Systematic review and meta-analysis of the low-dose versus high-dose cyclophosphamide regimen for induction treatment in lupus nephritis patients, evidence update 2010–2023

Abstract

Introduction

Lupus nephritis is a common complication in patients with systemic lupus erythematosus. Cyclophosphamide in induction treatment has a good efficacy profile, but greater toxicity.

Objective

To perform a systematic review and meta-analysis, updating the literature that compares the efficacy and safety of cyclophosphamide at high and low doses.

Materials and methods

PRISMA-P methodology, PROSPERO register number CRD42023485477, only randomized clinical trials were included from 2010 to October 2023, in patients over 16 years of age with lupus nephritis class III, IV, V, or V + III, V + IV. Complete response, partial response, relapses, infections, leukopenia, and amenorrhoea were measured, the RR was estimated for a 95% CI, based on the Mantel-Haenszel method, and heterogeneity by I² and Q test.

Results

415 articles were found, only three were included with 406 patients, 217 for the high dose group and 189 low doses. There were no statistically significant differences with respect to complete response (RR = 0.77, 95% CI 0.60–1.01, p = 0.06), partial response (RR = 0.88, 95% CI 0.67–1.16, p = 0.35), relapses (RR = 2.16, 95% CI 0.18–24.7, p = 0.52) and infections (RR = 0.68, 95% CI 0.45–1.04, p = 0.08), while leukopenia and amenorrhoea were significantly uncommon in the low dose group (RR = 0.41, 95% CI 0.22–0.77, p = 0.01) and (RR = 0.40, 95% CI 0.25–0.64, p = 0.002).

Conclusions

The use of low-dose cyclophosphamide is as effective as the high-dose regimen and the probability of leukopenia and amenorrhoea is lower.

Resumen

Introducción

La nefritis lúpica es una complicación frecuente en los pacientes con lupus eritematoso sistémico. La ciclofosfamida en el tratamiento de inducción tiene un buen perfil de eficacia, pero mayor toxicidad.

Objetivo

Llevar a cabo una revisión sistemática y un metaanálisis, actualizando la literatura que compare la eficacia y la seguridad de la ciclofosfamida a altas y bajas dosis.

Materiales y métodos

Metodología PRISMA-P, registro en PROSPERO: número CRD42023485477. Se incluyeron solamente ensayos clínicos aleatorizados desde el año 2010 hasta octubre de 2023, en pacientes mayores de 16 años con nefritis lúpica clases III, IV, V o V + III, V + IV. Se midió la respuesta completa, la respuesta parcial, las recaídas, las infecciones, la leucopenia y la amenorrea; se estimó el RR para un IC 95% a partir del método de Mantel-Haenszel y la heterogeneidad por I² y prueba Q.

Resultados

Se encontraron 415 artículos, incluyéndose únicamente tres con 406 pacientes: 217 para el grupo de altas dosis y 189 para bajas dosis. No hubo diferencias estadísticamente significativas con respecto a la respuesta completa (RR = 0,77 IC 95% 0,60−1,01, p = 0,06), parcial (RR = 0,88 IC 95% 0,67−1,16 p = 0,35), las recaídas (RR = 2,16 IC 95% 0,18–24,7 p = 0,52) y las infecciones (RR = 0,68 IC 95% 0,45−1,04 p = 0,08), mientras que la leucopenia y la amenorrea fueron significativamente más infrecuentes en el grupo de bajas dosis (RR = 0,41 IC 95% 0,22−0,77 p = 0,01) y (RR = 0,40 IC 95% 0,25−0,64 p = 0,002).

Conclusiones

El uso de ciclofosfamida a bajas dosis es tan efectivo como el esquema de altas dosis, con menor probabilidad de leucopenia y amenorrea.

Introduction

Lupus nephritis is a type of glomerulonephritis that affects about 50% of patients with systemic lupus erythematosus (SLE), and is defined as evidence of proteinuria (>2+ of proteins on test strip on any level of specific gravity, >1+ of mild gravity, proteinuria/creatininuria ratio [PCR] >500 mg/g) or abnormal sediment (acanthocytes >5% and red and white blood cell casts in urinalysis).¹ The histopathological study is necessary for its characterization, according to the classification of the International Society of Nephrology/Renal Pathology Society, which categorizes lupus nephritis into six classes: minimal mesangial, proliferative mesangial, focal proliferative, diffuse proliferative, membranoproliferative and sclerosing; in 2018, some aspects related to the inclusion of fibrinoid necrosis as an activity parameter were modified and global and segmental subclassification of diffuse proliferative nephritis was eliminated from the subclassification.²

There is a high toxicity related to cyclophosphamide (CYC) in induction treatment derived from the cumulative dose and its association with bladder toxicity in the context of hemorrhagic cystitis and ovarian failure. The alternative is mycophenolate; however, divergent results are found, and it has been described that it is not superior to cyclophosphamide in terms of efficacy measured by stabilization or improvement of serum creatinine and decrease in the proteinuria/creatinuria ratio.³ but in other studies it has shown to be more effective than cyclophosphamide in achieving remission at 6, 12 and 24 months.⁴ In others, it has not shown significant differences in clinical responses at 6 months,^5,6 which could be the result of heterogeneous studies, therefore, a recommendation with a high level of evidence cannot be made.

Due to the foregoing, cyclophosphamide is recommended as the initial treatment; however, the risk-benefit assessment based on the dose has generated such uncertainty that it has led to the performance of a systematic review and meta-analysis regarding to the treatment of lupus nephritis in 2016. These showed, indeed, that cyclophosphamide had an adequate efficacy profile with respect to progression to end-stage renal disease compared to standard doses of steroids. (CYC; OR = 0.49; 95% CI: 0.25−0.92) or CYC + azathioprine (AZA; OR = 0.18; 95% CI: 0.05−0.57), steroids (OR = 3.59; 95% CI: 1.30–9.86).⁷

Another systematic review explored the efficacy and toxicity of different induction treatment regimens and found that low doses of cyclophosphamide had a lower rate of relapse compared to high doses of the same treatment (RR = 0.465; 95% CI: 0.261−0.830; p = 0.010), which was associated with lower risk of infection (RR = 0.688; 95% CI: 0.523−0.905; p = 0.008).⁸

Tian et al.⁹ conducted a systematic review and meta-analysis to evaluate the differences in efficacy and safety of low-dose cyclophosphamide compared to high-dose cyclophosphamide in the management of lupus nephritis, in 7 clinical trials which included 665 patients, and found no differences in efficacy, but a favorable effect of the low-dose regimen on the risk of infection (RR = 0.74; 95% CI: 0.56−0.98; p = 0.03).

Previous systematic reviews included studies until 2017, with no age restriction. Subsequently, there was a publication in 2018 in which Mehra et al.¹⁰ reported that the response favors the scheme of high doses over the management with low doses of cyclophosphamide, which invites to conduct a new literature review to update the evidence, including studies from 2010 to 2023.

Consequently, given the low number of clinical trials specifically exploring the effect of this dose difference, we decided to conduct this review, with the aim of updating the available evidence regarding the efficacy and safety of the low-dose cyclophosphamide regimen versus the high-dose regimen in the induction treatment of patients with lupus nephritis. This study only includes part of the available evidence, as it does not include studies that are outside of the established dates. Therefore, it is not possible to make an estimation of the comparator effect with relevant studies previously published and, consequently, the data analyzed do not necessarily reflect the full evidence in the body of literature.

Materials and methods

Protocol and registration

This systematic review was conducted using the PRISMA-P (Preferred Reporting Items for Systematic Review and Meta Analysis Protocol) 2015 Checklist¹¹ methodology and was registered on the PROSPERO platform with the number CRD42023485477.

Search strategy

The search for the articles was carried out by a researcher (PQ) in the records of the MEDLINE and Cochrane (Central Register of Controlled Trials), SCIELO databases, from 2010 to October 2023. Only randomized clinical trials published in English and Spanish were selected; gray literature was included. For the search strategy, the following formula was used for MeSH terms as keywords: ((((lupus nephritis) AND (cyclophosphamide)) AND (treatment outcome). Subsequently, a list was developed with the authors, the year of publication and the titles/abstracts, which was independently reviewed by two authors (PQ, PP) for approval according to the eligibility criteria. In case of disagreement, it was evaluated by a third author (AA).

Eligibility criteria

Randomized clinical trials involving patients over 16 years of age with a diagnosis of lupus nephritis classes III, IV, V or V + III, V + IV, proven by renal biopsy, in which the two cyclophosphamide induction treatment regimens were compared: low-dose regimen and high-dose regimen according to the definition assigned to each study, were included. Studies with a follow-up period of at least 6 months after induction therapy were also included to evaluate relapses and adverse events related to the drug. We excluded those studies that included patients with classes I, II and VI lupus nephritis, since they do not require the immunosuppressive treatment under study. Given the high variability between the definition of high- and low-dose regimens as a consequence of a non-standardized definition in the studies, it was considered to adopt the definition assigned to each one individually. This difficulty is stated in the limitations of the study. A low-dose target value established in the Euro-lupus scheme of 3 g of cumulative dose in 3 months is taken as a reference.

Outcomes

Primary outcome

Efficacy measured by the proportion of patients who achieve complete response (CR), defined according to the criteria in force at the time of the clinical trial. The relative risk (RR) and 95% CI were estimated, respectively.

Secondary outcomes

Proportion of patients who achieve partial response (PR), defined as at least 50% reduction in proteinuria at 6 months of follow-up; proportion of drug-related adverse events: infections, leukopenia, relapses during the follow-up period, and amenorrhea.

Quality assessment

Two researchers (AA, JM) conducted the bias assessment using the Cochrane risk-of-bias tool, taking into account the five domains: biases in the randomization process, randomization sequence, incomplete data, outcome measurement, and selection of reported outcomes; were classified as "high risk", "some concern" and "low risk", independently by each of the reviewers, through the RoB2 Assessment Form.^12,13

Data extraction and synthesis

Information on the year, the author, the mean age, the treatment groups (low- and high-dose schedules), the proportion of patients who achieved complete response, partial response, relapses, and adverse events related to infections, leukopenia, and amenorrhea, was extracted from the selected studies. Relative risks (RRs) were estimated for 95% CI.

Statistical analysis

The weighted values of the effect measure for the group of dichotomous quantitative variables were estimated using the Mantel-Haenszel method. Heterogeneity was initially assessed by visual inspection of the Forest Plots, and statistically the Q test or chi-square (χ²) test statistics were used, considering the rejection of the homogeneity hypothesis with a p value ≤ 0.10. likewise, the I² test was applied to evaluate the percentage of heterogeneity (< 40% low, 30–60 % moderate, 50–90 % substantial, 75–100 % considerable). The overall effect measure was assessed according to the level of heterogeneity: when I² > 50%, the common-effect model was interpreted, and when I² ≤ 50%, the random-effect model was interpreted. Sensitivity analysis and analysis by subgroups were applied according to the subtype of nephritis included in the studies. The statistical software R Studio version 4.3.2.was used for the statistical analysis.

Results

A total of 415 articles were found as a result of the search, of which only 3 publications met the eligibility criteria (Fig. 1). A total of 406 patients were included, of which 217 were randomized to the high-dose group and 189 to the low-dose group. All studies had different CR criteria, but maintained levels of 24- hour proteinuria < 700 mg/dl (Table 1). It is clarified that the RC time was 6 months for the studies conducted by Zhang et al.¹⁴ and Mitwalli et al.,¹⁵ while for the one of Mehra et al.¹⁰ it was 12 months.

Fig. 1.

Information selection process.

Table 1.

Summary of the selected studies.

Author, year n	Complete remission	Definition of high doses of cyclophosphamide n	Definition of low doses of cyclophosphamide n	Complete response high doses cyclophosphamide n	Complete response low doses of cyclophosphamide n
Zhang et al., 2014^14n = 225	Proteinuria <0,3 g/24 h	Total: 8 pulses distributed as follows: 6 pulses of cyclophosphamide: monthly pulses of 500 mg/m^2, followed by 2 pulses of 250 mg/m^2 every 4 months n = 112	12 biweekly pulses at fixed doses of cyclophosphamide 400 mg and subsequently 6 monthly pulses of 400 mg n = 112	n = 35 (32.7%)	n = 30 (28%)
Mitwalli et al., 2011^15n = 117	Proteinuria <0.3 g/24 h or improvement in creatinine levels >50% of baseline	Cyclophosphamide 10 mg/kg weight monthly × 6 months, then every 2 months for 12 months n = 73	5 mg/kg/monthly for 6 months, then every 2 months × 36 months n = 44	n = 25 (34.2%)	n = 11 (25%)
Mehra et al., 2018^10n = 75	Proteinuria creatininuria ratio UPCR <0.5, GFR >90 ml/min, stable creatinine (<10% deterioration with respect to baseline)	Cyclophosphamide 750 mg/m^2 IV monthly for 6 months n = 38	Cyclophosphamide 500 mg every 15 days, 6 doses n = 37	n = 24 (65%)	n = 17 (44%)

Bias risk assessment

The RoB2 Assessment Form tool was used for the different domains: biases in the randomization process, randomization sequence, incomplete data, measurement of results, selection of reported results, independently performed by two evaluators (JM, AA), considering a “low risk” in the study conducted by Mehra et al.,¹⁰ "some concern" for the study by Zhang et al.,¹⁴ due to the lack of data regarding the randomization process and no data on the blind, some incomplete or unavailable data, and a "high risk" for the study conducted by Mitwalli et al.¹⁵ (Fig. 2).

Fig. 2.

Bias diagram.

The mean age was 31.2±5.36 years versus 30.1±1.55 in the low- and high-dose group; when evaluating the primary outcome "CR", the experimental group "low dose CYC" was less likely to achieve complete remission compared to the high dose group, with no statistical difference (RR = 0.77; 95% CI: 0.60–1.01; p = 0.06) in the common-effects model, given the homogeneity of the studies (I² = 0%; Q test = 0.55; p = 0.75). the sensitivity analysis for CR showed no significant differences (RR = 0.77; 95% CI: 0.59–1.01; p = 0.62) (Fig. 3).

Fig. 3.

Forest Plot for complete response (CR) and sensitivity analysis.

Common-effect model

Regarding the partial response (PR), it was achieved in 228 subjects, being more frequent in those who received high doses, with no significant differences between the studies in the random-effects model (RR = 0.88; 95% CI: 0.67–1.16; p = 0.35) due to the high heterogeneity of the studies (I² = 58.4%; Q = 4.81; p = 0.09) (Fig. 4a).

Fig. 4.

(a) Forest Plot for partial response (PR), sensitivity analysis (b) Forest Plot for partial response (PR), sensitivity analysis excluding the study conducted by Zhang et al.¹⁴

Random effect model

Given the high heterogeneity in the above-mentioned outcome, a sensitivity analysis was explored, excluding the study conducted by Zhang et al.¹⁴ for PR, and it was evident that it continued to occur in a lower proportion in the low-dose group, but with a significant difference (RR = 0.77; 95% CI: 0.61−0.97; p = 0.026) (Fig. 4b). When assessing the secondary outcomes, the proportion of infections was lower in the low-dose group, not significantly in the common-effect model (RR = 0.68; 95% CI: 0.45–1.04; p = 0.08) due to the homogeneity of the studies (I² = 12.5%; Q test = 2.27; p = 0.32). In the sensitivity analysis for infections, no differences were found between the studies either (RR = 0.69; 95% CI: 0.43–1.13; p = 0.08) (Fig. 5). To assess renal relapses, only two studies (Mehra et al.,¹⁰ Zhang et al.¹⁴) included this outcome, with fewer relapses in the high-dose group and no statistical significance in the random-effects model (RR = 2.16; 95% CI: 0.18–24.7; p = 0.52) (I² = 78%; Q test = 4.55; p = 0.03) (Fig. 6).

Fig. 5.

Forest Plot for infections and sensitivity analysis.

Fig. 6.

Forest Plot for renal relapses.

Leukopenia was significantly less severe in the low-dose group in the common-effect model (RR = 0.41; 95% CI: 0.22−0.77; p = 0.01) (I² = 0%; Q test = 1.38; p = 0.50). Amenorrhea also occurred more infrequently in the low-dose scheme, significantly (RR = 0.40; 95% CI: 0.25−0.64; p = 0.002) (I² = 0%; Q test = 0.02; p = 0.98) (Fig. 7).

Fig. 7.

Forest Plot for leukopenia and amenorrhea.

Subsequently, the analysis was carried out by subtypes of nephritis included in the clinical trials, the CR was evaluated according to the subtype of nephritis (III-IV, versus IV), and no significant difference was found (RR = 0.79; 95% CI: 0.60–1.03; p = 0.77) (I² = 0%; p = 0.84) (Fig. 8).

Fig. 8.

Analysis by subgroups according to the subtype of nephritis.

Since the temporal definition of CR showed differences in the three trials, as exposed above, a subgroup analysis of CR at 6 and 12 months, respectively was performed, and it was found that in the group of 6 months the weighted effect was RR = 0.81 (95% CI: 0.57–1.13; I² = 0%; p = 0.67), with no significant differences in the test of differences between subgroups (p = 0.55) (Fig. 9).

Fig. 9.

Analysis by subgroups according to the time to complete response (CR).

When carrying out the analysis according to heterogeneity, no significant differences were observed depending on the type of analysis used.

Discussion

The regimen of treatment with cyclophosphamide for lupus nephritis continues to be effective for different response criteria, such as relapse rate, complete response, partial response, and improved safety profile. There have been few meta-analytic reviews comparing the effect of low doses versus high doses of this drug, but they agree on their similar efficacy profile and greater safety with the use of low doses. Li et al.¹⁶ found in their systematic review that the low-dose regimen is associated with a decrease in 24 h proteinuria (MD: –0.45; 95% CI: 0.62−0.27) and a lower incidence of major infections (OR = 0.62; 95% CI: 0.40−0.95), while high doses were associated with a lower probability of doubling creatinine values. Our research made similar findings, although when updating the data to 2023 only one additional trial (Mehra et al.¹⁰) was included, in which no changes with respect to creatinine levels and dosage schedule were observed.

In the evaluation of the included articles, the study conducted by Zhang et al.¹⁴ had a prospective design, with active intervention (low-dose cyclophosphamide) and comparator group (high-dose cyclophosphamide) to evaluate results in terms of efficacy and safety, although it has many limitations from a point of view of design, that decrease reliability and power, such as non-randomization and lack of description of the blinding process, in addition to the lack of description of the sample size calculation process and the analysis by protocol. It was included in the meta-analysis because it met eligibility criteria, but the respective risk of bias is described, so it was removed when performing the sensitivity analysis.

Leukopenia and amenorrhea have occurred less frequently in patients exposed to low doses of cyclophosphamide and mycophenolate mofetil (MMF), as well as in multi-target therapies, which combine other immunosuppressants such as tacrolimus + mycophenolate + steroids, versus CYC + steroids, with a lower probability that these adverse events occur. Zhou et al.¹⁷ demonstrated in a systematic review a lower risk of leukopenia in those with multi-target therapies versus CYC (OR = 0.38; 95% CI: 0.22−0.67; p = 0.0009), similar to our findings, in addition to the lower proportion of menstrual irregularities, although this group had a higher risk of high blood pressure.

Although the available literature regarding to randomized trials comparing the two schemes of doses of cyclophosphamide, it can be objectively demonstrated that the measures of effect favor the use of low-dose schemes over high-dose schemes in terms of efficacy, without significant differences, and likewise it has been demonstrated overtime that at 10 years there is no variation in the risk of mortality, deterioration of serum creatinine, or progression to end-stage renal disease, between the two schemes.¹⁸

Low doses of cyclophosphamide continue to be used as a first measure in the different clinical practice guidelines, but, due to its efficacy, MMF in induction therapy has proven to be an equivalent option with a better safety profile. In a network meta-analysis conducted by Bae et al.¹⁹ low and high doses of cyclophosphamide were compared with MMF, and it was evidenced that the total response was similar in the low-dose and MMF group. The latter had a higher probability of reaching the total response (SUCRA = 0.74) followed by low doses of cyclophosphamide (SUCRA = 0.69). In terms of safety, low-dose schedules of cyclophosphamide and MMF were also favored.

The demonstrated short-term efficacy in the complete renal response has been evaluated in follow-up studies. Houssiau et al.²⁰ demonstrated that 16% of patients in the low-dose scheme presented treatment failure versus 20% in the high-dose group; nevertheless, although there was no difference in the likelihood of progression to chronic kidney disease and doubling of creatinine levels between the two groups, it has been suggested that early response at 6 months (75% reduction in proteinuria) was the best predictor of good long-term outcomes.²¹

In the 10 year follow-up of a cohort of patients exposed to the two regimens, the poor renal response (serum creatinine >1.4 mg/dl) did not differ between the groups (low doses 22% and high doses 23%), but there was an increase in cumulative doses of cyclophosphamide.¹⁸

Although the scheme of high doses of cyclophosphamide has been associated with higher cumulative doses, there is no-specific cut-off point that determines the increased risk. Bellomio et al.²² described the risk of infections in relation to the doses of cyclophosphamide, and demonstrated that the higher the cumulative dose, the greater the risk of adverse events, where infections corresponded to 45%, with a cumulative dose of 2.600 mg present at the time of the adverse event. Likewise, the risk of ovarian failure is estimated at 8 g/m², hematological neoplasia 36 g/m² and myelofibrosis 80 g/m².²³

There are other studies with different methodologies and representative samples, such as the one conducted by Houssiau et al.,²⁰ which also show a higher frequency of infections with high doses of cyclophosphamide, especially in patients with leukopenia, pulmonary hypertension and use greater than 15 mg day of steroid. Houssiau et al.²¹ described that high and low doses are comparable in early stages of treatment, but at 48 months the high-dose scheme had better remission rates; however, in the groups evaluated there were differences in the activity index, presence of thrombotic microangiopathy, and the cumulative dose of steroid was not described, which could have affected the results.

Mycophenolate versus cyclophosphamide has in turn demonstrated better control of immune activity. In another systematic review, Jiang et al.²⁴ showed that MMF was more effective than cyclophosphamide in increasing C3 levels (SMD = 0.47; 95% CI: 0.23−0.71) and higher probability of CR (RR = 1.23; 95% CI: 1.0–1.43), in addition to a lower risk of infection and leukopenia (RR = 0.19; 95% CI: 0.07−0.45). In our systematic review, it was significantly evidenced that low doses of cyclophosphamide were associated with less severe leukopenia, which also predisposes to a lower risk of opportunistic infections.

Although the efficacy of mycophenolate has been greater, multi-target therapy (MMF + tacrolimus) also has a better response than cyclophosphamide. Deng et al.²⁵ evidenced a greater CR (RR = 1.94; 95% CI: 1.61–2.33; p < 0.01) of multi-target therapy versus cyclophosphamide, as well as a lower proportion of infections and leukopenia. However, these data have limitations when it comes to external validation, especially because they were developed in Asian populations.

Similar findings about the effectiveness of mycophenolate were demonstrated by Chen et al.,²⁶ who made an evaluation of different systemic reviews that confirm the greater probability of complete response (RR = 4.33; 95% CI: 1.45–12.91) for this group of patients, improving the quality of the evidence regarding the use of low doses of cyclophosphamide and MMF as an excellent choice for the safe induction of patients with lupus nephritis.

Among other therapies available for the induction management of lupus nephritis, rituximab (RTX) has shown favorable results compared to the standard of care CYC + steroids. Li et al.²⁷ assessed, based on a network meta-analysis the efficacy and safety or the four drugs: RTX, CYC, MMF and tacrolimus, and evidenced that although there were no difference in achieving PR, RTX showed a higher probability of achieving CR than MMF (OR = 2.6; 95% CI: 1.0–7.1), significantly higher than CYC (OR = 4.2; 95% CI: 1.7–14), as well as the highest ranking with respect to the selection of the best therapy for RTX (SUCRA = 96.9%), but also the highest for infections (SUCRA = 74.9%).

A factor that may influence the response rate to existing therapies in the management of lupus nephritis is the racial group. In this sense, Mejía et al.²⁸ compared the efficacy of the different treatment regimens (mycophenolate, cyclophosphamide, azathioprine) in the Mexican population, with superior results in the mycophenolate group compared to cyclophosphamide (HR = 2.0; 95% CI: 1.23–3.25; p = 0.005) in relation to the complete renal response.

Other factors are related to a poor renal prognosis, such as histopathological findings due to extracapillary proliferation (HR = 2.70; 95% CI: 1.02–6.67; p = 0.0469), glomerular sclerosis (HR = 1.69; 95% CI: 1.34–2.21; p < 0.0001) and fibrous crescents (HR = 9.98; 95% CI: 1.12–79.4; p = 0.04).²⁹ The recommendations favor the use of a high-dose regimen of cyclophosphamide versus low-doses,³⁰ although there is no difference with respect to the use of mycophenolate, which favors a faster response.³¹

With reference to maintenance treatment, mycophenolate also favors the response in Hispanic and African-American populations. Contreras et al.³² demonstrated that mycophenolate and azathioprine have a higher probability of survival free of compound events: death or chronic kidney disease versus cyclophosphamide as maintenance (MMF = 89%, AZA = 80%, CYC = 45%).

The weaknesses of this systematic review and meta-analysis lie in the small number of studies included, given the variability found in the measurement of outcomes. On the other hand, the limited availability of information regarding the randomization process in two of three of the trials confers a high risk of bias and uncertainty. Only one study (Mehra et al.¹⁰) had a low risk of bias in the assessment conducted by domains, which encourages the periodic updating of evidence to properly guide clinical decision-making.

Another important limitation of our research lies in the lack of homogeneity in the definition of cyclophosphamide dosage schemes, given that in the low-dose groups were used protocols which had cumulative doses higher than 3 g, which could bias the estimation of the effect. Likewise, as there were differences in the temporal estimation of CR time between the trials, the effect could have been modified by maintenance doses in those patients in whom the response at 12 months was evaluated. which could bias the interpretation of the results. However, when doing the subgroup-weighted effect analysis according to response time, we did not find significant differences.

It is essential to conduct further studies that include all published trials, as this article only includes an update of the evidence.

Conclusion

The use of low doses of cyclophosphamide as induction therapy in lupus nephritis is as effective as the high-dose scheme in achieving the outcomes of complete response and partial response. In addition, the proportion of renal relapses in the follow-up period of up to 6 months confers a better safety profile in the context of a significantly lower probability of leukopenia and amenorrhea.

CRediT authorship contribution statement

PAQM: protocol writing, bibliographic search, selection of articles, statistical analysis, writing of the document.

JMMM: selection of articles, bias assessment, writing of the document.

PAPB: selection of articles.

AAA: evaluator of the selected articles, review of disagreements and evaluator of biases.

LMEM contribution in the elaboration of tables, writing of the discussion.

DIMR: professor supervisor of the project, corrections of the manuscript, contribution to the discussion.

Funding

The authors declare that they have not received any funding for the execution of this research.