Lupus nephritis (LN) is one of the most common clinical manifestations and serious complications of systemic lupus erythematosus (SLE) and is a leading cause of morbidity and mortality in SLE (1–3). About 40% of patients with lupus develop LN (2). The management of LN comprises two phases. The induction phase aims to induce remission. The maintenance phase aims to prevent relapse and progression to end-stage renal disease. Glucocorticoids are the first treatment and improve the renal outcomes of LN. Then, they are followed by cyclophosphamide (CYC), mycophenolate mofetil (MMF), and tacrolimus (TAC) (4, 5). CYC regimens have long been considered the gold standard for inducing renal remission and preventing relapse (4). Increasing evidence shows that MMF and TAC are at least equivalent to CYC for induction and maintenance treatment of severe LN (6–8). Nevertheless, the adverse effects of these drugs can be considerable and limiting. For example, the benefits of CYC are outweighed by treatment-related adverse effects, including gonadal toxicity (9). Although the renal remission rate with CYC or MMF regimens is up to 50%–80% in LN patients, many of these responses are partial (10). Therefore, it is urgent to search for better treatments that have better efficacy and fewer side effects (including sparing fertility) to manage LN.

Rituximab (RTX) is a chimeric monoclonal antibody that targets CD20. RTX offers an alternative or adjunctive option for SLE patients (11–13). Recently, RTX has been introduced as an induction drug for LN, and studies suggested that RTX seemed to be at least as effective as MMF and CYC regimens in inducing remission (14, 15). Still, few studies were conducted to compare the efficacy and safety of RTX with common therapeutic drugs, especially TAC. Therefore, it is important for medical decisions to assess the relative value between intervention and comparators. Network meta-analysis is an extension of traditional pairwise meta-analysis and simultaneously combines direct and indirect information about the relative efficacy of each treatment (16). Thus, using a network meta-analysis, the present systematic study aimed to compare the relative efficacy and safety (infection events) of RTX, TAC, and MMF as LN induction therapy.

It is important for induction of remission to achieve the best long-term outcomes among patients with LN. The present network meta-analysis combined and compared available evidence of the effectiveness and safety (infection events) of RTX, TAC, MMF, and CYC based on the number of patients who achieved CR, PR, and overall response and who suffered from infection events. Network meta-analysis can simultaneously combine direct and indirect information of the relative efficacy and safety of each treatment to determine the optimal treatment regimen even if there are no or insufficient head-to-head comparisons (37, 38).

In the present study, RTX was the most effective drug for inducing CR among LN patients, followed by TAC and MMF. CYC was the most successful medicine for inducing PR among patients with LN, followed by TAC and RTX. In terms of overall response, TAC was the most effective treatment drug for achieving overall response among LN patients, followed by RTX and MMF. As for safety, TAC was the safest treatment with the lowest likelihood of infection events. However, RTX showed the highest probability of experiencing infection without considering other adverse events.

In summary, RTX and TAC were the most effective drugs for inducing remission among LN patients, and TAC had the lowest probability of infection compared with the other drugs. These results were consistent with a previous network meta-analysis that compared the efficacy and safety of TAC, MMF, and CYC regimens (39). Lee and Song found that TAC was the most effective induction treatment for patients with LN and had the highest probability of decreasing the risk of serious infections (39). In addition, our results were in accordance with another meta-analysis in which the direct comparison reported that TAC was more effective and safer than CYC (40). A review showed that LN patients would benefit from RTX (41). A prior study found that RTX presented a satisfying efficacy and safety for SLE patients (42). Besides, RTX therapy has been used in vasculitis and rheumatoid arthritis. Although RTX is widely used in patients with LN, RTX as an induction treatment is less established than MMF and CYC. A possible reason is the lack of large-scale placebo-controlled clinical trials (41). In addition, the evidence for RTX use remains varied in terms of ethnicity. For example, an RCT comparing the efficacy of RTX with placebo among patients with extrarenal diseases found that there was no significant difference in disease response rate between the two groups (43). However, the RTX treatment group in a subgroup analysis showed a significantly higher renal response rate among Hispanic and African-American patients. In the present study, RTX showed the highest risk of experiencing infection adverse events. A study found that infection, infusion reaction, and neutropenia were the most common adverse events of RTX (44). Furthermore, with respect to infection, evidence reported that a dose-dependent relationship was found between the RTX dose and the frequency of infection (41). Besides, RTX, as induction therapy for LN, is mainly used in the Americas (USA, Mexico, etc.), Europe (the United Kingdom, Italy, Sweden, Spain, Greece, Netherlands, etc.), Asia (Japan, China, Singapore, etc.), and Australia. TAC is only used in Asian countries, and the use of TAC is not recommended by the European League Against Rheumatism (EULAR). The off-label use of RTX in SLE was first reported in 2002 and has been increasingly used in patients with SLE (45). Therefore, off-label use should be carefully considered.

Some limitations might be considered in this network meta-analysis. First, in addition to RCTs, observational studies were included, a problem particularly found in RTX since there were no RCTs of RTX. Besides, some RCTs included in our study had small sizes. Hence, more RCTs with a large sample size should be conducted to compare RTX, TAC, MMF, and CYC, especially head-to-head trials. Second, the data and the methods of reporting data in the primary trials limited the results in our study due to different definitions in each study. For instance, the CR and PR results were heterogeneous owing to variable definitions in existing studies. Still, our results found a low heterogeneity in the pooled analysis, which suggested that our network meta-analysis was appropriate. Furthermore, heterogeneity in patients’ characteristics of eligible studies in our meta-analysis affected the results of our study. For example, Mohan et al. reported that the response to treatment drugs for LN patients varied with ethnicity (46). In addition, Merrill et al. found that the RTX treatment group showed a significantly higher renal response rate between Hispanic and African-American patients (43). Besides, prior evidence found that diet and exercise influenced renal function (47, 48). However, these factors were not adjusted in the current meta-analysis owing to insufficient data in our study, which might have a potential effect on our results. Third, for some drugs, the data were limited in some direct comparisons. For example, in our study, direct comparison between RTX and TAC was insufficient, so indirect comparison among the two groups was limited. Fourth, with respect to adverse events, only infection events were considered in the present study. However, other adverse events, including gastrointestinal reaction, myelosuppression, liver damage, alopecia, leukopenia, and menstrual disorders, were not considered owing to insufficient data. Therefore, our results should be interpreted with caution.

Although these limitations have a potential impact on our results, our study had several strengths. First, the present study used a comprehensive literature search without date and language restriction. Second, though the sample sizes from individual trials ranged from 13 to 346, the pooled analysis in our study involved 1,566 LN patients. Finally, our study used a standardized analysis method to evaluate the confidence that might be held in our results. Network meta-analysis simultaneously combined direct and indirect information of the relative efficacy of each treatment even if there were a lack of direct head-to-head comparisons.

In conclusion, our network meta-analysis combined and compared available evidence of the effectiveness and safety of RTX, TAC, MMF, and CYC based on the number of patients who achieved CR, PR, and overall response and those who suffered from infection events. RTX and TAC were the most effective drugs for induction remission among LN patients. TAC had the lowest probability of infection compared with other drugs. RTX showed the highest probability of experiencing an infection. This meta-analysis could not conclude about other adverse events.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

KL and JG carried out the studies, participated in collecting data, and drafted the manuscript. KL and YG performed the statistical analysis and participated in its design. KL, FZ, and ZL participated in the acquisition, analysis, or interpretation of data and drafted the manuscript. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.