idiopathic (idiopathic membranous nephropathy, IMN), while the rest 20% are secondary to other diseases or exposures including systematic autoimmune diseases, infection, malignancy, drugs, and so forth [1].IMN is the most common cause of primary nephrotic syndrome in adults worldwide [2].Most IMNs are mediated by antibodies to the M-type phospholipase A2 receptor (anti-PLA2R) (85%) and the thrombospondin type 1 domain containing 7A (THSD7A) (3%-5%) [3].Clinically, 80% of patients with IMN present with nephrotic syndrome and 20% with non-nephrotic range proteinuria.If untreated, about onethird of IMN patients undergo spontaneous remission, one-third progress to end-stage renal disease over 10 years, and the rest remain non-progressive [4].
According to the Kidney Disease Improving Global Outcomes (KDIGO) 2021 clinical practice guideline [5], IMN patients with moderate risk or above were recommended to receive immunosuppressive therapy.Chlorambucil, cyclophosphamide, calcineurin inhibitors (CNIs), and rituximab were proven to show effectiveness in reducing proteinuria in IMN depending on previous studies [6].Except for immunosuppressive agents, glucocorticoids were recommended as either add-on or not [5].However, the optimal dosage of glucocorticoids balancing the effectiveness and the adverse events was not recommended in the KDIGO 2021 guideline.Few studies have directly compared the dosage of glucocorticoids in IMN patients.
Therefore, we conducted a systematic review and network meta-analysis of randomized controlled trials (RCTs) to evaluate the efficacy and adverse effects of glucocorticoids with different dosages when combined with recommended immunosuppressants.

Review Materials & methods
This network meta-analysis was conducted according to the Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) statement [7] and the extension for network meta-analysis [8].The study was registered prospectively with the International Prospective Register of Systematic Reviews (PROSPERO; CRD42021291150).

Data Sources and Search Strategy
A systematic review of the literature was conducted in PubMed, Embase, Cochrane Library, and ClinicalTrials.govfrom inception until January 2022 with relevant records retrieved.RCTs for investigating the treatment of IMN in adults with nephrotic syndrome were targeted, limiting to therapeutic strategies of supportive care, glucocorticoid, cyclophosphamide, chlorambucil, CNIs, and rituximab.Details of the search strategies [9] and results are shown in supplementary table S1.

Study Selection
The inclusion criteria were displayed as follows: 1) RCTs of patients over 14 years old diagnosed as IMN by renal biopsy, excluding secondary membranous nephropathy; 2) therapeutic arms limited to supportive therapy; glucocorticoid alone or combined with immunosuppressive agents (cyclophosphamide, chlorambucil, CNIs including either cyclosporine or tacrolimus, or rituximab); immunosuppressants mentioned above alone; 3) at least one of the following outcomes were mentioned: complete remission (CR) rate, partial remission (PR) rate, change in 24-hour urine total protein (24-hour UTP), relapse rate, and adverse event (infection) rate; 4) literature published in English.Depending on the dosage of glucocorticoid, regimens of immunosuppressive drugs with glucocorticoid were divided into zero-to low-dose glucocorticoid group (prednisone 5 mg/kg.d - or equivalent) and moderate-to high-dose glucocorticoid group (prednisone ≥0.5 mg/kg.d - or equivalent).CR was defined as 24-hour UTP <0.3 g (urine protein: creatinine ratio [uPCR]<300 mg/g or <30 mg/mmol), along with normal serum creatine (SCr).PR was defined as 24-hour UTP <3.5 g (uPCR < 3500 mg/g or <350 mg/mmol) and at least 50% reduction from peak values, accompanied by a stable SCr.Relapse was defined as the development of proteinuria >3.5 g/day after CR or PR [5,10].
Studies were excluded as follows: 1) secondary membranous nephropathy; 2) patients with kidney transplantation; 3) treatments including agents with uncertain outcomes in IMN, such as traditional Chinese medicine, mycophenolate mofetil, mizoribine, leflunomide, azathioprine, and adrenocorticotropic hormone; 4) containing more than one immunosuppressive agent in a single therapeutic arm except for glucocorticoid; 5) comparison of the dosage or administration way of the same immunosuppressant except for glucocorticoid; 6) reviews, meta-analysis, case reports, case series, retrospective cohort studies, singlearm studies, animal studies, and in vitro studies (Table 1).

Randomized controlled trials
Reviews, meta-analysis, case reports, case series, retrospective cohort studies, and

Outcome Assessment
The primary outcome was the total remission rate (TRR), defined as the rate of either CR or PR.The secondary outcomes included adverse event (infection) rate, change in 24-hour UTP, and relapse rate.

Data Extraction and Quality Assessment
Two researchers (YL and ZG) extracted data independently from the included studies.Any disagreement would be resolved by a third reviewer (JZ).The following information was collected: study information (author, year of publication, country, single or multicenter, study type, sample size, intervention, control arm, and follow-up time), characteristics of the patients (age, sex, baseline of 24-hour UTP, and SCr), and primary and secondary outcomes (TRR, infection rate, change in 24-hour UTP, and relapse rate).In studies that reported median and range of 24-hour UTP, the following formula was used to transform it into mean and standard deviation (SD): mean ≈ , and SD ≈ (a for the minimum value; b for the maximum value; m for the median; n for the sample size; -1 for -1( ) , the upper th percentile of the standard normal distribution) [11,12].The mean difference (MD) was calculated as the difference of mean values between the follow-up and baseline.The formula of SDChange = was used to calculate the SD of change scores.R represented the Pearson correlation coefficient, inputted with 0.5 [13].Quality assessment of the included studies was evaluated independently by two researchers (YL and ZG) under the indications recommended in the Cochrane Handbook [14].

Statistical Analysis
Conventional pairwise meta-analysis was initially performed considering the available head-to-head comparisons with a standard random-effect model.Categorical data was estimated by risk ratio (RR) with 95% confidence interval (CI), while continuous variable was evaluated with MD with 95%CI.Between-study heterogeneity was assessed by I 2 statistic, when I 2 <50% represented no significant heterogeneity.
A random-effect network meta-analysis was performed by a Bayesian network framework with a Monte Carlo Markov chain (MCMC) model [15].Four chains were set simultaneously, and the number of tuning iterations and simulation iterations were 20,000 and 50,000, respectively.For all outcomes, evidence from the included studies was summarized in network relation graphs.Outcomes of different therapeutic regimens were ranked through the surface under the cumulative ranking area (SUCRA) curve and demonstrated in league tables.Larger SUCRA scores indicated greater probability [16].A global heterogeneity was assessed by I 2 statistic.To evaluate the consistency between direct and indirect comparisons, node node-splitting approach was performed.A p-value <0.05 was considered statistically significant [17].All analyses were conducted using R V4.1.2(R Foundation for Statistical Computing, Vienna, Austria) (pairwise meta-analysis, network meta-analysis, league tables, and assessment of heterogeneity and consistency) and STATA V15.

Study Selection and Characteristics
A total of 1,129 studies were retrieved after the initial search.After removing 325 duplicate studies, 661 studies were excluded from the screening of titles and abstracts.Then after full text was assessed, 115 were removed.Ultimately, 28 RCTs of 1,830 patients were included in the network meta-analysis (Figure 1).Depending on the dosage of glucocorticoids and the type of immunosuppressive agents, therapeutic regimens were divided as follows: a moderate to high dose of glucocorticoids plus CNIs (HMSCn), a moderate to high dose of glucocorticoids plus cyclophosphamide (HMSCt), a moderate to high dose of glucocorticoids plus chlorambucil (HMSCh), zero to a low dose of glucocorticoids plus CNIs (LNSCn), zero to a low dose of glucocorticoids plus cyclophosphamide (LNSCt), rituximab alone (R), glucocorticoids alone (SE), and supportive care alone (SP).Supportive care included non-immunosuppressive antiproteinuric treatment, anticoagulation, and so on.Both supportive care had been administrated or not were eligible in all the immunosuppressive regimens.The main characteristics of the 28 RCTs are summarized in Table 2.   a Indicates that the unit is mg/dl; bIndicates that the unit is umol/L.

Risk-of-Bias Assessment
According to the Cochrane instrument, more than half of the included studies were judged to have a low risk of bias for random sequence generation.Most of the studies were regarded as low risk of bias for incomplete outcome data and selective reporting, and all studies were at low risk of bias for blinding of outcome assessment.Three studies that used blinding methods were judged to have a low risk of bias.More details of quality assessment are demonstrated in supplementary figure S1.

Heterogeneity Analysis and Evaluation of Inconsistency
For heterogeneity analysis, the global I2 value was 46.56% (TRR), 61.27% (infection rate), 39.28% (change in 24-hour UTP), and 0.00% (relapse rate), respectively.With regard to the local inconsistency, the node-split model revealed no significant discrepancy between the direct and indirect comparisons in outcomes of relapse rate.However, a significant difference was found in one comparison of TRR, one comparison of infection rate, and two comparisons of change in 24-hour UTP (supplementary tables S2-S5).

Conventional Pairwise Meta-Analysis
By conventional pairwise meta-analysis, it was discovered that TRR in groups of HMSCh (RR: 1.37, 95%CI:  2A.Compared with SP, HMSCh, HMSCn, HMSCt, LNSCn, and R all had a higher probability of TRR (Table 3).With regard to the SUCRA ranking curves, HMSCn seemed to have the highest probability of TRR (Table 4, Figure 2B).

TABLE 3: The league table of all comparisons of total remission rate and adverse event (infection) rate
The league table of all comparisons of total remission rate (on the lower triangle) and adverse event (infection) rate (on the upper triangle).The summary estimates are risk ratios (RRs) and 95% confidence intervals.For total remission rate, the column-defining treatment is compared to the row-defining treatment, and RRs >1 favor the column-defining treatment.For adverse event (infection) rate, the column-defining treatment is compared to the rowdefining treatment, and RRs <1 favor the row-defining treatment.Significant results are in bold.

Adverse Event (Infection) Rate
Various adverse events were mentioned in the 28 RCTs, such as gastrointestinal symptoms, glucose intolerance, diabetes mellitus, anemia, hypertension, and so forth.However, we mainly focused on infections (pneumonia, urinary tract infection, skin infection, etc.).A total of 204 records of infection in 1,247 patients were mentioned in 20 RCTs, with eight intervention regimens included: HMSCh (six trials, 23 patients with infections), HMSCn (five trials, 35 patients with infections), HMSCt (11 trials, 67 patients with infections), LNSCn (seven trials, 33 patients with infections), LNSCt (one trial, 0 patient with infections), R (three trials, 25 patients with infections), SE (one trial, 0 patient with infections), SP (eight trials, 21 patients with infections).The network plot is demonstrated in Figure 3A.SE was associated with a lower probability of infections compared with HMSCh, HMSCn, and HMSCt when other comparations showed no significant differences (Table 3).Details of SUCRA ranking are displayed in Table 5 and Figure 3B.6).By ranking the SUCRA value, HMSCn seemed to be the most effective regimen in reducing 24-hour UTP (Table 7 and

Relapse Rate
A total of 10 studies involving 96 relapsed in 484 patients during the follow-up.Seven therapeutic regimens were involved: HMSCh (one trial, 11 patients with relapse), HMSCn (three trials, 10 patients with relapse), HMSCt (eight trials, 36 patients with relapse), LNSCn (two trials, 12 patients with relapse), R (one trial, three patients with relapse), SE (three trials, 15 patients with relapse), SP (two trials, nine patients with relapse).The network map is demonstrated in supplementary figure S2A.No significant difference was found among the therapeutic regimens by network meta-analysis (supplementary table S10).Details of SUCRA ranking are demonstrated in supplementary table S11 and figure S2B.

Discussion
In our network meta-analysis on treatments of IMN, it is found that regimens of moderate-to high-dose glucocorticoids plus either chlorambucil, cyclophosphamide, or CNIs increased the probability of remission compared to solely supportive care.Moreover, the regimen of zero to low-dose glucocorticoids plus CNIs and the regimen of rituximab alone also showed more effectiveness than supportive care regarding remission.Moderate-to high-dose glucocorticoids plus CNIs showed the highest probability of total remission and reduction of proteinuria.Moderate-to high-dose glucocorticoids were not demonstrated to increase the infection rate compared with zero to low-dose when added to either cyclophosphamide or CNIs.
Glucocorticoids have been widely administrated in glomerular diseases.Glucocorticoids exert antiinflammatory and immunosuppressive functions through genomic and nongenomic effects [18].The dosage of glucocorticoids varies in different research on IMN.Large-dose impulsion sequential by a medium dose of glucocorticoids plus immunosuppressants was applied in several studies [19,20].Moreover, an initial high dose of glucocorticoid plus immunosuppressants has also been mentioned [21,22].Besides, a regimen of low-dose or even no glucocorticoids with immunosuppressants especially CNIs has shown effectiveness in remission of IMN [23,24].Seldom studies have compared the dosage of glucocorticoids directly and the optimal dosage of glucocorticoids hasn't been recommended in 2021 KDIGO guideline [5].In our study, different dosages of glucocorticoids with the same immunosuppressants (cyclophosphamide or CNIs) were compared through network meta-analysis.Regimens of CNIs regardless of the dosage of glucocorticoids had a higher probability of total remission compared with supportive care alone, while CNIs plus moderate-to high-dose glucocorticoids had the highest probability.Concerning reduction of proteinuria, moderate-to high-dose glucocorticoids showed superiority to zero to low-dose when accompanied with CNIs.It is naturally understood that the higher the dosage of the glucocorticoid, the more the effectiveness of disease remission.
Side effects of glucocorticoids are usually dose-and time-dependent [18].It is reported that the side effects of the administration of glucocorticoids plus cyclosporine were significantly higher than those of cyclosporine alone [24].In our study, we focused on the adverse effects of infection.However, we did not find that moderate-to high-dose glucocorticoids showed a higher probability of infection rate compared with zero to low-dose when added to either cyclophosphamide or CNIs.The result might be due to the rapid reduction of protein leakage after administration of adequate glucocorticoids or the quick tapering of glucocorticoids after remission, which might reduce the risk of infection, in the regimens of moderate-to high-dose glucocorticoids plus cyclophosphamide or CNIs.
CNIs, such as cyclosporine and tacrolimus, exert their immunosuppressive effect by indirectly affecting Bcell function.It has been widely proven from clinical studies that CNIs-based management increased the probability of PR and CR in IMN patients [6].It was reported that the reduction of proteinuria was more significant in the CNIs group compared with the cyclophosphamide group [6,25].Our network meta-analysis demonstrated that moderate-to high-dose glucocorticoids combined with CNIs seemed to show the highest probability of total remission and be the best regimen in reducing 24-hour UTP, which was consistent with the previous studies.The superiority of CNIs in the reduction of proteinuria may be owing to the direct target on the podocyte [26].
Some limitations should be taken into account when interpreting our results.First, the outcome of renal function after follow-up has not been investigated due to the insufficiency of evidence for some regimens.
Secondly, the serum level of anti-PLA2R antibody has not been estimated because only a few studies published recently had mentioned this specific biomarker of IMN.Thirdly, the sample sizes of some studies were only pretty limited, which reduced the level of evidence in our study.Hence, more new studies need to be included in the future.

FIGURE 1 :
FIGURE 1: Flowchart of the study selection procedure.RCT, randomized clinical trial.

FIGURE 2 :
FIGURE 2: Network map and SUCRA plot for total remission rate in IMN.(A) Network map.Nodes indicate the treatment regimens that are evaluated in the study.The size of the node is proportional to the number of trials evaluating the treatment.Lines represent head-to-head comparisons of the two treatment regimens indicated by the connected nodes.The thickness of the lines is weighted according to the number of trials comparing the two connected treatment regimens.(B) SUCRA plot.SUCRA, surface under the cumulative ranking area; IMN, idiopathic membranous nephropathy; HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderate-to high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to lowdose glucocorticoid plus calcineurin inhibitors; R, rituximab alone; SE, glucocorticoid alone; SP, supportive care alone.

FIGURE 3 :
FIGURE 3: Network map and SUCRA plot for infection rate in IMN.(A) Network map.Nodes indicate the treatment regimens that are evaluated in the study.The size of the node is proportional to the number of trials evaluating the treatment.Lines represent head-to-head comparisons of the two treatment regimens indicated by the connected nodes.The thickness of the lines is weighted according to the number of trials comparing the two connected treatment regimens.(B) SUCRA plot.SUCRA, surface under the cumulative ranking area; IMN, idiopathic membranous nephropathy; HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderate-to high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to lowdose glucocorticoid plus calcineurin inhibitors; R, rituximab alone; SE, glucocorticoid alone; SP, supportive care alone.

FIGURE 4 :
FIGURE 4: Network map and SUCRA plot for change in 24-hour UTP in IMN.(A) Network map.Nodes indicate the treatment regimens that are evaluated in the study.The size of the node is proportional to the number of trials evaluating the treatment.Lines represent head-to-head comparisons of the two treatment regimens indicated by the connected nodes.The thickness of the lines is weighted according to the number of trials comparing the two connected treatment regimens.(B) SUCRA plot.SUCRA, surface under the cumulative ranking area; 24-hour UTP, 24-hour urine total protein; IMN, idiopathic membranous nephropathy; HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderateto high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to low-dose glucocorticoid plus calcineurin inhibitors; R, rituximab alone; SE, glucocorticoid alone; SP, supportive care alone.

FIGURE 6 :
FIGURE 6: Supplementary figure 2. Network map and SURCA plot for relapse rate in IMN.(A) Network map.Nodes indicate the treatment regimens that are evaluated in the study.The size of the node is proportional to the number of trials evaluating the treatment.Lines represent head-to-head comparisons of the two treatment regimens indicated by the connected nodes.The thickness of lines is weighted according to the number of trials comparing the two connected treatment regimens.(B) SURCA plot.

TABLE 1 : Inclusion and exclusion criteria for the systematic review and network meta-analysis
IMN, idiopathic membranous nephropathy.

TABLE 6 : The league table of all comparisons of change in 24-hour UTP The
summary estimates are mean differences (MDs) and 95% confidence intervals.The column-defining treatment is compared to the row-defining treatment, and MDs <0 favor the column-defining treatment.Significant results are in bold.24-hourUTP, 24-hour urine total protein; HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderate-to high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to low-dose glucocorticoid plus calcineurin inhibitors; LNSCt, zero to low-dose glucocorticoid plus cyclophosphamide; SE, glucocorticoid alone; SP, supportive care alone.

TABLE 9 : Supplementary table S2. Evaluation of local inconsistency for total remission rate with node-split model.
The ratio of risk ratios (RRs) for direct and indirect estimates with their 95% confidence intervals (CIs).A CI including 1 indicates there is no statistically detectable inconsistency between direct and indirect treatment estimates.Bold values denote statistical significance.HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderate-to high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to low-dose glucocorticoid plus calcineurin inhibitors; LNSCt, zero to low-dose glucocorticoid plus cyclophosphamide; R, rituximab alone; SE, glucocorticoid alone; SP, supportive care alone.

TABLE 10 : Supplementary table S3. Evaluation of local inconsistency for adverse event (infection) rate with node-split model.
The ratio of risk ratios (RRs) for direct and indirect estimates with their 95% confidence intervals (CIs).A CI including 1 indicates there is no statistically detectable inconsistency between direct and indirect treatment estimates.Bold values denote statistical significance.

TABLE 11 : Supplementary table S4. Evaluation of local inconsistency for change in 24-hour UTP with node-split model.
The mean difference (MDs) for direct and indirect estimates with their 95% confidence intervals (CIs).A CI including 0 indicates there is no statistically detectable inconsistency between direct and indirect treatment estimates.Bold values denote statistical significance.

TABLE 12 : Supplementary table S5. Evaluation of local inconsistency for relapse rate with node- split model.
The ratio of risk ratios (RRs) for direct and indirect estimates with their 95% confidence intervals (CIs).A CI including 1 indicates there is no statistically detectable inconsistency between direct and indirect treatment estimates.HMSCn, moderate-to high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to low-dose glucocorticoid plus calcineurin inhibitors; SE, glucocorticoid alone; SP, supportive care alone.

TABLE 13 : Supplementary table S6. Outcomes of conventional pairwise meta-analysis for total remission rate.
HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderate-to high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to low-dose glucocorticoid plus calcineurin inhibitors; LNSCt, zero to low-dose glucocorticoid plus cyclophosphamide; R, rituximab alone; SE, glucocorticoid alone; SP, supportive care alone; NA, no available.Tau 2 represents between-study heterogeneity characterized by standard deviation.

TABLE 17 : Supplementary table S10. The league table of all comparisons of relapse rate.
The league table of all comparisons of relapse rate.The summary estimates are risk ratios (RRs) and 95% confidence intervals.The column-defining treatment is compared to the row-defining treatment, and RRs >1 favor the column-defining treatment.HMSCh, moderate-to high-dose glucocorticoid plus chlorambucil; HMSCn, moderate/high-dose glucocorticoid plus calcineurin inhibitors; HMSCt, moderate-to high-dose glucocorticoid plus cyclophosphamide; LNSCn, zero to low-dose glucocorticoid plus calcineurin inhibitors; R, rituximab alone; SE, glucocorticoid alone; SP, supportive care alone.