Abstract
Introduction
Randomized controlled trials (RCTs) are considered the gold standard for assessing treatment efficacy. However, sampling bias can affect the generalization of results to routine clinical practice. Here we assessed whether patients with lupus nephritis (LN) seen in routine clinical practice would have satisfied entry criteria to the major published RCTs in LN.
Methods
A systematic literature search from January 1974 to May 2015 was carried out, identifying all RCTs investigating LN induction treatment. Patients diagnosed with proliferative or membranous LN between 1995 and 2013 were identified from the Barts Lupus Centre database; baseline characteristics were compared with each RCT’s entry criteria to assess hypothetical inclusion or exclusion.
Results
Of 363 articles, 33 RCTs met inclusion criteria. Of 137 patients newly diagnosed with LN (111 with proliferative/mixed proliferative and 26 with pure membranous LN), 32% would have been excluded from RCT entry (range 8%–73%). The main reasons for exclusion would have been too severe disease, too mild disease, or prior immunosuppressant use, which were exclusion criteria in 26, 20, and 22 RCTs, respectively. A total of 27 patients with LN (20%) were re-biopsied due to flare; 68% of these would have been ineligible to enter RCTs.
Conclusion
Published RCTs do not truly reflect the heterogeneity of patients with LN in routine practice at our lupus center. The external validity of RCTs could be improved by including more representative patient cohorts. RCTs should be used as a guide but consideration should be given to similarities between individual patients and the characteristics of the trial cohorts before treatment decisions being made.
Keywords: glomerulonephritis, immunosuppression, lupus, randomized controlled trial
Tremendous advances in immunosuppressive treatment of lupus nephritis (LN) have been made over the past 4 decades. Fortunately, most treatment decisions made in routine clinical practice are now based on results from well-designed clinical trials rather than anecdotal evidence. Randomized controlled trials (RCTs) are considered the “gold standard” to assess treatment efficacy,1, 2 as they have the best potential to minimize bias through allocating participants to study arms in a random fashion. Despite their rigorous design, the applicability of results to individual patients has been debated thanks to other types of bias, some of which lead to RCTs not adequately reflecting the heterogeneity of patients in routine clinical practice.3 Examples include “selection bias” due to poor allocation concealment, “sampling bias” when physicians consider only a certain type of patient for trials and exclude patients on the basis of the lack of personal equipoise, or “literacy bias” for those who fail to understand the consent form.4
In LN RCTs, a particular problem may be “severity of illness bias.”4 To achieve a more homogeneous study population in whom the study drug is most likely to work and less likely to harm, patients must meet a range of prespecified entry requirements.5 Inclusion criteria to LN RCTs often specify the target range of biochemical variables and serological activity markers, and the histological evidence of LN with restriction to 1 or 2 particular classes of glomerulonephritis. Moreover, exclusion criteria intentionally eliminate patients with mild or, more commonly, severe disease; those with recent major infections or with preexisting comorbidities; and tend to prohibit certain immunosuppressive drugs before enrollment. As a consequence, patients recruited to most LN RCTs have a relatively moderate disease activity excluding those with too mild or severe disease and those developing active LN while on immunosuppressive therapies.
In this study, we aimed to evaluate the external validity of published LN RCTs. We assessed whether a multiethnic cohort of patients with lupus seen in routine clinical practice at a single center would have been represented adequately by assessing RCT inclusion and exclusion criteria.
Methods
Patients and Study Design
This study was conducted at Barts Health NHS Trust, the largest tertiary teaching hospital in the United Kingdom. Patients with biopsy-proven proliferative (class III or IV) or membranous (class V) LN diagnosed between 1995 and 2013 were included. LN classes based on glomerular pathology were defined according to the International Society of Nephrology/Renal Pathology Society 2003 classification.6, 7 All patients fulfilled at least 4 of the American College of Rheumatology revised classification criteria for systemic lupus erythematosus.8 Clinical data were collected retrospectively from electronic patient records; baseline variables included serum creatinine, glomerular filtration rate (GFR), serum albumin, proteinuria, double-stranded DNA, and complement C3 and C4 levels. Baseline immunosuppressive drugs at time of LN onset were recorded.
Identification of Eligible Clinical Studies
We selected RCTs that aimed to measure the efficacy of immunosuppressive agents in proliferative (class III, IV) and/or membranous (class V) LN induction treatment. Literature searches were performed using PubMed, Medline, and EMBASE databases (from 1974 to May 2015) and the Cochrane Controlled Trial Register (up to May 2015) using the OVID search engine, to identify relevant studies. Key search words included “lupus nephritis” and “randomized” and “lupus nephritis” and “clinical trial.” Additional studies were sought through bibliographic notations. Studies published in full-text English literature were included if they met the following criteria: (i) prospective controlled trial with treatment allocation by random assignment; (ii) compared at least 2 arms for LN induction treatment; (iii) involved proliferative and/or membranous LN patient population; and (iv) included adult patients. We excluded trials that published data on extended follow-up of already published RCTs, studies of maintenance therapies, or refractory LN.
Evaluation of Hypothetical Noninclusion Rate to RCTs
Baseline characteristics of patients in the Barts LN cohort were compared with patients participating in some of the most important RCTs of LN induction treatment from the past 2 decades.
For the purpose of this study, factors leading to exclusion included severe renal impairment with predefined GFR or serum creatinine in the selection criteria of the RCT; mild disease with predefined levels of proteinuria, serum albumin, GFR, or serum creatinine; or prohibited baseline immunosuppressive drugs used at time of LN onset. We did not analyze any other exclusion criteria, including confounding factors not related to their renal disease.
When trials involved patients with proliferative glomerulonephritis (class III or IV ± V), we evaluated only the patients with proliferative LN from our cohort (n = 111); where trials recruited patients with membranous LN, we evaluated only our membranous LN cohort (n = 26), whereas our entire cohort was compared (n = 137) when both classes were represented in the RCT.
Results
Characteristics of RCTs Included in the Analysis
Our systematic literature search yielded 363 articles, of which 309 articles were deemed unsuitable after title or abstract review (Figure 1). Of the remaining, 5 trials on LN induction treatment had to be excluded, as the study design was insufficiently described and no comment was made on inclusion and exclusion criteria used. The final analysis composed of 33 RCTs is shown in Table 1.9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41
Figure 1.
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram.
Table 1.
An overview of 33 randomized controlled trials of lupus nephritis induction treatments published since 1970
| Author, Year | Study design | Follow-up | n | LN classes | Treatment arms |
|---|---|---|---|---|---|
| Steinberg AD et al.9 1971 | Double-blind single center (USA) |
10 wk | 13 | Biopsy not required | G1: p.o. CYC G2: placebo |
| Steinberg and Decker10 1974 | Double-blind single center (USA) |
10 wk | 38 | Diffuse proliferative (IV) | G1: p.o. CYC G2: AZA G3: placebo |
| Ginzler E et al.11 1976 | Double-blind single center (USA) |
12 mo | 14 | Diffuse proliferative (IV) or membranous (V) | G1: AZA G2: p.o. CYC + AZA |
| Donadio JV et al.12 1978 | Open label single center (USA) |
Mean 43 mo | 50 | Diffuse proliferative (IV) | G1: p.o. CYC + p.o. corticosteroid G2: p.o. corticosteroid |
| Dinant H et al.13 1982 | Open label single center (USA) |
Mean 42 mo | 46 | Biopsy not required | G1: p.o. corticosteroid G2: p.o. CYC + AZA G3: i.v. CYC |
| Boumpas DT et al.14 1992 | Open label single center (USA) |
30 mo | 65 | Proliferative (III or IV) or membranous (V) | G1: i.v. corticosteroid G2: short course i.v. CYC G3: long course i.v. CYC |
| Lewis EJ et al.15 1992 | Open label multicenter (USA) |
Mean 136 wk | 86 | Proliferative (III or IV) | G1: p.o. corticosteroid + p.o. CYC G2: p.o. corticosteroid + p.o. CYC + PE |
| Doria A et al.16 1994 | Open label single center (Italy) |
Mean 23 mo | 18 | Diffuse proliferative (IV) | G1: AZA G2: AZA + PE |
| Sesso R et al.17 1994 | Open label single center (Brazil) |
Mean 15 mo | 29 | Biopsy not required | G1: i.v. CYC G2: i.v. corticosteroid |
| Gouerly MF et al.18 1996 | Open label single center (USA) |
Minimum 5 yr | 82 | Proliferative (III or IV) | G1: i.v. CYC G2: i.v. corticosteroid G3: i.v. CYC + i.v. corticosteroid |
| Wallace DJ et al.19 1998 | Open label international |
24 mo | 19 | Proliferative (III or IV) | G1: i.v. CYC G2: i.v. CYC + PE |
| Chan TM et al.20 2000 | Open label single center (Hong Kong) |
12 mo | 42 | Diffuse proliferative (IV) | G1: MMF G2: p.o. CYC |
| Houssiau FA et al.21 2002 | Open label international (Europe) |
Median 41 mo | 90 | Proliferative (III or IV) | G1: high dose i.v. CYC G2: low dose i.v. CYC |
| Yee CS et al.22 2004 | Open label international (Europe) |
Mean 3.5 yr | 32 | Proliferative (III or IV) | G1: i.v. CYC G2: p.o. CYC |
| Ong LM et al.23 2005 | Open label national (Malaysia) |
6 mo | 54 | Proliferative (III or IV) | G1: MMF G2: i.v. CYC |
| Ginzler EM et al.24 2005 | Open label national (USA) |
24 wk | 140 | Proliferative (III or IV) or membranous (V) | G1: MMF G2: i.v. CYC |
| Grootscholten C et al.25 2006 | Open label national (Netherlands) |
Median 5.7 yr | 87 | Proliferative (III or IV) | G1: i.v. CYC G2: AZA + i.v. corticosteroid |
| Bao H et al.26 2008 | Open label single center (China) |
9 mo | 40 | Diffuse proliferative and membranous (IV + V) | G1: MMF + Tac G2: i.v. CYC |
| Appel GB et al.27 2009 | Open label international |
24 wk | 370 | Proliferative (III or IV) or membranous (V) | G1: MMF G2: i.v. CYC |
| Austin HA et al.28 2009 | Open label single center (USA) |
12 mo | 42 | Membranous (V) | G1: p.o. corticosteroid G2: i.v. CYC G3: CsA |
| Li EK et al.29 2009 | Open label single center (Hong Kong) |
48 wk | 19 | Proliferative (III or IV) | G1: RTX G2: RTX + i.v. CYC |
| Sabry A et al.30 2009 | Open label single center (Egypt) |
12 mo | 46 | Diffuse proliferative (IV) | G1: high dose i.v. CYC G2: low dose i.v. CYC |
| El-Shafey EM et al.31 2010 | Open label single center (Egypt) |
24 wk | 47 | Proliferative (III or IV) | G1: MMF G2: i.v. CYC |
| Zavada J et al.32 2010 | Open label, international (Czech Republic, Slovakia) |
18 mo | 40 | Proliferative (III or IV) | G1: i.v. CYC G2: CsA |
| Chen W et al.33 2011 | Open label national (China) |
6 mo | 81 | Proliferative (III or IV) or membranous (V) | G1: Tac G2: i.v. CYC |
| Li X et al.34 2011 | Open label single center (China) |
24 wk | 62 | Proliferative (III or IV) or membranous (V) | G1: MMF G2: Tac G3: i.v. CYC |
| Rovin BH et al.35 2012 | Double-blind, international (USA, Latin-America) |
52 wk | 144 | Proliferative (III or IV) | G1: MMF G2: MMF + RTX |
| Yap DY et al.36 2012 | Open label national (China) |
24 mo | 16 | Membranous (V) | G1: MMF G2: Tac |
| Mysler EF et al.37 2013 | Double-blind international |
48 wk | 381 | Proliferative (III or IV) | G1: MMF or i.v. CYC G2: low dose OCR + MMF or i.v. CYC G3: high dose OCR + MMF or i.v. CYC |
| The ACCESS Trial group38 2014 | Double-blind international (USA, Mexico) |
52 wk | 137 | Proliferative (III or IV) | G1: ABT + i.v. CYC G2: i.v. CYC |
| Furie R et al.39 2014 | Double-blind, international |
52 wk | 300 | Proliferative (III or IV) | G1: MMF G2: low dose ABT + MMF G3: high dose ABT + MMF |
| Liu Z et al.40 2014 | Open label national (China) |
24 wk | 544 | Proliferative (III or IV) or membranous (V) | G1: Tac + MMF G2: i.v. CYC |
| Mok CC et al.41 2014 | Open label national (Hong Kong) |
6 mo | 150 | Proliferative (III or IV) or membranous (V) | G1: MMF G2: Tac |
ABT, abatacept; AZA, azathioprine; CsA, cyclosporin; CYC, cyclophosphamide; G, group; LN, lupus nephritis; MMF, mycophenolate mofetil; n, number of patients; OCR, ocrelizumab; PE, plasmapheresis; RTX, rituximab; Tac, Tacrolimus.
Of the 33 RCTs, 19 journal articles (58%) failed to report the methods of randomization and/or the protocols used to assign participants to comparison groups. Furthermore, only 4 RCTs27, 31, 39, 40 published the rate of screening failure, ranging from 19.6% to 39.4%. There were only 2 of 33 RCTs elaborating the reasons for ineligibility.27, 31
The most frequently prespecified inclusion and exclusion criteria encountered in the selected RCTs are detailed in Table 2. Nearly all RCTs (97%) required fulfillment of systemic lupus erythematosus classification criteria, but a positive antinuclear antibody/lupus erythematosus test was rarely desired (18%). Three trials (9%) did not require biopsy evidence of LN; 75% of RCTs required patients to have a renal function that met predefined values, and 67% of studies demanded a minimal level proteinuria. Prohibited concomitant or prior immunosuppressive treatment was defined in 76% of RCTs.
Table 2.
Commonly used eligibility and exclusion criteria of the 33 RCTs
| Eligibility criteria | n (%) |
|---|---|
| Fulfillment of systemic lupus erythematosus criteria according to ARA or American College of Rheumatology | 32 (97.0) |
| Predefined serum creatinine or glomerular filtration rate | 25 (75.8) |
| Predefined minimal proteinuria | 22 (66.7) |
| Age limitation | 19 (57.6) |
| Active urinary sediment | 12 (36.4) |
| Predefined histopathological changes (other than class) | 10 (30.3) |
| Positive antinuclear antibody/lupus erythematosus test | 6 (18.2) |
| Abnormal lupus activity markers, double-stranded DNA, complements | 4 (12.1) |
| Exclusion criteria | n (%) |
|---|---|
| Prohibited immunosuppressive medication | 25 (75.8) |
| Pregnancy | 24 (72.7) |
| Major infection | 22 (66.7) |
| Malignancy | 11 (33.3) |
| Cytopenia | 11 (33.3) |
| Hypersensitivity to a study drug | 10 (30.3) |
| Diabetes | 10 (30.3) |
| Gastrointestinal hemorrhage, peptic ulcer disease | 8 (24.2) |
| Liver disease | 7 (21.2) |
| Cerebral lupus | 6 (18.2) |
| Nonlupus renal disease | 5 (15.2) |
| Unwilling to use contraception | 5 (15.2) |
| Anticipation of poor compliance | 4 (12.1) |
ARA, American Rheumatism Association; RCT, randomized controlled trial.
Characteristics of Barts LN Patient Cohort
From our database, we identified 137 patients with a new diagnosis of biopsy-proven active LN; their baseline characteristics are shown in Table 3. Eighty-one percent of our LN cohort had proliferative (n = 111, 30% class III, 51% class IV, with or without membranous component), and 19% had pure membranous LN (n = 26). Overall, patients with LN in our cohort tended to have more severe renal disease compared with patients in landmark RCTs of LN induction (Table 4)42; they had a mean proteinuria of 5.5 g/24 h (± 5.1 SD), mean serum albumin of 27.3 g/l (±7.4 SD), and mean serum creatinine of 159 μmol/l (± 166 SD). Although 33% (n = 45) had a normal GFR (>90 ml/min per 1.73 m2), 18% (n = 25) had a GFR <30 ml/min per 1.73 m2. Serological evidence of lupus activity was present in half of the cohort with strongly positive double-stranded DNA and low complements. Twenty-three patients with systemic lupus erythematosus (16.7%) were taking immunosuppressants other than hydroxychloroquine or oral prednisolone (either mycophenolate mofetil or azathioprine) at time of LN diagnosis.
Table 3.
Baseline characteristics of our LN cohort
| Characteristic | Barts LN cohort, n = 137 |
|---|---|
| Gender, n (%) | |
| Male | 20 (15.6) |
| Female | 117 (85.4) |
| Ethnicity, n (%) | |
| White Caucasian | 22 (16.1) |
| Black | 69 (50.4) |
| Asiana | 46 (33.5) |
| Age, yr, mean (SD) | |
| at SLE diagnosis | 30.8 (11.3) |
| at LN diagnosis | 33.6 (11.5) |
| Serology, n (%) | |
| ANA positive | 131 (97.0) |
| ENA antibody positive | 97 (75.8) |
| RNP antibody positive | 57 (44.5) |
| Sm antibody positive | 40 (31.3) |
| Ro antibody positive | 54 (42.2) |
| aPL antibodies | 30 (25.4) |
| Renal histology class, n (%) | |
| Class III LN (± class V) | 41 (30.0) |
| Class IV LN (± class V) | 70 (51.1) |
| Class V LN | 26 (19.0) |
| Biochemical variables, mean ± SD | |
| Proteinuria (g/24 h) | 5.5 ± 5.1 |
| Serum albumin (g/l) | 27.3 ± 7.4 |
| Serum creatinine (μmol/l) | 159.2 ± 166.1 |
| Range of GFR (ml/min per 1.73 m2), n (%) | |
| ≥90 | 45 (32.8) |
| ≥60 to <90 | 31 (22.6) |
| ≥30 to <60 | 36 (26.3) |
| ≥15 to <30 | 11 (8.0) |
| <15 | 14 (10.2) |
| Lupus activity markers, n (%) | |
| Range of anti-dsDNA titer, IU/ml | |
| <30 (negative) | 37 (30.1) |
| 30 to 60 (low positive) | 14 (11.4) |
| 60 to 200 (positive) | 15 (12.2) |
| >200 (strong positive) | 57 (46.3) |
| Low complement C3 | 78 (68.4) |
| Low complement C4 | 82 (69.5) |
ANA, antinuclear antibody; aPL, antiphospholipid; dsDNA, double-stranded DNA; ENA, extractable nuclear antigens; GFR, glomerular filtration rate; LN, lupus nephritis; n, number of patients; RNP, ribonucleoprotein; SLE, systemic lupus erythematosus; Sm, Smith.
South Asians (n = 39) and Oriental Asians (n = 7).
Table 4.
Overview of patient cohorts in important RCTs of LN induction
| Characteristic | Barts cohort | Gouerly MF et al.18 1996 | Chan TM et al.20 2000 | Houssiau FA et al.21 2002 | Ginzler EM et al.24 2005 | Appel GB et al.27 2009 | Rovin BH et al.35 2012 | Furie et al.39 2014 |
|---|---|---|---|---|---|---|---|---|
| Gender, n (%) | ||||||||
| Male | 20 (15.6) | 14 (16.7) | 3 (7.1) | 6 (7.0) | 14 (10.0) | 57 (15.4) | 14 (9.7) | 47 (15.8) |
| Female | 117 (85.4) | 68 (82.9) | 39 (92.9) | 84 (93.0) | 126 (90.0) | 313 (84.6) | 130 (90.3) | 251 (84.2) |
| Ethnicity, n (%) | ||||||||
| White Caucasian | 22 (16.1) | 57 (69.5) | 0 | 76 (84.4) | 24 (17.1) | 147 (39.7) | 45 (31.3) | 111 (37.2) |
| Black | 69 (50.4) | 18 (22.0) | 0 | 8 (8.9) | 79 (56.4) | 46 (12.4) | 40 (27.8) | 14 (4.7) |
| Asian | 46 (33.5) | 2 (2.4) | 42 (100) | 6 (7) | 8 (5.7) | 123 (33.2) | 7 (4.9) | 164 (55.0) |
| Renal histology class, n (%) | ||||||||
| Class III LN (± V) | 41 (30.0) | 62 (78.5) | 0 | 21 (23.3) | NK | 58 (15.7) | 49 (34.0) | 78 (26.2) |
| Class IV LN (± V) | 70 (51.1) | 17 (21.5) | 42 (100) | 62 (68.9) | NK | 252 (68.1) | 95 (66.0) | 220 (73.8) |
| Class V LN | 26 (19.0) | 0 | 0 | 7 (7.8) | 39 (27.9) | 60 (16.2) | 0 | 0 |
| Biochemical variables, mean ± SD | ||||||||
| Proteinuria (g/24 h) | 5.5 ± 5.1 | 4.0 ± 0.7a | 4.8 ± 3.2b | 3.04 ± 2.4 | 4.2 ± 2.4b | 4.1 ± 3.7 | 4.0 ± 2.9 | 4.4 ± 5.5b |
| Serum albumin (g/l) | 27.3 ± 7.4 | NK | 28.0 ± 5.5b | 30.3 ± 6.1 | 27.5 ± 7.6b | NK | 26.5 ± 7.5b | NK |
| Serum creatinine (μmol/l) | 159.2 ± 166.1 | 102.0 ± 9.9a | 106.1 ± 39.8b | 101.7 ± 58.3 | 94.6 ± 45.1b | 101.0 ± 1.1 | 88.4 ± 44.2 | 73.6 (35–239)c |
LN, lupus nephritis; n, number of patients; NK, not known.
Combined mean ± SE.
Combined mean ± SD.
Combined median (range).
Hypothetical Noninclusion Rate of Newly Diagnosed Patients With LN to LN Induction RCTs
On average, one-third (31.8%) of our newly diagnosed biopsy-proven active LN cohort would have been ineligible to enter the RCTs (range 8.1%–72.9%) (Figure 2). These patients would have been excluded because of factors directly related to their renal disease or the concomitant use of immunosuppressive treatment. Severe renal impairment was a prespecified exclusion criterion in 26 RCTs (79%) leading to ineligibility in up to 61.4% of our LN cohort. Twenty RCTs (63%) would have left out patients with milder disease based on mildly impaired renal function or lower level of proteinuria, resulting in a hypothetical exclusion rate reaching up to 44.1%. Additionally, 22 RCTs (67%) prohibited the use of various immunosuppressive drugs resulting in exclusion rates of 16.1% of our LN cohort.
Figure 2.
Noninclusion rates to landmark lupus nephritis (LN) randomized controlled trials (RCTs). The proportion of our patients with LN that would not have been included in RCTs is shown. Overall on average 32% of our patients with LN would not have been eligible to participate in the RCTs listed (range 8%–73%).
Hypothetical Noninclusion Rate of Relapsed Patients With LN to LN Induction RCTs
Next, we analyzed hypothetical exclusion rates of relapsed patients with LN with active LN evidenced by a repeat kidney biopsy. Twenty-nine of 137 patients with LN (21.2%) had a repeat kidney biopsy. The median time to repeat biopsy was 28 months (interquartile range 13–47 months). Of those, 27 had active LN, 24 proliferative (class III or IV with or without class V, 88.9%), and 3 pure membranous (class V, 11.1%). Two patients had end-stage LN (class VI, 7.4%) and because indication for repeat biopsy was persistent disease activity rather than flare, they therefore were excluded from the following analysis.
We found that due to strict trial designs, on average, 67.7% of patients with LN with renal flares would have been ineligible to participate in these RCTs (range 7.4%–100%). Severe disease excluded up to 63.2% of the repeat biopsy cohort (average 19.2%), and mild disease caused ineligibility rates up to 51.9% (average 12.9%). Overall, 57.5% of patients would have been prevented from entering the 33 RCTs because of prior use of immunosuppressive agents, with 11 RCTs (33.3%) excluding more than 80% of our cohort.
Discussion
Current standard of care in LN induction remains suboptimal and kidney damage occurs despite using the best available therapies. The rationale behind choosing one immunosuppressive drug over another or in combination is based on knowledge gained from trials conducted in the past 5 decades. Among those, explanatory RCTs are most valuable, but can be susceptible to bias, using narrow inclusion and exclusion criteria to recruit those most likely to benefit from a drug and least likely to experience harm.4, 5
RCTs must not only be internally valid but clinically useful and relevant to lupus patient populations. The present study is the first to evaluate eligibility of patients with LN seen in everyday practice for inclusion into published RCTs. Our results indicate that LN RCT populations may only partially reflect real-world LN cohorts and caution should be exercised in extrapolating trial data to patient subgroups that have not been adequately represented in studies. Overall, 32% of our LN cohort would have been excluded from the reviewed 33 RCTs because of the severity of their renal disease and/or the use of an immunosuppressive drug at the time of diagnosis. Eligibility criteria and hence exclusion rates showed a wide range of variability among the examined trials, and reached up to 61% due to severe disease (low GFR or high serum creatinine), up to 44% due to mild disease (low-level proteinuria or preserved renal function), and up to 16% due to concomitant immunosuppressive drugs at time of LN onset. These findings suggest that RCT results are less likely to be relevant to patients with severe LN who would have never made it to the trial. Exclusion rates were even higher, on average 68%, when examining our existing LN population, with new renal flares having had a repeat kidney biopsy proving active LN. This further queries the application of RCT results to patients with severe LN and those who are nonresponders to other immunosuppressive treatments.
In this study, we focused only on exclusion criteria directly related to LN: disease severity and immunosuppressive status at time of LN diagnosis. There are other factors listed in trial designs that would have further increased exclusion rates if examined. The most important ones are previous immunosuppression, pregnancy, and medical conditions associated with safety of study subjects: infections, malignancy, and comorbidities, such as cytopenia, diabetes, peptic ulcer disease, and cerebral lupus. Some of these criteria further weaken the representation of a real-life LN population; nevertheless, there is no solution that could address the problem without jeopardizing patient safety. However, implementing less conservative exclusion criteria, particularly with regard to comorbidities, in late-stage clinical trials might improve the generalizability.
Stringent eligibility criteria are likely to limit the external validity of RCTs; nevertheless, clinicians should be able to use their judgment when interpreting the results of published trials. Having said that, inadequate reporting of trials can often mislead decision making in prescribing practice. We found a large variation in the quality of reporting of LN RCTs. None of the RCTs published the “screening rate,” the number of patients who were diagnosed with LN and assessed for eligibility using inclusion/exclusion criteria. What is more disappointing is that only 4 of the 33 RCTs published “ineligibility rates,” and only 2 published the factors leading to nonparticipation. In one of the most cited and influential LN articles, in the Aspreva Lupus Management Study (ALMS) cohort, 13% of screened patients with LN did not meet entry criteria due to mild clinical activity or prohibited concurrent medications.30 In our cohort, the hypothetical rate of exclusion from the ALMS trial for these reasons would have reached 29%, despite that our LN cohort seemed to have a more severe disease compared with ALMS study subjects. An explanation for the discrepancy could be the ethnic background of the participating subjects, with a higher proportion of white patients (40%) in the ALMS versus our cohort (16%), and the low proportion of black patients (12% vs. 50%, ALMS vs. our LN cohort, respectively) who generally tend to present with more severe LN. Both gender and ethnicity are factors affecting the incidence and severity of LN, as well as treatment outcomes. There is evidence that Hispanic and African-American individuals have a higher incidence and prevalence rates of LN and they may also show different response to treatment.43, 44 Clinical studies should account for these ethnic disparities, as indicated by the National Institutes of Health Revitalization Act of 1993.45 Some issues could certainly be improved if demographic profiles would be incorporated in trial designs, and medical systems were able to address poverty, lack of education, impaired access to trials, and language barrier. Ultimately, more diverse recruitment strategies should be used when enrolling minority groups to achieve a more representative patient population, which would lead to a better generalizability of the results.
The lack of transparency due to poor reporting rate can eventually cause serious damage; by preventing readers from evaluation of the validity and reliability of these trials, which may result in biased estimates of treatment effects. The CONSORT 2010 statement (Consolidated Standards of Reporting Trials) gives a clear guidance on the reporting of RCTs using an essential checklist.46 The checklist includes the reporting of exclusion rates. Since this CONSORT was published, sadly, only 1 journal article included in this analysis followed these recommendations. We recommend that editors improve the reporting quality of clinical trials for transparency by giving authors clear guidance on items to be reported.
In summary, the present study was the first to evaluate how representative participants in published LN RCTs were in comparison with patients seen in usual clinical practice from a large tertiary lupus center. A third of newly diagnosed real-world patients with LN would have been excluded from LN RCTs due to strict eligibility criteria solely related to the severity of their renal disease. Although some trials have excellent validity, many do not; and factors determining their validity are rarely reported, preventing the clinician from using thoughtful judgment. Specifically, we highlight the need for more pragmatic trials designed for those with more severe LN.
Disclosure
All the authors declared no competing interests.
References
- 1.Riegelman R.K. 4th ed. Lippincott Williams & Wilkins; Philadelphia, PA: 2000. Studying a Study and Testing a Test: How to Read the Medical Evidence. [Google Scholar]
- 2.Nallamothu B.K., Hayward R.A., Bates E.R. Beyond the randomized clinical trial: the role of effectiveness studies in evaluating cardiovascular therapies. Circulation. 2008;118:1294–1303. doi: 10.1161/CIRCULATIONAHA.107.703579. [DOI] [PubMed] [Google Scholar]
- 3.Kennedy-Martin T., Curtis S., Faries D. A literature review on the representativeness of randomized controlled trial samples and implications for the external validity of trial results. Trials. 2015;16:495. doi: 10.1186/s13063-015-1023-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jadad A.R., Enkin M., Jadad A.R. 2nd ed. Blackwell Pub; Malden, MA: 2007. Randomized Controlled Trials:Questions, Answers, and Musings. [Google Scholar]
- 5.Roland M., Torgerson D.J. What are pragmatic trials? BMJ. 1998;316:285. doi: 10.1136/bmj.316.7127.285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Churg J., Sobin L.H. Renal Disease: Classification and Atlas of Glomerular Disease. Tokyo, Japan: Igaku-Shoin. 1982 [Google Scholar]
- 7.Weening J.J., D'Agati V.D., Schwartz M.M. The classification of glomerulonephritis in systemic lupus erythematosus revisited. Kidney Int. 2004;65:521–530. doi: 10.1111/j.1523-1755.2004.00443.x. [DOI] [PubMed] [Google Scholar]
- 8.Tan E.M., Cohen A.S., Fries J.F. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum. 1982;25:1271–1277. doi: 10.1002/art.1780251101. [DOI] [PubMed] [Google Scholar]
- 9.Steinberg A.D., Kaltreider H.B., Staples P.J. Cyclophosphamide in lupus nephritis:a controlled trial. Ann Intern Med. 1971;75:165–171. doi: 10.7326/0003-4819-75-2-165. [DOI] [PubMed] [Google Scholar]
- 10.Steinberg A.D., Decker J.L. A double-blind controlled trial comparing cyclophosphamide, azathioprine and placebo in the treatment of lupus glomerulonephritis. Arthritis Rheum. 1974;17:923–937. doi: 10.1002/art.1780170602. [DOI] [PubMed] [Google Scholar]
- 11.Ginzler E., Diamond H., Guttadauria M. Prednisone and azathioprine compared to prednisone plus low-dose azathioprine and cyclophosphamide in the treatment of diffuse lupus nephritis. Arthritis Rheum. 1976;19:693–699. doi: 10.1002/1529-0131(197607/08)19:4<693::aid-art1780190406>3.0.co;2-2. [DOI] [PubMed] [Google Scholar]
- 12.Donadio J.V., Jr., Holley K.E., Ferguson R.H. Treatment of diffuse proliferative lupus nephritis with prednisone and combined prednisone and cyclophosphamide. N Engl J Med. 1978;299:1151–1155. doi: 10.1056/NEJM197811232992102. [DOI] [PubMed] [Google Scholar]
- 13.Dinant H.J., Decker J.L., Klippel J.H. Alternative modes of cyclophosphamide and azathioprine therapy in lupus nephritis. Ann Intern Med. 1982;96:728–736. doi: 10.7326/0003-4819-96-6-728. [DOI] [PubMed] [Google Scholar]
- 14.Boumpas D.T., Austin H.A., 3rd, Vaughn E.M. Controlled trial of pulse methylprednisolone versus two regimens of pulse cyclophosphamide in severe lupus nephritis. Lancet. 1992;340:741–745. doi: 10.1016/0140-6736(92)92292-n. [DOI] [PubMed] [Google Scholar]
- 15.Lewis E.J., Hunsicker L.G., Lan S.P. A controlled trial of plasmapheresis therapy in severe lupus nephritis. The Lupus Nephritis Collaborative Study Group. N Engl J Med. 1992;326:1373–1379. doi: 10.1056/NEJM199205213262101. [DOI] [PubMed] [Google Scholar]
- 16.Doria A., Piccoli A., Vesco P. Therapy of lupus nephritis. A two-year prospective study. Ann Med Intern (Paris) 1994;145:307–311. [PubMed] [Google Scholar]
- 17.Sesso R., Monteiro M., Sato E. A controlled trial of pulse cyclophosphamide versus pulse methylprednisolone in severe lupus nephritis. Lupus. 1994;3:107–112. doi: 10.1177/096120339400300209. [DOI] [PubMed] [Google Scholar]
- 18.Gourley M.F., Austin H.A., 3rd, Scott D. Methylprednisolone and cyclophosphamide, alone or in combination, in patients with lupus nephritis. A randomized, controlled trial. Ann Intern Med. 1996;125:549–557. doi: 10.7326/0003-4819-125-7-199610010-00003. [DOI] [PubMed] [Google Scholar]
- 19.Wallace D.J., Goldfinger D., Pepkowitz S.H. Randomized controlled trial of pulse/synchronization cyclophosphamide/apheresis for proliferative lupus nephritis. J Clin Apher. 1998;13:163–166. doi: 10.1002/(sici)1098-1101(1998)13:4<163::aid-jca4>3.0.co;2-1. [DOI] [PubMed] [Google Scholar]
- 20.Chan T.M., Li F.K., Tang C.S. Efficacy of mycophenolate mofetil in patients with diffuse proliferative lupus nephritis. Hong Kong-Guangzhou Nephrology Study Group. N Engl J Med. 2000;343:1156–1162. doi: 10.1056/NEJM200010193431604. [DOI] [PubMed] [Google Scholar]
- 21.Houssiau F.A., Vasconcelos C., D'Cruz D. Immunosuppressive therapy in lupus nephritis: the Euro-Lupus Nephritis Trial, a randomized trial of low-dose versus high-dose intravenous cyclophosphamide. Arthritis Rheum. 2002;46:2121–2131. doi: 10.1002/art.10461. [DOI] [PubMed] [Google Scholar]
- 22.Yee C.S., Gordon C., Dostal C. EULAR randomised controlled trial of pulse cyclophosphamide and methylprednisolone versus continuous cyclophosphamide and prednisolone followed by azathioprine and prednisolone in lupus nephritis. Ann Rheum Dis. 2004;63:525–529. doi: 10.1136/ard.2002.003574. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ong L.M., Hooi L.S., Lim T.O. Randomized controlled trial of pulse intravenous cyclophosphamide versus mycophenolate mofetil in the induction therapy of proliferative lupus nephritis. Nephrology. 2005;10:504–510. doi: 10.1111/j.1440-1797.2005.00444.x. [DOI] [PubMed] [Google Scholar]
- 24.Ginzler E.M., Dooley M.A., Aranow C. Mycophenolate mofetil or intravenous cyclophosphamide for lupus nephritis. N Engl J Med. 2005;353:2219–2228. doi: 10.1056/NEJMoa043731. [DOI] [PubMed] [Google Scholar]
- 25.Grootscholten C., Ligtenberg G., Hagen E.C. Azathioprine/methylprednisolone versus cyclophosphamide in proliferative lupus nephritis. A randomized controlled trial. Kidney Int. 2006;70:732–742. doi: 10.1038/sj.ki.5001630. [DOI] [PubMed] [Google Scholar]
- 26.Bao H., Liu Z.H., Xie H.L. Successful treatment of class V+IV lupus nephritis with multitarget therapy. J Am Soc Nephrol. 2008;19:2001–2010. doi: 10.1681/ASN.2007121272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Appel G.B., Contreras G., Dooley M.A. Mycophenolate mofetil versus cyclophosphamide for induction treatment of lupus nephritis. J Am Soc Nephrol. 2009;20:1103–1112. doi: 10.1681/ASN.2008101028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Austin H.A., 3rd, Illei G.G., Braun M.J. Randomized, controlled trial of prednisone, cyclophosphamide, and cyclosporine in lupus membranous nephropathy. J Am Soc Nephrol. 2009;20:901–911. doi: 10.1681/ASN.2008060665. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Li E.K., Tam L.S., Zhu T.Y. Is combination rituximab with cyclophosphamide better than rituximab alone in the treatment of lupus nephritis? Rheumatology. 2009;48:892–898. doi: 10.1093/rheumatology/kep124. [DOI] [PubMed] [Google Scholar]
- 30.Sabry A., Abo-Zenah H., Medhat T. A comparative study of two intensified pulse cyclophosphamide remission-inducing regimens for diffuse proliferative lupus nephritis:an Egyptian experience. Int Urol Nephrol. 2009;41:153–161. doi: 10.1007/s11255-007-9325-4. [DOI] [PubMed] [Google Scholar]
- 31.El-Shafey E.M., Abdou S.H., Shareef M.M. Is mycophenolate mofetil superior to pulse intravenous cyclophosphamide for induction therapy of proliferative lupus nephritis in Egyptian patients? Clin Exp Nephrol. 2010;14:214–221. doi: 10.1007/s10157-010-0270-7. [DOI] [PubMed] [Google Scholar]
- 32.Zavada J., Pesickova S., Rysava R. Cyclosporine A or intravenous cyclophosphamide for lupus nephritis: the Cyclofa-Lune study. Lupus. 2010;19:1281–1289. doi: 10.1177/0961203310371155. [DOI] [PubMed] [Google Scholar]
- 33.Chen W., Tang X., Liu Q. Short-term outcomes of induction therapy with tacrolimus versus cyclophosphamide for active lupus nephritis: a multicenter randomized clinical trial. Am J Kidney Dis. 2011;57:235–244. doi: 10.1053/j.ajkd.2010.08.036. [DOI] [PubMed] [Google Scholar]
- 34.Li X., Ren H., Zhang Q. Mycophenolate mofetil or tacrolimus compared with intravenous cyclophosphamide in the induction treatment for active lupus nephritis. Nephrol Dial Transplant. 2012;27:1467–1472. doi: 10.1093/ndt/gfr484. [DOI] [PubMed] [Google Scholar]
- 35.Rovin B.H., Furie R., Latinis K. Efficacy and safety of rituximab in patients with active proliferative lupus nephritis:the Lupus Nephritis Assessment with Rituximab study. Arthritis Rheum. 2012;64:1215–1226. doi: 10.1002/art.34359. [DOI] [PubMed] [Google Scholar]
- 36.Yap D.Y., Yu X., Chen X.M. Pilot 24 month study to compare mycophenolate mofetil and tacrolimus in the treatment of membranous lupus nephritis with nephrotic syndrome. Nephrology. 2012;17:352–357. doi: 10.1111/j.1440-1797.2012.01574.x. [DOI] [PubMed] [Google Scholar]
- 37.Mysler E.F., Spindler A.J., Guzman R. Efficacy and safety of ocrelizumab in active proliferative lupus nephritis:results from a randomized, double-blind, phase III study. Arthritis Rheum. 2013;65:2368–2379. doi: 10.1002/art.38037. [DOI] [PubMed] [Google Scholar]
- 38.The ACCESS Trial group Treatment of lupus nephritis with abatacept: the Abatacept and Cyclophosphamide Combination Efficacy and Safety Study. Arthritis Rheum. 2014;66:3096–3104. doi: 10.1002/art.38790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Furie R., Nicholls K., Cheng T.T. Efficacy and safety of abatacept in lupus nephritis:a twelve-month, randomized, double-blind study. Arthritis Rheum. 2014;66:379–389. doi: 10.1002/art.38260. [DOI] [PubMed] [Google Scholar]
- 40.Liu Z., Zhang H., Liu Z. Multitarget therapy for induction treatment of lupus nephritis:a randomized trial. Ann Intern Med. 2015;162:18–26. doi: 10.7326/M14-1030. [DOI] [PubMed] [Google Scholar]
- 41.Mok C.C., Ying K.Y., Yim C.W. Tacrolimus versus mycophenolate mofetil for induction therapy of lupus nephritis: a randomised controlled trial and long-term follow-up. Ann Rheum Dis. 2016;75:30–36. doi: 10.1136/annrheumdis-2014-206456. [DOI] [PubMed] [Google Scholar]
- 42.Corapi K.M., Dooley M.A., Pendergraft W.F., 3rd Comparison and evaluation of lupus nephritis response criteria in lupus activity indices and clinical trials. Arthritis Res Ther. 2015;17:110. doi: 10.1186/s13075-015-0621-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Contreras G., Lenz O., Pardo V. Outcomes in African Americans and Hispanics with lupus nephritis. Kidney Int. 2006;69:1846–1851. doi: 10.1038/sj.ki.5000243. [DOI] [PubMed] [Google Scholar]
- 44.Isenberg D., Appel G.B., Contreras G. Influence of race/ethnicity on response to lupus nephritis treatment:the ALMS study. Rheumatology. 2010;49:128–140. doi: 10.1093/rheumatology/kep346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Freedman L.S., Simon R., Foulkes M.A. Inclusion of women and minorities in clinical trials and the NIH Revitalization Act of 1993–the perspective of NIH clinical trialists. Control Clin Trials. 1995;16:277–285. doi: 10.1016/0197-2456(95)00048-8. discussion 286–279, 293–309. [DOI] [PubMed] [Google Scholar]
- 46.Moher D., Hopewell S., Schulz K.F. CONSORT 2010 explanation and elaboration:updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c869. doi: 10.1136/bmj.c869. [DOI] [PMC free article] [PubMed] [Google Scholar]