Rheumatology 2009;48:944–952
Advance Access publication 3 June 2009
doi:10.1093/rheumatology/kep120
Mycophenolate mofetil is as efficacious as, but safer than,
cyclophosphamide in the treatment of proliferative lupus nephritis:
a meta-analysis and meta-regression
Anselm Mak1, Alicia A. C. Cheak1, Jason Y. S. Tan1, Hnin Cho Su1, Roger C. M. Ho2
and Chak Sing Lau3
KEY
WORDS:
Mycophenolate, Cyclophosphamide, Lupus, Nephritis, Meta-analysis, Meta-regression.
efficacious than CYC for inducing renal remission, preventing
flares and prolonging survival is still uncertain because results
from these trials which included relatively small numbers of
patients with short duration of observation, have not been in
complete agreement. Although three systemic reviews have been
published in an attempt to answer these questions [26–28], the
results of these meta-analyses are inconsistent, chiefly because of
the small number of trials included, small sample sizes of the trials
and the lack of uniformity of methodology used in these analyses.
Additionally, heterogeneity of the trials and the potential factors
which contribute to heterogeneity were not examined in the previous meta-analyses. We therefore carried out this meta-analysis
with an aim to compare the efficacy and safety of MMF and CYC
in the treatment of proliferative lupus GN by including the most
recently published large randomized controlled trials (RCTs),
conducting sensitivity analyses to test the robustness of the
effect sizes and exploring trial- and patient-related factors
which potentially contribute to heterogeneity of the trials by
meta-regression.
Introduction
SLE is a multisystemic autoimmune disease with protean clinical
manifestations and organ involvement. Proliferative glomerulonephritis (GN) is one of the most important organ manifestations of
lupus affecting between one-half and two-thirds of patients during
their disease course, leading to significant morbidity and mortality
[1–3]. Although cyclophosphamide (CYC)-containing regimens
have long been considered a gold standard in inducing renal
remission and preventing renal flares for patients with proliferative lupus GN [3, 4], its significant toxicities raise a number of
concerns [5]. Further, while CYC induces renal remission in a
significant proportion of patients with proliferative lupus GN,
the rate of disease relapse is considerable. Observational studies
have demonstrated that the 5-year cumulative rate of renal relapse
was as high as 44% [6, 7] and up to 15% of the patients with
diffuse proliferative GN were refractory to CYC [8, 9].
Mycophenolate mofetil (MMF), a lymphocyte selective antiproliferative agent which was first introduced into clinical use
over a decade ago for preventing solid organ transplant rejection,
has emerged as an efficacious and safe immunosuppressive agent
for a number of lupus-related conditions [10–14]. The use of
MMF in proliferative lupus GN has been repeatedly tested in a
number of controlled trials [15–25]. While the majority of these
trials showed that MMF was well tolerated and had a more
favourable safety profile than CYC, whether MMF is more
Materials and methods
Search strategy
We performed an extensive literature search using the relevant
keywords of ‘mycophenolate’, ‘mycophenolic’, ‘lupus nephritis’,
‘nephritis’ and ‘glomerulonephritis’ to identify RCTs published
in the English language from different computerized databases:
PubMed (1966 to April 2008), EMBASE (1980 to January 2008)
and the Cochrane Central Register of Controlled Trials
(2nd quarter of 2008). Abstracts presented in major international
conferences (Annual European Congress of Rheumatology,
International Congress on SLE and ACR meeting) over the past
10 years were manually searched. We also scanned the articles
from the bibliographies of the retrieved trials and review articles.
1
Division of Rheumatology, Department of Medicine, 2Department of Psychological
Medicine, National University of Singapore, Singapore and 3Department of
Medicine and Therapeutics, University of Dundee, Dundee, UK.
Submitted 10 November 2008; revised version accepted 15 April 2009.
Correspondence to: Anselm Mak, Yong Loo Lin School of Medicine, National
University of Singapore, Department of Medicine, National University Hospital,
5 Lower Kent Ridge Road, Singapore 119074. E-mail: [email protected]
944
ß The Author 2009. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
Objective. Since mycophenolate mofetil (MMF) has emerged as an immunosuppressant for treating proliferative lupus nephritis, the role of
cyclophosphamide (CYC)-containing regimens is being challenged. Efficacy data from randomized controlled trials (RCTs) and previous
meta-analyses comparing these two agents for treating lupus nephritis have been inconsistent as they were heterogeneous in design and of
small sample size. An updated meta-analysis is therefore required.
Methods. Publications in the English literature were searched with the keywords ‘mycophenoate’, ‘mycophenolic’, ‘lupus nephritis’, ‘nephritis’
and ‘glomerulonephritis’ for RCTs in electronic databases. Primary outcome was relative risk (RR) of renal remission at 6 months. Secondary
outcome included RRs of mortality, development of end-stage renal failure (ESRF) and side effects. Meta-regression was performed to
identify factors explaining the heterogeneity of the effect sizes.
Results. Ten eligible RCTs involving 847 patients were included. MMF offers similar efficacy in inducing renal remission as CYC (RR 1.052;
95% CI 0.950, 1.166) and the risks of death (RR 0.709; 95% CI 0.373, 1.347) and ESRF (RR 0.453; 95% CI 0.183, 1.121) were comparable.
Significantly fewer patients receiving MMF developed amenorrhoea (RR 0.212; 95% CI 0.094, 0.479) and leucopenia (RR 0.473; 95% CI
0.269, 0.832) while the risks of herpes infection and pneumonia tended to be lower and that of diarrhoea appeared higher in the MMF groups.
Meta-regression revealed that the non-white and non-Asian ethnicities contributed significantly to the heterogeneity of the effect sizes of
renal remission.
Conclusion. MMF offers similar efficacy in renal remission and survival as CYC. MMF appears safer than CYC in the treatment of
proliferative lupus nephritis.
Meta-analysis of MMF and CYC in lupus nephritis
The authors of correspondence were contacted for obtaining
information essential for this meta-analysis which was lacking in
the published articles.
Criteria for selecting articles included in this meta-analysis
Outcome measures of this meta-analysis
The primary outcome was the proportion of patients who
achieved complete and partial renal remission at 6 months after
induction therapy with MMF or CYC. The definition of complete
and partial renal remission of this meta-analysis was based on
individual studies’ remission criteria, which were essentially
similar to one another. Secondary outcome measures included
the proportion of patients who died, developed ESRF and
other adverse effects including amenorrhoea, leucopenia, herpes
zoster, diarrhoea and pneumonia at the end of the respective
studies.
Assessment of quality of trial
The quality of each trial was assessed according to a standard
scoring system proposed by Jadad et al. [29]. The assessment
was based on: (i) whether the randomization method was
appropriate, (ii) whether double blindness is mentioned in the
trial and whether it was appropriately performed; and (iii) whether
the number of patients of and the reasons for withdrawal and
drop-outs were clearly stated. The score ranges from 0 to 5 with
higher scores denoting better quality of a trial.
In order to comply with the Cochrane Handbook for
Systematic Reviews of Interventions in terms of quality assessment of the RCTs [30], we additionally assessed whether proper
allocation concealment during randomization was performed in
the RCTs eligible for this meta-analysis because allocation
concealment is not assessed in the system suggested by Jadad
et al. [29].
Statistical analysis
The proportion of patients who achieved renal remission at
6 months following induction therapy with either MMF or CYC
was pooled for combined relative risk (RR). The RRs of the predefined secondary outcomes at the end of the study period were
pooled. Effect sizes of both the primary and secondary outcomes
were expressed as RR and the corresponding 95% CI.
Heterogeneity was assessed by the Cochran Q-test. Since the
number of trials of this meta-analysis is relatively limited, the
Cochran Q-test for heterogeneity may yield a low statistical
power [31]. A value of significance at 10% (P 4 0.1) was therefore
considered statistically significant for heterogeneity [32].
Furthermore, we assessed heterogeneity with I2, which describes
the percentage of total variation across studies caused by heterogeneity rather than chance. High values of I2 suggest increased
heterogeneity. The fixed effects model was used for statistically
significant homogeneous studies. If considerable heterogeneity
(arbitrarily if I2 > 40) was encountered, we applied the random
effects model with the method suggested by DerSimonian and
Laird [33].
For models with considerable heterogeneity, meta-regression
was performed to identify patient- and trial-related factors
which might contribute to the heterogeneity [34]. Trial-related
factors included the sample size, duration of study, proportion
of patients who were on CYC and quality of trials while
patient-related factors were gender, age, duration of lupus or
lupus nephritis, proportion of patients with Class IV nephritis
and ethnicities (proportions of non-white and non-Asian
patients). Chronicity and activity indices of renal biopsy were
not included for meta-regression because they were not reported
in a number of studies [19, 22–24]. Since the covariates chosen
were not expected to explain all the heterogeneity of the trials,
mixed-model regression was used in considering the presence of
‘residual heterogeneity’ [34]. The regression coefficients and
the associated standard error (S.E.), the z-score and P-values
were reported for the meta-regression analysis.
Sensitivity analyses were performed for testing the robustness
of the effect sizes and the analyses were divided into two parts.
In the first part, we evaluated the robustness of the pooled effect
sizes based on elimination of publication which is the extension of
an original cohort, route of administration of CYC (i.e. intravenous and oral administrations) and quality of trials. Because the
observation time and duration of treatment differed between trials
which assessed MMF and CYC as maintenance agents in the
treatment of lupus nephritis, in the second sensitivity analysis,
we tested the robustness of the RRs of the secondary outcome
by comparing them with their corresponding incidence rate
ratios (IRRs), which were derived by pooling the incidence per
patient-years follow-up. This is important because theoretically
the occurrence of events in the secondary outcomes would be
higher with longer duration of treatment and/or observation.
The patient-years follow-up was estimated from the product
of the number of patients and mean duration of follow-up in
respective studies.
Publication bias was not assessed because it has been proven to
be unhelpful [35]. All statistical analyses in this meta-analysis were
performed using the Comprehensive Meta-analysis Programme,
Version 2 (Biostat, Englewood, NJ, USA). The QUORUM
statement for improving the quality and result of meta-analyses
of RCTs was adhered where appropriate [36].
Results
We initially identified 438 articles through database searches. All
abstracts were scanned and 16 studies and 3 abstracts were
retrieved for detailed scrutiny after we rejected 419 abstracts
because they were non-human studies, observational studies,
case reports and letters to the editors or editorials. We further
excluded nine publications because they were reports of a metaanalysis (n ¼ 3), an uncontrolled trial (n ¼ 1), a non-randomized
trial (n ¼ 1) and RCTs which were not comparing MMF and CYC
(n ¼ 2), review of an RCT (n ¼ 1) and an introduction of an RCT
(n ¼ 1). We identified one trial from the bibliographies of one of
the retrieved articles, but it was subsequently excluded due to
insufficient analysable information available in that report [25].
We realized that one of the RCTs [20] was partly an extension of
observation of a previously reported cohort [15]. However, since
there were some new patients in the latter trial and invaluable
information on safety profile, ESRF development and mortality
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
All trials which randomly assigned patients to receive either MMF
or CYC for the management of proliferative lupus GN were
included. RCTs were included if they met the following criteria:
(i) compare MMF and CYC as induction and/or maintenance
therapy; (ii) concomitant therapy with corticosteroids equally to
both treatment arms; and (iii) at least one of the following four
outcomes were reported: mortality, end-stage renal failure
(ESRF), achievement of complete and partial renal remission
at 6 months and common adverse effects including herpes zoster
infection, pneumonia, leucopenia, amenorrhoea and diarrhoea.
Four investigators (A.M., J.Y.S.T., R.C.M.H. and H.C.S.)
independently assessed the papers generated for relevancy and
papers with the following exclusion criteria were excluded:
(i) not written in English for abstract; (ii) not comparing MMF
and CYC in treating proliferative lupus GN; and (iii) investigating
patients with pure membranous lupus GN. Data were independently extracted into a standard electronic data extraction form.
Any discrepancies were resolved by consensus. If consensus could
not be reached, the principal investigator (A.M.) would make
the final decision for trial eligibility and data extraction. The
senior author (C.S.L.) is the overall advisor of this meta-analysis.
945
Anselm Mak et al.
946
438 abstracts
retrieved under
search engines
419 abstracts
rejected
19 abstracts
accepted
4 studies are not
controlled trials
1 study from
15 studies are
controlled trials
bibliographies
5 studies were
rejecteda
10 studies were
eligible for
analyses
FIG. 1. Results of literature search. aReasons of rejection: (i) uncontolled trial; (ii) review of an RCT; (iii) introduction of an RCT; and (iv) two RCTs not comparing MMF
and CYC.
TABLE 1. Characteristics and quality of controlled trials comparing MMF and CYC in patients with proliferative lupus nephritis
Study duration (weeks)
Study (ref.)
Publication Study
type
design
Chan et al. [15]
Hu et al. [16]
Contreras et al. [17]
Full text
Full text
Full text
RCT
RCT
RCT
Ong et al. [18]
Ginzler et al. [19]
Chan et al. [20]
Wang et al. [21]
Flores-Suarez et al. [22]
Isenberg et al. [23]
Mulic-Bacic et al. [24]
Full text
Full text
Full text
Full text
Abstract
Abstract
Abstract
RCT
RCT
RCT
RCT
RCT
RCT
RCT
Comparisona
n
MMF (n ¼ 21) vs oral CYC (n ¼ 21)
42
MMF (n ¼ 23) vs IV CYC (n ¼ 23)
46
59
IV CYC (n ¼ 59) and then
MMF (n ¼ 20) or Aza (n ¼ 19) or
IV CYC (n ¼ 20)
MMF (n ¼ 19) vs IV CYC (n ¼ 25)
44
MMF (n ¼ 71) vs IV CYC (n ¼ 66)
137
MMF(n ¼ 33) vs oral CYC (n ¼ 31)
64
MMF (n ¼ 9) vs IV CYC (n ¼ 11)
20
MMF (n ¼ 10) vs IV CYC (n ¼ 10)
20
MMF (n ¼ 185) vs IV CYC (n ¼ 185) 370
MMF (n ¼ 20) vs IV CYC (n ¼ 25)
45
Patient-years follow-up
Age
(mean)
Induction
Maintenance
37.5
28.8
32.7
24
24
28
24
NA
–b
21
11.5
48.3
21
11.5
41.6
1
1
2d
30.9
31.8
39.9
31.5
NA
NA
NA
24
24
24
24
24
24
24
NA
NA
–c
NA
NA
NA
NA
9.5
36.2
144.6
4.5
10
92.5
10.5
12.5
28.1
144.6
5.5
10
92.5
12.5
3d
3d
3
2d
1
2
1
MMF group CYC group
Jadad score
(0–5)
a
All patients received corticosteroids. bThe median duration of treatment was 25 months and 29 months in the CYC group and MMF group, respectively. Maintenance therapies after CYC induction
were MMF, AZA and quarter IV CYC. cThe mean duration of treatment was 63.9 and 52.2 months in the CYC and MMF groups, respectively. Maintenance therapy in the MMF could be MMF or AZA
while that of the CYC group was AZA. dAppropriate allocation concealment was performed during randomization process (see text for details). IV; intravenous; NA, not available.
could be retrieved from this extended long-term study which
spanned a median follow-up of 63 months [20], we decided to
include it in the meta-analysis. The definitive analysis in this
meta-analysis included 10 RCTs that were published between
2000 and 2008. The result of the literature search is shown
in Fig. 1.
Among the 10 studies, 7 were published in full text while 3 were
published in abstract form. Two of the 10 studies involved
comparison of MMF and CYC in both the induction and
maintenance phases [15, 20] and one compared MMF, CYC and
AZA as maintenance therapies after successful induction with
CYC for a duration between 4 and 7 months [17]. Apart from
the two studies which involved the use of oral CYC [15, 20], all
tested intravenous CYC against MMF. A total of 847 patients
participated in the controlled trials. Amongst them, 411 were
randomly assigned to receive MMF and 417 received CYC.
Nineteen patients received AZA as maintenance therapy in one
of the trials [17]. Seventy-four and 20 patients continued MMF
and quarterly intravenous CYC as maintenance therapy, respectively, as noted in the three maintenance trials. The number of
participants in individual trials ranged from 20 to 370. The mean
(range) age of the patients was 33.3 (28.8–39.9) years and the
majority of the patients were female (ranging from 82.6 to
94.0%). The calculated mean (range) patient-years follow-up
was 38.9 (4.5–144.6) for MMF therapy and 38.0 (5.5–144.6) for
CYC therapy. We used the median to estimate the patient-years
follow-up for one of the trials because the mean duration of
follow-up was not provided [17] (Table 1).
For the quality of the studies as assessed by the Jadad
method [29], the mean Jadad score was 2 (range 1–3) and the
majority of the trials (six studies) were of low quality (Jadad
scale <2) (Table 1). Double blindness was impossible for most
of the studies because patients received either CYC intravenously
or MMF orally while the use of double dummy was not
mentioned in any of the study protocols. Although two studies
involved the use of oral CYC and MMF, implementation of
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
Insufficient
information,
excluded
10 controlled
trials accepted
Meta-analysis of MMF and CYC in lupus nephritis
double dummy was also not described [15, 20]. In terms of allocation concealment during randomization, only four out of the
10 studies recruited may have performed it appropriately. Three
studies randomized subjects with the use of sealed envelopes at
central sites [17, 19, 21], whereas one used randomization codes
that were generated separately for each participating centre by
using the random permuted block method with randomly varying
block size [18] (Table 1). Despite these, authors of these trials
did not explicitly state whether any investigators involved in
the recruitment and randomization processes had any role in
generating those sealed envelopes or randomization codes.
Agreement between investigators
Primary outcome
Data of eight studies were involved in pooling estimates for renal
remission rate during induction therapy because one trial was a
maintenance study which did not compare MMF and CYC as an
induction therapy [17] and another one did not report data on
renal remission [16]. MMF is comparable with CYC in the ability
to induce complete and partial remission at 6 months after
commencing the induction therapy (RR 1.052; 95% CI 0.950,
1.166; random effects model) (Fig. 2). Cochran’s Q and I2 statistics revealed a moderate degree of heterogeneity in these
trials (I2 ¼ 40.775; P ¼ 0.1). The random effects model was therefore used.
Secondary outcome
During the whole study period of all the trials, the RR of all-cause
mortality was comparable between patients who received MMF
and CYC (RR 0.709; 95% CI 0.373, 1.347; fixed effects model).
Two studies reported no mortality and they were therefore not
estimable for effect sizes [16, 21]. As far as the development of
ESRF is concerned, the risk of ending up with ESRF at the end of
study period was comparable between patients who took MMF
and CYC (RR 0.453; 95% CI 0.183, 1.121; fixed effects model).
Although patients in the MMF tended to have lower risks
of death and development of ESRF, it did not reach statistical
FIG. 2. Forest plot of the primary outcome: complete and partial remission.
significance. Only four studies were pooled for effect sizes of
ESRF because no ESRF was reported in three trials [15, 16, 21]
while no data were available in three studies [22–24].
Heterogeneity is minimal when effect sizes of mortality and
ESRF were assessed (Fig. 3).
Adverse events
Significantly fewer patients who received MMF developed
amenorrhoea (RR 0.212; 95% CI 0.094, 0.479, fixed effects
model) and leucopenia (RR 0.473; 95% CI 0.269, 0.832).
Patients in the MMF group tended to have a lower risk of developing herpes infection (RR 0.7; 95% CI 0.391, 1.251, fixed effects
model) and pneumonia (RR 0.565; 95% CI 0.235, 1.360, fixed
effects model). However, patients receiving MMF appeared to
have a higher risk to experience diarrhoea (RR 2.078; 95% CI
0.982, 4.397, fixed effects model), though it did not reach statistical significance. Apart from the assessment of the effect size
of diarrhoea where mild heterogeneity was found (I2 ¼ 23.158),
heterogeneity was undetectable when the effect sizes of other
side effects were evaluated (I2 ¼ 0). The fixed effects model was
thus used (Fig. 4).
Sensitivity analyses
The results of the primary outcome and the secondary outcomes
including the RRs of mortality, ESRF and side effects such as
herpes infection, pneumonia and amenorrhoea remained generally
consistent upon sensitivity analyses based on exclusion of a study
which is an extended observation of an original cohort, route of
administration of CYC and quality of trials (Table 2). When
only high-quality trials were pooled for analysis, the incidence
of leucopenia became insignificant. Interestingly, the RR of
diarrhoea was significantly worse in patients who received
MMF when only high-quality trials were analysed. Additionally,
the risk of death of patients who took MMF was much lower
when only studies of high-quality and trials testing oral CYC
were pooled for analyses, although the pooled effect size remained
statistically insignificant. Nevertheless, these results must be
interpreted with caution because the effect sizes were generated
from a small number of high-quality studies (n ¼ 3) and trials only
involved oral CYC (n ¼ 2), respectively [15, 20] (Table 2).
In the second part of the sensitivity analysis where effect sizes in
terms of RRs and IRRs were compared, all the results remained
robust except for the higher heterogeneity of the effect sizes of
diarrhoea, herpes and pneumonia (Table 3). Different treatment
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
The inter-rater reliability agreement of the four investigators
(A.M., J.Y.S.T., R.C.M.H and H.C.S) in terms of inclusion and
exclusion of studies and quality of papers assessed by the Jadad’s
method were 0.86 and 0.80, respectively, calculated based on the
Fleiss’ -statistic [37]. Such level of agreement is considered to be
substantially almost perfect [38].
947
948
Anselm Mak et al.
and observation time between these trials could partly explain the
increased heterogeneity when IRRs were pooled. Due to the
small number of trials which demonstrated moderate heterogeneity (n ¼ 5; df ¼ 4), meta-regression was therefore not performed.
Meta-regression
Meta-regression was performed for assessment of heterogeneity of
the effect size of renal remission because a moderate degree of
heterogeneity was evident (I2 ¼ 40.775) (Fig. 2). Meta-regression
demonstrated that the proportion of non-white and non-Asian
patients contributed significantly to the heterogeneity of the
effect size of renal remission (P ¼ 0.050). Other factors such as
the mean age of patients, sample size, duration of study, route
of CYC administration, trial quality, proportion of Class IV
lupus nephritis and duration of SLE/nephritis did not significantly
contribute to the heterogeneity of the renal remission effect
size (Table 4). Although the effect sizes (IRRs) of diarrhoea and
pneumonia were quite heterogeneous, meta-regression was
not performed because of the small number of trials involved
(Table 4).
Discussion
The current meta-analysis of 10 RCTs involving over 800 patients
with proliferative lupus GN indicates that MMF has similar
efficacy when compared with CYC in terms of induction of
disease remission. In addition, MMF is not superior to CYC in
conferring survival benefit in these patients. Our meta-analysis
generally agrees with previously published RCT reports and systemic reviews, in that MMF does not provide additional benefits
in reducing the incidence of ESRF when compared with CYC
although MMF is a safer agent because of its more favourable
safety profile. Additionally, meta-regression revealed that the nonwhite and non-Asian ethnicities contributed significantly to the
heterogeneity of renal remission.
Uncontrolled experience in the use of MMF in patients with
proliferative lupus GN was first reported in the late 1990s when
clinicians recognized that CYC was neither an absolutely safe nor
a completely efficacious agent [39, 40]. Despite the moderately
high renal remission rate following treatment with CYC, up to
15% of the patients with proliferative lupus GN were refractory to
treatment and as many as 50% of the patients developed ESRF
on this treatment [8, 9, 41, 42]. In addition, the safety profile of
CYC raised significant concerns, in particular amenorrhoea,
leucopenia and infection [43]. Thus, clinicians have long yearned
for a more efficacious and safer agent in the treatment of this
serious yet manageable condition of mostly women of reproductive age. Since the report of the first RCT comparing MMF with
CYC in proliferative lupus GN was published in 2000 [15], MMF
has been suggested to be safer than and at least as efficacious as
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
FIG. 3. Forest plots of the secondary outcomes: death and ESRF.
Meta-analysis of MMF and CYC in lupus nephritis
949
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
FIG. 4. Forest plots of the secondary outcomes (adverse events).
Anselm Mak et al.
950
TABLE 2. Results of sensitivity analyses (trial exclusion)
Combined RR (95% CI)
Intravenous CYC vs oral CYC
All trials
Intravenous CYC only
Oral CYC only
Extension study
All trials
Excluding Chan [15]
Excluding Chan [15]
Quality
All trials
High (3)
Low (<3)
Diarrhoea
Herpes zoster
Leucopenia
Amenorrhoea
10
8
2
2.08 (0.98, 4.40)
2.03 (0.94, 4.40)
3.00 (0.13, 69.70)
0.70 (0.39, 1.25)
0.75 (0.39, 1.45)
0.56 (0.17, 1.87)
0.47 (0.27, 0.83)*
0.54 (0.30, 0.97)*
0.10 (0.01, 0.78)*
0.21 (0.09, 0.48)*
0.26 (0.10, 0.67)*
0.11 (0.02, 0.58)*
10
9
9
2.08 (0.98, 4.40)
2.22 (0.68, 7.24)
2.25 (0.86, 5.89)
0.70 (0.39, 1.25)
0.67 (0.37, 1.24)
0.77 (0.41, 1.45)
0.47 (0.27, 0.83)*
0.49 (0.28, 0.87)*
0.52 (0.29, 0.92)*
0.21 (0.09, 0.48)*
0.22 (0.09, 0.51)*
0.25 (0.10, 0.60)*
10
3
7
2.08 (0.98, 4.40)
6.78 (1.60, 28.66)*
1.34 (0.56, 3.23)
0.70 (0.39, 1.25)
0.71 (0.30, 1.69)
0.69 (0.32, 1.51)
0.47 (0.27, 0.83)*
0.63 (0.32, 1.21)
0.22 (0.07, 0.65)*
0.21 (0.09, 0.48)*
0.16 (0.04, 0.69)*
0.24 (0.09, 0.64)*
Pneumonia
ESRF
Mortality
Remission
10
8
2
0.57 (0.24, 1.36)
0.59 (0.24, 1.48)
0.33 (0.01, 7.74)
0.45 (0.18, 1.12)
0.50 (0.19, 1.28)
0.19 (0.01, 3.77)
0.71 (0.37, 1.35)
0.81 (0.41, 1.59)
0.19 (0.02, 1.61)
1.05 (0.95, 1.17)
1.17 (0.94, 1.45)
1.01 (0.94, 1.10)
10
9
9
0.57 (0.24, 1.36)
0.59 (0.24, 1.48)
0.57 (0.24, 1.36)
0.45 (0.18, 1.12)
0.45 (0.18, 1.12)
0.50 (0.19, 1.28)
0.71 (0.37, 1.35)
0.75 (0.39, 1.46)
0.76 (0.39, 1.46)
1.05 (0.95, 1.17)
1.07 (0.94, 1.22)
1.10 (0.95, 1.28)
10
3
7
0.57 (0.24, 1.36)
0.66 (0.12, 3.55)
0.42 (0.12, 1.48)
0.45 (0.18, 1.12)
0.48 (0.18, 1.31)
0.33 (0.04, 2.94)
0.71 (0.37, 1.35)
0.47 (0.17, 1.30)
0.94 (0.41, 2.16)
1.05 (0.95, 1.17)
1.17 (0.86, 1.60)
1.01 (0.93, 1.11)
*P < 0.05.
TABLE 3. Results of sensitivity analyses (RR vs IRR)
Analysis using RR
Outcome
ESRF
Death
Amenorrhoea
Diarrhoea
Herpes zoster
Leucopenia
Pneumonia
RR (95% CI)
0.45
0.71
0.21
2.08
0.70
0.47
0.57
(0.18,
(0.37,
(0.09,
(0.98,
(0.39,
(0.27,
(0.24,
1.12)
1.35)
0.48)*
4.40)
1.25)
0.83)**
1.36)
Analysis using IRR
2
df
I
3
7
5
4
7
7
4
0.00
16.11
0.00
23.16
0.00
0.00
0.00
IRR (95% CI)
0.41
0.68
0.19
1.82
0.56
0.27
0.37
(0.16,
(0.35,
(0.16,
(0.65,
(0.32,
(0.16,
(0.11,
1.04)
1.31)
0.21)*
5.11)
1.01)
0.44)*
1.23
df
I2
3
7
5
4
7
7
4
0.00
18.26
0.00
44.92
29.40
17.78
66.46
*P < 0.0001; **P ¼ 0.009. I2: degree of heterogeneity.
TABLE 4. Meta-regression analysis of potential factors contributing to heterogeneity
of renal remission
Regression coefficient (S.E.)
Study characteristics
Sample size
Duration of study
IV CYC, %
Jadad score
Patient characteristics
Age
Male sex, %
Class IV nephritis, %
Duration of SLE/nephritis
Non-white or non-Asians, %
Z-score
P-value
0.0001
0.0007
–0.0009
0.0131
(0.001)
(0.004)
(0.001)
(0.073)
0.244
–0.177
–0.658
0.179
0.807
0.859
0.510
0.858
–0.0149
–0.0098
–0.0037
–0.0007
0.0046
(0.015
(0.016)
(0.004)
(0.004)
(0.002)
–1.025
–0.630
–0.993
–0.177
1.952
0.306
0.529
0.321
0.859
0.050
CYC in inducing disease remission and preventing renal relapses
in a number of subsequently published controlled trials [15–25].
Probably due to the small sample size and short duration of
observation of these trials, there is no total agreement as to
whether MMF is superior to CYC in terms of its efficacy and
effects on survival benefit. Three meta-analyses which attempted
to address these questions have been published so far [26–28].
However, data regarding the risk of death, ESRF, amenorrhoea,
leucopenia and infections between patients who received MMF
and CYC for treating proliferative lupus GN were inconsistent.
For example, Walsh et al. [26] and Moore et al. [28] respectively
demonstrated striking 80 and 56% risk reductions in death or
ESRF in patients who received MMF when compared with
those who had CYC. This alarmingly encouraging benefit, however, cannot be reproduced by Zhu et al. [27]. As far as side effects
are concerned, the significant reduction in leucopenia, infection
and amenorrhoea demonstrated by Moore et al. [28] could not be
repeated by Walsh et al. (besides infection) [26] and Zhu et al. [27].
These discrepancies largely reflect the small number of studies
included in these meta-analyses and lack of uniformity in pooling
data for analysis. In the current meta-analysis, apart from
including the two recently published RCTs, one of which,
the Aspreva Lupus Management Study (ALMS) is the latest
and largest one to date, we also implemented a few steps
of precautions with an aim to augment the robustness of the
pooled estimates. First, we realized that treatment duration
differed between studies when comparing MMF and CYC as
maintenance agents [15, 17, 20]. If only the RRs of certain
conditions were reported, a potential bias may arise because for
studies of longer duration, a higher proportion of patients would
experience certain events including death. To eliminate this bias,
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
Intravenous CYC vs oral CYC
All trials
Intravenous CYC only
Oral CYC only
Extension study
All trials
Excluding Chan [15]
Excluding Chan [15]
Quality
All trials
High ( 3)
Low (< 3)
No. of trials
Meta-analysis of MMF and CYC in lupus nephritis
CYC dosing regimens used in these RCTs were either based on
the protocol adopted by the US National Institutes of Health
(the NIH protocol) or daily oral CYC (2.5 mg/kg/day) during
the induction phase. These regimens incur a higher cumulative
CYC dose than that used in the Euro-Lupus Nephritis Trial
(ELNT), which comprises six intravenous fortnightly CYC
pulses of 500 mg [50]. With a comparable efficacy in terms of
renal response, the ELNT demonstrated that patients who were
exposed to lower CYC dose experienced half of the incidence
of severe infection than those who received higher CYC dose
(6 monthly intravenous CYC pulses initiating at 0.5 gm/m2, with
subsequent increment to the highest dose of 1500 mg/pulse) [50].
This finding poses an interesting question as to whether a lower
CYC dose or a shorter duration of CYC exposure can narrow the
side effect gap between MMF and CYC. Confirmation by further
controlled trials is necessary to provide the answer.
In conclusion, our meta-analysis, which is based on the largest
number of trials and patients to date, suggests that MMF is
comparable with CYC in terms of efficacy in inducing renal remissions, prevention of development of ESRF and survival benefit
in the treatment of patients with proliferation lupus GN.
Nevertheless, MMF is a safer medication for its lower incidence
of adverse effects including leucopenia and amenorrhoea. From
our results, we conclude that the treatment selection for proliferative lupus GN should be individualized based on patient-,
physician- and disease-related factors. Factors such as patients’
gender, age and ethnicity, presence of comorbidity, financial capability, physicians’ experience and patients’ concerns, especially
side effects and fertility issues are to be considered and discussed
with patients in order to provide them with the most appropriate
management strategy. Without scepticism, larger RCTs with
longer duration of observation and cost-effective analyses can
invariably shed more light on the ever existing question of whether
MMF is truly better than CYC in the treatment of proliferative
lupus GN.
Rheumatology key messages
MMF offers similar efficacy as CYC regarding renal remission,
patient and renal survival.
While being similarly efficacious, MMF possesses a more
favourable side effect profile than CYC.
There exists a differential response of lupus nephritis towards
immunosuppressive therapies between different ethnic groups.
Acknowledgements
The results presented in this article have not been published
previously in a full-length article, except that at the time of
submission, the data presented in this article were accepted for
presentation in abstract form at the Congress of the Asia Pacific
League of Associations for Rheumatology in Yokohama, Japan,
in September 2008, and the ACR/ARHP Annual Scientific
Meeting in San Francisco, USA, in October 2008.
Disclosure statement: C.S.L. was a member of the Aspreva Asia
Pacific Advisory Board between 2004 and 2006 and has received
an unrestricted grant from Aspreva for research. All other authors
have declared no conflicts of interest.
References
1 Boumpas DT, Sidiropoulos P, Bertsias G. Optimum therapeutic approaches for lupus
nephritis: what therapy and for whom? Nat Clin Pract Rheumatol 2005;1:22–30.
2 Mak A, Mok CC, Chu WP et al. Renal damage in systemic lupus erythematosus: a
comparative analysis of different age groups. Lupus 2007;16:28–34.
3 Mok CC, Wong RW, Lai KN. Treatment of severe proliferative lupus nephritis: the
current state. Ann Rheum Dis 2003;62:799–804.
4 Austin HA 3rd, Klippel JH, Balow JE et al. Therapy of lupus nephritis. Controlled trial
of prednisone and cytotoxic drugs. N Engl J Med 1986;314:614–9.
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
we converted all proportions into incidences (IRR) before pooling
the data and compared them with the RRs in the sensitivity analyses. Although the IRRs were roughly estimated from the mean
duration of treatment and observation in respective trials, the
results of the sensitivity analyses invariably confirmed the robustness of the pooled RRs. Secondly, response to either MMF or
CYC could not be assumed uniform between patients and trials
because patient- and trial-related factors may contribute to heterogeneity in treatment response. While being absent in the previous meta-analyses, we performed meta-regression to identify
factors which explained the heterogeneity of the effect sizes. Our
finding that the non-white and non-Asian race contributed to the
heterogeneity of renal remission further underscores the existence
of a differential response of nephritis towards immunosuppressive
therapies between different ethnicities. Indeed, it has been well
documented in a few observational studies that the white race is
favourable, whereas the African–American ethnicity is an unfavourable predictor for renal response and remission [44, 45]. The
ALMS, the only RCT which attempted to compare the renal
response with MMF amongst difference races and ethnicities,
showed that significantly more non-Caucasian and non-Asian
patients (composed mainly of black and Hispanic patients)
responded to MMF [46]. This is an interesting observation since
factors such as access to healthcare, poverty and non-compliance
are less influential in controlled trials, which are essentially protocol-based, and in addition, there were no clinically notable differences in the baseline disease-related characteristics amongst the
patients that might explain the response variation by race and
ethnicity. Therefore, difference in genetic makeup and immunopathology might play a more impactful role on the inter-ethnic
difference in treatment response of nephritis [47, 48]. While further
clinical and cost–effectiveness studies are required to address the
influence of race and ethnicity on renal response to MMF, based
on the current data, MMF is encouraged in non-Caucasian and
non-Asian patients with severe lupus nephritis, at least for induction therapy.
Despite our added precautions, there are limitations to this
study which mainly stems from factors intrinsic to systemic
reviews, missing data of some of the trials, poor trial quality
and the limitations of the original RCTs. Though statistically
significant results with regard to several side effects were obtained,
these should be interpreted with caution because the current
meta-analysis is still based only on a few controlled trials and is
subjected to random error. Analysable data with respect to side
effects were not available in the ALMS and one of the latest trials
published in abstract form [23, 24]. These may over- or underestimate the incidence of certain adverse effects of both MMF and
CYC. In addition, the poor quality of the trials mainly resulting
from the lack of double-blindness and double-dummy might
further compromise the validity of the results. In spite of this,
one point to note is that the quality of trials published in abstract
form might be underestimated because details regarding the
process of randomization and drop-outs are usually not mentioned due to a tight word limit. Finally, apart from the ALMS,
all trials were of small sample size and none of the studies were
double-blinded. Therefore, results of this albeit comprehensive
meta-analysis might not truly reflect the relative efficacy and
safety profile of MMF and CYC and we should therefore interpret
the results with caution. Nevertheless, every effort has been
implemented to eliminate potential bias and the sensitivity analysis helped prove the robustness of the pooled estimates. Lastly,
although we used meta-regression and were able to identify that
the non-white and non-Asian population significantly explained
the heterogeneity of the RR of renal remission, it was based on
a few studies and the results might be potentially biased.
Furthermore, the result was subject to aggregation bias because
the analysis was not based on subjects’ individual data [49].
Although MMF has a more favourable side effect profile than
CYC as shown in this meta-analysis, one point to note is that the
951
952
Anselm Mak et al.
27 Zhu B, Chen N, Lin Y et al. Mycophenolate mofetil in induction and maintenance
therapy of severe lupus nephritis: a meta-analysis of randomized controlled trials.
Nephrol Dial Transplant 2007;22:1933–42.
28 Moore RA, Derry S. Systematic review and meta-analysis of randomised trials
and cohort studies of mycophenolate mofetil in lupus nephritis. Arthritis Res Ther
2006;8:R182.
29 Jadad AR, Moore RA, Carroll D et al. Assessing the quality of reports of randomized
clinical trials: is blinding necessary? Control Clin Trials 1996;17:1–12.
30 Cochrane Handbook for Systematic Reviews of Interventions. http://www.cochranehandbook.org/ [18 March, 2009, date last accessed]
31 Hardy RJ, Thompson SG. Detecting and describing heterogeneity in meta-analysis.
Stat Med 1998;17:841–56.
32 Fletcher J. What is heterogeneity and is it important? Br Med J 2007;334:94–6.
33 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials
1986;7:177–88.
34 Thompson SG, Higgins JP. How should meta-regression analyses be undertaken
and interpreted? Stat Med 2002;21:1559–73.
35 Sterne JA, Gavaghan D, Egger M. Publication and related bias in meta-analysis:
power of statistical tests and prevalence in the literature. J Clin Epidemiol
2000;53:1119–29.
36 Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality
of reports of meta-analyses of randomised controlled trials: the QUOROM statement.
Quality of reporting of meta-analyses. Lancet 1999;354:1896–900.
37 Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull
1971;76:378–82.
38 Landis JR, Koch GG. The measurement of observer agreement for categorical data.
Biometrics 1977;33:159–74.
39 Glicklich D, Acharya A. Mycophenolate mofetil therapy for lupus nephritis refractory
to intravenous cyclophosphamide. Am J Kidney Dis 1998;32:318–22.
40 Dooley MA, Cosio FG, Nachman PH et al. Mycophenolate mofetil therapy in lupus
nephritis: clinical observations. J Am Soc Nephrol 1999;10:833–9.
41 Sesso R, Monteiro M, Sato E et al. A controlled trial of pulse cyclophosphamide
versus pulse methylprednisolone in severe lupus nephritis. Lupus 1994;3:107–12.
42 Belmont HM, Storch M, Buyon J et al. New York University/Hospital for Joint
Diseases experience with intravenous cyclophosphamide treatment: efficacy in
steroid unresponsive lupus nephritis. Lupus 1995;4:104–8.
43 Dooley MA, Ginzler EM. Newer therapeutic approaches for systemic lupus
erythematosus: immunosuppressive agents. Rheum Dis Clin North Am 2006;32:
91–102.
44 Korbet SM, Schwartz MM, Evans J. Lewis EJ; Collaborative Study Group. Severe
lupus nephritis: racial differences in presentation and outcome. J Am Soc Nephrol
2007;18:244–54.
45 Illei GG, Takada K, Parkin D et al. Renal flares are common in patients with severe
proliferative lupus nephritis treated with pulse immunosuppressive therapy: long-term
followup of a cohort of 145 patients participating in randomized controlled studies.
Arthritis Rheum 2002;46:995–1002.
46 Ginzler EM, Appel GB, Dooley MA et al. Mycophenolate mofetil and intravenous
cyclophosphamide in the Aspreva Lupus Management Study (ALMS): efficacy by
racial group. Arthritis Rheum 2007;56:4308.
47 Bouali H, Nietert P, Nowling TM et al. Association of the G-463A myeloperoxidase
gene polymorphism with renal disease in African Americans with systemic lupus
erythematosus. J Rheumatol 2007;34:2028–34.
48 Rovin BH, Lu L, Zhang X. A novel interleukin-8 polymorphism is associated with
severe systemic lupus erythematosus nephritis. Kidney Int 2002;62:261–5.
49 Morgenstern H. Uses of ecological analysis in epidemiologic research. Am J Public
Health 1982;72:127–30.
50 Houssiau FA, Vasconcelos C, D’Cruz D et al. Immunosuppressive therapy in lupus
nephritis: the Euro-Lupus Nephritis Trial, a randomized trial of low-dose versus
high-dose intravenous cyclophosphamide. Arthritis Rheum 2002;46:2121–31.
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on February 6, 2015
5 Buhaescu I, Covic A, Deray G. Treatment of proliferative lupus nephritis—a critical
approach. Semin Arthritis Rheum 2007;36:224–37.
6 Illei GG, Takada K, Parkin D et al. Renal flares are common in patients with
severe proliferative lupus nephritis treated with pulse immunosuppressive therapy:
long-term follow-up of a cohort of 145 patients participating in randomized controlled
studies. Arthritis Rheum 2002;46:995–1002.
7 Mok CC, Ying KY, Tang S et al. Predictors and outcome of renal flares after successful cyclophosphamide treatment for diffuse proliferative lupus glomerulo-nephritis.
Arthritis Rheum 2004;50:2559–68.
8 Valeri A, Radhakrishnan J, Estes D et al. Intravenous pulse cyclophosphamide treatment of severe lupus nephritis: a prospective five-year study. Clin Nephrol
1994;42:71–8.
9 Mok CC, Ho CTK, Siu YP et al. Treatment of diffuse proliferative lupus glomerulonephritis: a comparison of two cyclophosphamide-containing regimens. Am J Kidney
Dis 2001;38:256–64.
10 Mok CC, Mak A, To CH. Mycophenolate mofetil for lupus related myelopathy. Ann
Rheum Dis 2006;65:971–3.
11 Mak A, Mok CC. Mycophenolate mofetil for refractory haemolytic anemia in systemic
lupus erythematosus. Lupus 2005;14:856–8.
12 Moder KG, Amin S, Mazlumzadeh M et al. The effect of mycophenolate mofetil on
patients with active non-renal SLE. Clin Exp Rheumatol 2007;25:932.
13 Schanz S, Ulmer A, Rassner G et al. Successful treatment of subacute
cutaneous lupus erythematosus with mycophenolate mofetil. Br J Dermatol
2002;147:174–8.
14 Pisoni CN, Karim Y, Cuadrado MJ. Mycophenolate mofetil and systemic lupus
erythematosus: an overview. Lupus 2005;14(Suppl. 1):S9–11.
15 Chan TM, Li FK, Tang CS et al. Efficacy of mycophenolate mofetil in patients with
diffuse proliferative lupus nephritis. Hong Kong-Guangzhou Nephrology Study
Group. N Engl J Med 2000;343:1156–62.
16 Hu W, Liu Z, Chen H et al. Mycophenolate mofetil vs cyclophosphamide therapy for
patients with diffuse proliferative lupus nephritis. Chin Med J 2002;115:705–9.
17 Contreras G, Pardo V, Leclercq B et al. Sequential therapies for proliferative lupus
nephritis. N Engl J Med 2004;350:971–80.
18 Ong LM, Hooi LS, Lim TO et al. Randomized controlled trial of pulse intravenous
cyclophosphamide versus mycophenolate mofetil in the induction therapy of
proliferative lupus nephritis. Nephrology 2005;10:504–10.
19 Ginzler EM, Dooley MA, Aranow C et al. Mycophenolate mofetil or intravenous
cyclophosphamide for lupus nephritis. N Engl J Med 2005;353:2219–28.
20 Chan TM, Tse KC, Tang CS et al. Long-term study of mycophenolate mofetil as
continuous induction and maintenance treatment for diffuse proliferative lupus
nephritis. J Am Soc Nephrol 2005;16:1076–84.
21 Wang J, Hu W, Xie H et al. Induction therapies for class IV lupus nephritis with
non-inflammatory necrotizing vasculopathy: mycophenolate mofetil or intravenous
cyclophosphamide. Lupus 2007;16:707–12.
22 Flores-Suarez LF, Villa AR. Preliminary results of an open label randomized clinical
trial comparing mycophenolate mofetil (MMF) vs intravenous cyclophosphamide as
induction therapy for severe lupus nephritis. Arthritis Rheum 2004;50(Suppl.
1):S1029.
23 Isenberg D, Appel G, Dooley MA et al. Mycophenolate mofetil compared with
intravenous cyclophosphamide as induction for lupus nephritis. Ann Rheum Dis
2008;67(Suppl. II):S53.
24 Mulic-Bacic S, Antic D, Krizic M et al. Mycophenolate mofetil or intravenous
cyclophosphamide in treatment of lupus nephritis. Ann Rheum Dis 2008;67
(Suppl. II):S349.
25 Ye Z, Tan YH, Hong XP et al. MMF vs CTX in the treatment of severe SLE patients.
Lupus 2001;10(Suppl. 1):S99.
26 Walsh M, James M, Jayne D et al. Mycophenolate mofetil for induction therapy of
lupus nephritis: a systematic review and meta-analysis. Clin J Am Soc Nephrol
2007;2:968–75.