Introduction

Up to 60% of patients with systemic lupus erythematosus (SLE) have renal impairment, and lupus nephritis remains the major cause of morbidity and mortality in SLE [29, 35]. The treatment strategy for lupus nephritis consists of an induction phase to induce remission and a maintenance phase to prevent recurrence and the development of end-stage renal disease [1]. Cyclophosphamide (CYC) therapy has long been considered the gold standard for inducing renal remission and preventing flare-ups. However, significant drug-related adverse effects, such as an increased risk of severe infections, bone marrow suppression, malignancy, and ovarian toxicity, outweigh these benefits [30]. Other immunosuppressive drugs used to treat lupus nephritis include mycophenolate mofetil (MMF), tacrolimus, and azathioprine (AZA).

Several studies have examined the efficacy and safety of calcineurin inhibitor (CNI) MMF, and AZA as maintenance therapies in patients with lupus nephritis [5,6,7,8, 12, 17, 19, 20, 28, 36]. All these drugs have demonstrated significant efficacy as maintenance therapies in randomized controlled trials (RCTs) for lupus nephritis. One meta-analysis found that MMF was as effective as AZA when administered for maintenance therapy and less hazardous for the treatment of lupus nephritis [13]. A second study, however, found no benefit of using MMF over AZA as maintenance therapy for this indication [26]. CNI, MMF, and AZA were investigated in a small number of RCTs to examine their relative efficacy and safety as maintenance treatments for lupus nephritis, but the findings were equivocal owing to the small sample sizes. Even if the available data from head-to-head comparisons are insufficient, network meta-analysis may assess the comparative effectiveness of various medicines and integrate data from a network of RCTs to help in decision-making [23, 24]. The present study employed a network meta-analysis to compare the efficacy and safety of CNI, MMF, and AZA as maintenance treatments for lupus nephritis.

Methods

Identification of eligible studies and data extraction

We performed an exhaustive search for studies that examined the efficacy and safety of CNI, MMF, and AZA as maintenance treatments for patients with lupus nephritis. A literature search was performed in PubMed, EMBASE, and the Cochrane Controlled Trials Register to identify available articles (up to December 2022). The following keywords and subject terms were used in the search: “lupus nephritis,” “maintenance treatment,” “tacrolimus,” “cyclosporine,” “mycophenolate mofetil,” and “azathioprine.” All references were reviewed to identify additional studies that were not included in the electronic databases. RCTs were included if the study met the following criteria: 1) compared CNI or MMF with AZA, or CNI with MMF as maintenance therapy for lupus nephritis; 2) provided endpoints for the efficacy and safety of CNI, MMF, and AZA; and 3) included patients with biopsy-proven lupus nephritis class III, IV, or V. The exclusion criteria were as follows: 1) inclusion of duplicate data and 2) lack of adequate data for inclusion. The efficacy outcome was relapse. The definitions of relapse were based on the criteria used in each trial (Table 1 and 2). The safety outcome was the number of patients withdrawn due to adverse events (AEs), infection, leukopenia, or doubling of serum creatinine. The following information was extracted from each study: first author, year of publication, study region, and kidney biopsy class; number of patients treated with CNI, MMF, or AZA; and the efficacy and safety outcomes of the drugs. We quantified the methodological qualities of the four studies using the Jadad scores [18]. Network meta-analysis was conducted in accordance with the guidelines provided by the PRISMA statement [27].

Table 1 Characteristics of individual studies included in the meta-analysis
Full size table
Table 2 Characteristics of individual studies included in the meta-analysis
Full size table

Evaluation of statistical associations for network meta-analysis

The efficacy and safety of CNI, MMF, and AZA for maintenance therapy of lupus nephritis were ordered according to the probability of being ranked as the best-performing agent. A random effects model was used as a conservative method, and a Bayesian network meta-analysis was conducted using NetMetaXL [2] and WinBUGS statistical analysis program version 1.4.3 (MRC Biostatistics Unit, Institute of Public Health, Cambridge, UK). The Markov chain Monte Carlo method was used to obtain a pooled effect size [3]. All chains were run with 10,000 burn-in iterations followed by 10,000 monitoring iterations. The information on relative effects was converted to the probability that a treatment was best, second best, and so on, with the ranking of each treatment (called the surface under the cumulative ranking curve or SUCRA) expressed as a percentage, ranging between 100% and 0%, with 100% and 0% indicating that the treatment is the best and worst, respectively [31]. The summary estimates were presented in league tables by ranking the treatments in order of the most pronounced impact on the outcome under consideration based on the SUCRA [31]. We reported the pairwise odds ratio (OR) and 95% credible interval (CrI) adjusted for the multiple-arm trials. Pooled results were considered statistically significant if 95% of the CrI did not contain a value of 1.

Test for inconsistency and sensitivity analysis

Inconsistency refers to the extent of disagreement between the direct and indirect evidence [10]. Assessment of inconsistency is important for conducting a network meta-analysis [16]. We plotted the posterior mean deviance of the individual datapoints in the inconsistency model against their posterior mean deviance in the consistency model to assess network inconsistency between the direct and indirect estimates in each loop [34]. A sensitivity test was performed by comparing the random and fixed effects models.

Results

Studies included in the meta-analysis

A total of 967 studies were identified using electronic or manual searches, of which 16 were selected for full-text review based on the title and abstract details. However, six of the 16 studies were excluded: one because it included patients with lupus nephritis without renal biopsy, [9] one was not an RCT (cohort [4]), four studies included data only on AZA without a comparison group including tacrolimus, MMF, or AZA [11, 28], and two studies included duplicate data [32, 33]. Thus, 10 RCTs, including a total of 884 patients, met the inclusion criteria ([5,6,7,8, 12, 17, 19, 20, 28, 36]; Fig. 1). There were three interventions, including eight studies of MMF, nine of AZA, and three of CNI, for the network meta-analysis. The Jadad scores for all studies except two were 3–4, indicating high study quality. The relevant features of the studies included in this meta-analysis are presented in Table 1 and 2.

Fig. 1
figure 1

Evidence network diagram of comparisons for network meta-analysis. The width of each edge is proportional to the number of randomized controlled trials comparing each pair of treatments, and the size of each treatment node is proportional to the number of randomized participants (sample size). AZA azathioprine, CNI calcineurin inhibitor, MMF mycophenolate mofetil

Full size image

Network meta-analysis of the efficacy of CNI, MMF, and AZA in RCTs

We considered the number of relapses as the efficacy outcome. MMF was listed at the top left of the diagonal of the league table (Fig. 2) because it was associated with the most favorable SUCRA for efficacy outcomes, whereas AZA was listed at the bottom right of the diagonal of the league table because it was associated with the least favorable results (Fig. 2; Table 3). MMF showed a trend toward a lower relapse rate compared with AZA (odds ratio [OR] 0.72, 95% credible interval [CrI] 0.45–1.22; Fig. 2 and 3). The number of relapses were lower in the CNI group than in the AZA group (OR 0.85, 95% CrI 0.34–2.00; Fig. 2 and 3). However, the number of relapses did not differ significantly between the three drugs. This may be partly explained by the low statistical power from a relatively small number of patients who had relapses in each group and from the small number of studies in this network meta-analysis. The ranking probability based on SUCRA indicated that MMF had the highest probability of being the best treatment based on the renal relapse rate, followed by CNI and AZA (Table 3).

Fig. 2
figure 2

League tables showing the results of the network meta-analyses comparing the effects of all drugs, including odds ratio (OR) and 95% credible interval (CrI). OR < 1 means the treatment in the top left is better. a Relapse; b withdrawal due to adverse events; c infection; d leukopenia; e doubling serum creatine. MMF mycophenolate mofetil, AZA azathioprine

Full size image
Table 3 Rank probability of efficacy and safety of CNI, MMF, and AZA based on renal relapse, withdrawal due to AEs, number of infections, and leukopenia, and doubling serum creatine
Full size table
Fig. 3
figure 3

Bayesian network meta-analysis results of randomized controlled studies on the relative efficacy and safety of calcineurin inhibitor (CNI), mycophenolate mofetil (MMF), and azathioprine (AZA) based on renal relapse, withdrawal due to adverse events (AEs), number of infections, leucopenia, and doubling serum creatine. a Relapse, b withdrawal due to AE, c infection, d leukopenia, e doubling serum. OR odds ratio, CrI 95% credible interval

Full size image

Network meta-analysis of the safety of CNI, MMF, and AZA in RCTs

We considered the number of patient withdrawals due to AEs, infection, leukopenia, and doubling of serum creatinine as the safety outcome. The number of patient withdrawals due to AEs did not differ significantly between the interventions (Fig. 2 and 3). However, the incidence of infection in the CNI and MMF groups tended to be lower than that in the AZA group (OR 0.39, 95% CrI 0.09–1.92; OR 0.56, 95% CrI 0.19–1.63; Fig. 2 and 3). The ranking probability based on SUCRA indicated that CNI had the highest probability of being the most tolerable treatment, followed by MMF and AZA (Table 3). Fewer patients with infections were observed in the CNI and MMF groups than in the AZA group, although the difference was not statistically significant (Fig. 2 and 3). The incidence of leukopenia in the MMF and CNI groups was significantly lower than that in the AZA group (OR 0.12, 95% CrI 0.04–0.34; OR 0.16, 95% CrI 0.04–0.50; Fig. 2 and 3). The ranking probability based on SUCRA indicated that MMF had the highest probability of being the safest treatment based on leukopenia incidence, followed by CNI and AZA (Table 3). The doubling of serum creatinine levels did not differ significantly among the interventions (Fig. 2; Table 3).

Inconsistency and sensitivity analysis

Inconsistency plots assessing network inconsistencies between direct and indirect estimates showed a low possibility that these inconsistencies may significantly affect the network meta-analysis results. In addition, the random and fixed effects model results provided the same trend for OR, indicating that the results of this network meta-analysis are robust (Fig. 3).

Discussion

The best long-term results for lupus nephritis require identification of the optimal maintenance medication after remission induction [1]. The relative effectiveness and safety of CNI, MMF, and AZA as maintenance therapies in patients with lupus nephritis were examined in this network meta-analysis. MMF, which is probably better than AZA, may be linked to the highest likelihood of preventing renal recurrence in these patients as maintenance therapy, followed by CNI. However, there were no significant differences in the efficacy. In terms of safety, there was no difference in the number of withdrawals owing to AEs among the drugs. Tolerability seemed to be higher with CNI and MMF because AE withdrawals were less common than with AZA. MMF and CNI were similarly linked to a significantly lower prevalence of leukopenia than AZA.

Unlike the previous network meta-analysis by Lee et al., [21] the current analysis eliminated CYC and included four new trials. Only one study has been conducted on CYC. However, in terms of relative effectiveness, our findings are comparable to those of the CYC study [21]. Furthermore, our findings are consistent with recent meta-analyses that found that MMF was less harmful than AZA [13]. Similar studies have been published on the management of lupus nephritis [14, 15]. Both studies were meta-analyses of RCTs on the induction and maintenance of lupus nephritis. MMF prevented relapse more successfully than AZA during maintenance treatment, with no increase in clinically meaningful AEs. In contrast to the two previous studies, our network meta-analysis focused on lupus nephritis maintenance treatment and included analyses of the relative effectiveness and safety of immunosuppressive medications such as MMF, AZA, and CNI. Our findings are consistent with those of previous studies comparing MMF to AZA in terms of renal relapse and safety. However, our findings provide important information that was missing from earlier conventional meta-analyses about the relative effectiveness and safety of CNI, MMF, and AZA as maintenance therapies for lupus nephritis.

However, due to several limitations of our research, our findings should be considered with caution. First, there were only two RCTs on tacrolimus in the included literature, resulting in limited sample sizes. Second, diversity in the study design and patient characteristics may have influenced the results of this network meta-analysis. Third, this study focused only on effectiveness and safety, measuring the number of patients who experienced renal relapse, severe infection, leukopenia, doubling blood creatine, and withdrawal due to AEs. Consequently, the findings of this study did not address all elements of drug efficacy and safety. Fourth, the results are complicated by a number of factors, including the percentage of patients with class IV lupus nephritis, varying definitions of renal relapse, induction therapy employed, severity of the illness, drug dosage, and duration of follow-up. We were not able to adjust the variables because of a lack of data.

Nevertheless, this meta-analysis has a few advantages. In various trials, patients with lupus nephritis ranged in number from 42 to 227, while this particular investigation comprised 884 participants. By improving the statistical power and resolution, we were able to generate data that were more precise than those revealed in individual research. When direct head-to-head comparisons are either unavailable or inadequate, network meta-analysis allows for an indirect comparison of different treatments, maximizing the use of all available data [22, 25]. The relative effectiveness and safety of MMF, CNI, and AZA in maintenance therapy of lupus nephritis are shown in this study. In conclusion, we discovered that CNI and MMF were the most effective maintenance therapies for patients with lupus nephritis and that MMF had the best chance of minimizing the risk of leukopenia, infections, and withdrawal due to AEs. MMF and CNI maintenance treatments are preferable over AZA in these patients owing to the outcome of decreased renal relapse rates. Further research is needed to definitively assess the relative effectiveness and safety of CNI, MMF, and AZA in a broader group of patients in order to find the best maintenance treatment for lupus nephritis.