Therapy of lupus nephritis: lessons learned from clinical research and daily care of patients

Despite numerous randomized clinical trials over the last three decades for identifying the optimal treatment option for lupus nephritis, renal involvement still significantly impacts the survival and quality of life of patients with lupus and the search for the ideal immunosuppressive regimen is far from complete. The purpose of this review is to summarize the major recent achievements in the field. More specifically, the following topics will be discussed: intravenous cyclophosphamide versus mycophenolate mofetil (MMF) for induction; azathioprine versus MMF for maintenance; targeted therapies. The review will address clues for optimal global care, such as the need for complete initial evaluation, the importance of patient education, the unmasking of non-compliance to therapy, the reason for an early treatment switch in non-responding patients, the need for prolonged immunosuppression, optimal renal protection, and prevention of cardiovascular disease and other comorbidities.

glomerular necrosis, should guide therapy at the bedside at least as much as the ISN/RPS class itself.

Education and obsessional follow-up
At diagnosis, physicians, dedicated nurse practitioners, and patient associations should provide adequate educational material that replaces inaccurate and mostly frightening uncontrolled websites and forums that create misconceptions. Follow-up must be obssessional, follow a standardized scheme, and directly involve patients (for example, by allowing them direct access to their own medical chart, as already practiced in some Scandinavian countries and as tested in our own lupus clinic).

Unmasking non-compliance to therapy
Non-adherence to pharmacological therapy is one of the major reasons for treatment failure. It should be discussed with the patient from the very fi rst visit and always suspected in 'refractory' cases. Measurements of whole-blood hydroxychloroquine titers may help in detecting poor adherence, based on the long elimination half-life of the drug, thereby bypassing the so-called 'white coat compliance' (in which the patient takes a few pills before the visit!). Using this approach as an objective marker of poor adherence to therapy, a recent study could demonstrate that non-compliant patients with lupus experienced more lupus fl ares [5].

Less glucocorticoid
No drug has improved survival of patients with LN as dramatically as glucocorticoids (GCs). However, no drug has been responsible for as many side eff ects! In randomized trials, many of the severe infections attributed to one immunosuppressive regimen or another may well be due to GCs, their common denominator. Intriguingly, the dose of GCs needed to control LN is an understudied topic. Recent trials have successfully used a starting oral prednisolone daily dose of 0.5 mg/kg (associated with a few intravenous (IV) methylprednisolone (MP) pulses) instead of the 'classical' 1 mg/kg oral daily dose [6,7]. Th at oral GC therapy can be completely left out of an induction immunosuppressive regimen is far from proven but is suggested by a recent uncontrolled study showing that 13 of 46 patients (28%) with LN went into complete renal remission at 6 months after treatment with 2 g of rituximab (1 g given intravenously twice 2 weeks apart) combined with IV MP (500 mg given twice 2 weeks apart) and followed by MMF with no oral GCs [8].

Individualized therapy
As researchers, we rightly consider that results of random ized trials should be translated as such at the bedside. But as clinicians, we know that many variables other than evidence-based medicine will guide therapy.
Ethnicity, age, doubts about compliance, severity of the clinical presentation, renal pathology fi ndings, and comorbidities will obviously be taken into account. More and more, patient-oriented issues such as pregnancy plans or even fear of certain side eff ects will contribute to the decision.

Early treatment switch
We would hope that all patients with LN would respond to the same fi rst-line immunosuppressive therapy. Again, as clinicians, we know that this assumption is not valid and we are still very bad at predicting who is going to respond to which drug, in particular because we largely ignore pharmacogenomic issues and the pathogenic pathways at work in a given patient. In other words, we proceed by a 'trial and error' approach. One way to overcome this shortcoming is to evaluate the early response to therapy and then switch to another treatment in case of an unsatisfactory response. We indeed demonstrated, in a long-term prospective trial, that an early response to immunosuppressive therapy at 6 months was the best predictor for a good long-term renal outcome in LN [9]. Th us, patients with a good long-term renal outcome, defi ned by a serum creatinine value of not more than 1.4 mg/dL after a follow-up of more than 5 years, experienced a prompt and dramatic drop in proteinuria after 3 and 6 months of therapy, and this was not the case in patients who later developed renal impairment. Th e positive predictive value of a 75% decrease in proteinuria at 6 months for good long-term renal outcome was 90%.

Prolonged maintenance immunosuppression
Although we lack prospective trials specifi cally aimed at comparing diff erent lengths of maintenance immunosuppressive regimes, the very high renal relapse rate (35% in most long-term series) favors prolonged immunosuppression [10]. In a small but interesting withdrawal study performed in patients with LN, an inverse correlation between the length of treatment/remission before withdrawal and the risk of relapse was found [11]. Interestingly, in Europe, the current practice (as refl ected by the 10-year follow-up data from the ELNT) is to keep patients long-term on AZA (or MMF) and on low-dose GCs. Th us, after 10 years, 73% and 56% of patients randomly assigned in the trial were still on low-dose prednisolone (mean daily dose of 5.5 mg) and on another immunosuppressant, respectively [1]. Th e possibility that the excellent long-term outcome in these patients (only 7% ESRD at 10 years) is linked to prolonged immunosuppression is far from proven yet is not too far fetched.

Optimal renal protection
In LN, as in other chronic renal diseases (see guidelines of the Kidney Disease Outcomes Quality Initiative and of Kidney Disease: Improving Global Outcomes), blood pressure must be tightly controlled and maximal systolic/ diastolic values should not exceed 120/80 mm Hg. Antiprotenuric therapy should be prescribed in all patients with LN by combining, if needed, angiotensin-converting enzyme inhibitor (ACEI), angio tensin recep tor blocker (ARB), and loop diuretics.

Prevention of cardiovascular disease
Pivotal data by the Toronto group fi rst demonstrated a bimodal mortality pattern in patients with lupus by showing an early peak related to lupus itself and its therapy and a second peak due to cardiovascular disease [12]. Th is observation was confi rmed in patients with lupus by the demonstration of accelerated atheroma as assessed by carotid Doppler studies (plaques and intimamedia thickness) [13] and by coronary computerized tomography scan studies (calcifi cations) [14]. In this respect, control of all cardiovascular risk factors, such as smoking avoidance, weight control, exercise, and cholesterol levels, is of the utmost importance. As for diabetes mellitus, we set the low-density lipoprotein cholesterol cutoff value for treatment with a cholesterol-lowering drug as low as 115 mg/dL.

Prevention of other comorbidities
Several studies have demonstrated that many patients with lupus suff er from GC-induced osteoporosis, which should always be prevented by calcium salts and vitamin D 3 supplements and, in selected cases, by bisphos phonates [15]. All patients with lupus should be immunized against Streptococcus pneumoniae given the incidence and severity of the infections caused by this pathogen [16]. Th e higher prevalence of uterine cervix carcinoma and human papillomavirus infection in patients with lupus in comparison with a control population [17] strongly argues in favor of early immunization in teenage patients with lupus.

Achievements and failures over the last decade
Mayo Clinic and National Institutes of Health (NIH) investigators deserve the credit for running the fi rst controlled trials in LN and for stressing the need for long-term follow-up to unmask diff erences between treatment arms. For decades, we have been prescribing the so-called 'NIH regimen' consisting of monthly (for 6 months) and then quarterly (for up to 1 year after complete remission) high-dose (0.75 to 1.5 g/m 2 ) intermittent IV CY pulses, combined mostly with IV MP pulses [18][19][20]. Although this regimen was shown to be superior to GCs at preserving renal function in the long run, the associated high incidence of premature gonadal failure made it unpopular among young female patients with LN and thereby generated alternative proposals, including the use of lower doses of IV CY (the Euro-Lupus regimen) or the use of MMF.

The Euro-Lupus regimen: 20 years later
Low-dose IV CY was introduced by Graham Hughes at St. Th omas' Hospital in London in order to minimize the side eff ects of the drug. A sequential approach consisting of prescribing a potentially toxic drug (IV CY) for a short period of time before switching to a more patient-friendly drug, namely AZA, for a long period of chronic immunosuppression was suggested [21]. Th e concept was further tested in a controlled trial, the ELNT, taking advantage of a European network of physicians who had trained together in Hughes's lupus unit.
Briefl y, we compared an NIH-like IV CY regimen (monthly pulses for 6 months and two additional quarterly pulses) and a low-dose IV CY regimen (six fortnightly pulses of a fi xed dose of 500 mg), both of which were followed by AZA (started at week 12 in the low-dose group and at week 44 in the high-dose group) [6]. All patients received three IV MP pulses of 750 mg and a moderate dose of oral prednisolone (starting mostly at 0.5 mg/kg per day) with a stringent tapering regimen. To summarize, after 10 years of follow-up, the low-dose regimen achieved results comparable to those of the NIH-like protocol, and there was a very low rate of ESRD (7%) [1]. Whether this conclusion, based on a European trial performed in a popu lation composed mainly of Caucasians, can be extra polated to other ethnic groups is currently unknown. It has been said that patients included in the ELNT did not suff er from severe renal disease. While this is true based on the percentage of patients presenting with nephrotic syndrome (28%) or renal impairment (22%) or both, it should be stressed that all suff ered from biopsy-proven proliferative LN and that 47% of them had partial circumferential crescents on central pathology review. In this respect, the possibility that patients were captured in very early clinical stages of LN is not too far fetched. Most patients were indeed followed in rheumatology-based lupus clinics, and early kidney involvement was therefore promptly detected by very regular dipstick examinations. Such a recruitment obviously diff ers from that of a nephrology clinic, where patients with LN present mostly with acute renal failure, implying that the underlying process is already at work for weeks or months before it is detected. In this subset of patients, the lack of signi fi cant extra-renal disease activity may delay the diagnosis of systemic lupus erythematosus.
Although some caveats against an indiscriminate use of the low-dose IV CY regimen can be raised, the many advantages of this regimen must be emphasized: no need for prolonged in-patient stay (500 mg of IV CY can be administered as a 30-minute drip infusion in a day clinic), no need for hyperhydration, no need for mesna, no need to monitor white blood cell count, optimal compliance (IV route), and -last but not least -virtually no risk of premature gonadal failure.

Mycophenolate mofetil: the challenger saga
No other drug has generated as much hope in LN as MMF over the last decade. Yet more than 10 years after the fi rst randomized controlled trial comparing oral CY and MMF [22], the jury is still out. Th e drug was introduced as an immunosuppressant in transplant medicine, in which it was shown to be superior to AZA at preventing graft rejection [23], although not all transplant teams share this view [24]. Logically, MMF has been tested in LN as an induction immunosuppressant and as a maintenance drug.
In a randomized LN trial (in which more than half of the participants were African-American) in the US, MMF was found to be superior to NIH IV CY at inducing complete remission of renal signs at 6 months (22.5% in the MMF group versus 5.8% in the IV CY arm) [25]. However, in a subsequent multiethnic trial (the Aspreva Lupus Management Study, or ALMS), the two immunosuppressants were found to be equivalent at inducing a renal response at 6 months (56.2% in the MMF group versus 53.0% in the IV CY arm), and MMF showed an advantage over IV CY in non-Caucasian/non-Asian patients [26]. To summarize, MMF can be con sidered to be at least equivalent to CY at inducing an early renal response, although we still lack long-term data (that is, 10-year follow-up) on patients induced with MMF. As mentioned above, the decision to start MMF or CY at the bedside will be infl uenced by many variables. Th us, if one anticipates non-observance to therapy, low-dose IV CY may be the ideal choice. Conversely, in a 38-year-old African-American woman with pregnancy plans, it may be wiser to prescribe MMF. Th e relatively low rate of renal response and, a fortiori, the very low rate of renal remission at 6 months (regardless of the initial therapy and despite the use of high-dose GCs) clearly illustrate the need for induction regimes that are more effi cacious. On the other hand, the availa bility of at least two drugs, besides GCs, to initiate immunosuppression leaves open the possibility of switch ing patients not responding to one drug to the other, although this has not been supported by data so far.
MMF was also proven eff ective for maintenance immunosuppression in LN. In a fi rst trial, Contreras and colleagues [27] compared quarterly high-dose IV CY pulses (NIH), AZA, and MMF for maintenance therapy after induction with monthly IV CY pulses. Mortality (in particular, mortality due to infections) was signifi cantly higher in the IV CY group in comparison with the two other groups, a result that largely contributed to a ban on quarterly IV CY pulses. Two other randomized trialsthe maintenance phase of the ALMS [28] and the MAINTAIN Nephritis Trial [7] -have compared AZA and MMF for main te nance immunosuppression.
Patients who were randomly assigned in the ALMS and who had responded to MMF or IV CY at 6 months (± 50% of the original cohort) were randomly assigned again to receive AZA or MMF for an additional period of 3 years. Th e primary outcome measure for the maintenance phase was time to treatment failure, a composite endpoint that consisted of renal fl are, sustained doubling of serum creatinine, initiation of rescue therapy for LN, ESRD, and death. According to Kaplan-Meier analysis, time to treatment failure was statistically shorter for AZA patients in comparison with MMF patients. Superiority of MMF over AZA was demon strated not only for patients who had achieved a response at 6 months with MMF (36% and 21% failure rates in the AZA and MMF maintenance groups, respectively) but also for patients induced with IV CY (28% and 11%, respectively). Of note, the lowest failure rate was observed in patients who had been induced with IV CY and maintained with MMF. Side eff ects were comparable [28].
In the MAINTAIN Nephritis Trial [7], which was performed in a Caucasian European population, all patients received the same induction regimen, namely GCs (IV MP and oral) and six fortnightly pulses of IV CY (according to the Euro-Lupus regimen). At month 3, irres pectively of the magnitude of their renal response, patients were switched to AZA or MMF on the basis of the random assignment performed at baseline. Time to renal fl are (the primary endpoint of the trial) did not diff er after a 5-year follow-up period: the renal fl are rates were 19% and 25% for MMF and AZA patients, respectively, and this diff erence was not statistically signifi cant. Side eff ects, except for transient cytopenias (which were more frequent in the AZA group), did not diff er between the two groups. Repeat renal biopsies performed 2 years after random assignment in a representative set of patients failed to reveal an advantage of one drug over the other in terms of activity/chronicity indices [29].
Th e design and patient population of the two MMF/ AZA maintenance LN trials are so diff erent that one should avoid comparing them head-to-head. Rather, we would suggest interpreting the results in a positive way since at least two drugs are available for long-term use in patients with LN and have a reasonable effi cacy and toxicity profi le, thereby -again -suggesting that patients failing on one drug could be rescued by the other.

Targeted therapies: sparks twinkling in the night or falling stars?
Several biologics have been, are, or will be evaluated in LN. Results of randomized trials are available for rituxi mab (RTX) (anti-CD20 mAb), ocrelizumab (OCR) (humanized anti-CD20 mAb), and abatacept (CTLA4-Ig). Sadly, these results were not conclusive.
Th e LUNAR (Lupus Nephritis Assessment with Rituximab) trial was a phase III, randomized, doubleblinded, placebo-controlled, multicenter study of 144 class III or IV patients with LN [30]. All patients received GCs (2 g of IV MP followed by 0.75 mg/kg per day of prednisolone) and MMF (up to 3 g/day) as background therapy. RTX (two infusions of 1 g, 15 days apart with retreatment 6 months later) or placebo was given as add-on treatment. Th e primary outcome (that is, the proportion of patients achieving complete or partial renal response at week 52) was met by 57% and 46% of the patients randomly assigned in the RTX and placebo groups, respect ively. Th is diff erence was not statistically signifi cant. A trend in favor of RTX was noted in African-Americans (70% versus 45%; P = 0.20).
Th e BELONG trial was aimed at testing the effi cacy and safety of OCR (400 or 1,000 mg given on days 1 and 15 and weeks 16 and 32) in patients with active ISN/RPS class III or IV LN treated with standard of care (SOC) consisting of GCs (0.5 to 0.75 mg/kg per day; with taper to not more than 10 mg/day over the course of 10 weeks) and either MMF (up to 3 g/day; 63% of the patients) or the Euro-Lupus regimen (IV CY 500 mg six times every 2 weeks, followed by AZA 2 mg/kg per day; 37% of the patients), as selected by investigators [31]. Th e study was terminated early in September 2009 (after complete enrollment) because an interim analysis by the Data and Safety Monitoring Board revealed safety concerns. Th us, serious infections (that is, requiring IV antibiotics) were two times more common in the 400 mg OCR group (but, surprisingly, not in the 1,000 mg OCR group) in comparison with placebo. Intriguingly, this was the case only in the subgroup of patients given MMF as SOC background therapy (not on Euro-Lupus back ground). In a modifi ed intent-to-treat population, 63.3% of the OCR patients achieved complete or partial renal response in comparison with 51.4% of the placebo patients (P = 0.07). Interestingly, the diff erence was more pronounced, in favor of OCR, in patients given the Euro-Lupus regimen as SOC background treatment (59.4% versus 42.9%).
Th e negativity of these two B-cell blockade LN trials, together with the negative results of the EXPLORER (Exploratory Phase II/III Systemic Lupus Erythematosus Evaluation of Rituximab) trial performed in patients with non-renal lupus [32], contrasts with the clinical practice of experienced physicians who have been successfully using RTX in LN cases who had failed on conventional therapy [33], thereby raising concerns about the design of the two RTX LN studies. Did concomitant therapy (with high doses of GCs and other immunosuppressants) mask the eff ects of RTX? Were the trials too short-term to unmask the benefi ts of RTX? Were the doses of RTX appropriate? Should RTX be prescribed only in a subset of early refractory LN cases that do not respond to SOC after 3 to 6 months? Th ese are some of the many questions that should be addressed before RTX can be considered a falling star! In this respect, the recent positive results of EMBLEM [34], a short-term phase II epratuzumab (anti-CD22) trial performed in patients with lupus without severe renal/central nervous system involvement, further stress the potential value of a B-cell blockade approach in lupus.
Two randomized trials were designed in order to test the effi cacy of abatacept as add-on induction therapy for LN. Th e molecule, also called CTLA4Ig, is a selective T-cell co-stimulation modulator that binds to B7 (expressed on antigen-presenting cells) and thereby blocks its interaction with CD28, expressed on T cells, preventing T-cell activation. An industry-sponsored trial testing abatacept on a GC and MMF background was recently reported to be inconclusive [35]. We now eagerly await the results of ACCESS (Abatacept and Cyclo phosphamide Combina tion: Effi cacy and Safety Study), a study that was spon sored by the Immune Tolerance Network and the National Institute for Allergy and Infectious Diseases and that was aimed at comparing abatacept and placebo on a Euro-Lupus regimen background.
Other biologics, such as anti-BLyS belimumab, anti-CD22 epratuzumab, anti-interleukin 6, or anti-type I interferons, will most likely be tested in LN. Th e very recent US Food and Drug Administration (FDA) approval of belimumab for the treatment of moderately severe lupus, after the success of two phase III pivotal trials performed in patients with non-renal lupus [36,37], will likely trigger a proper LN trial with the drug.

Treatment of membranous lupus nephritis
Membranous lupus glomerulonephritis (ISN/RPS class V LN) is characterized by subepithelial immune deposits. It is usually associated with proliferative disease, namely ISN/RPS class III (focal) or IV (diff use) LN. In this case, most physicians consider that the presence of proliferative lesions guides therapy. However, some patients suff er from isolated ISN/RPS class V membranous LN either as the initial manifestation of renal involvement or later in their evolution. Th us, patients treated for proliferative disease sometimes switch to membranous LN unmasked by a control kidney biopsy performed because of persisting proteinuria despite immunosuppression.
Treatment of pure membranous LN probably does not diff er from that of idiopathic membranous nephropathy. Based on the nephrology experience, a distinction must be made between nephrotic and subnephrotic patients. Th us, in subnephrotic patients, a 'watchful waiting' approach with optimal blockade of the angiotensin renin aldosterone system is usually appropriate. In patients with nephrotic range proteinuria, immuno suppressants are usually added by combining GCs and one of the following drugs: CY, MMF, AZA, or calcineurin inhibitors. In a controlled trial performed at the NIH, 42 patients with pure membranous LN received (a) cyclosporine and alternate-day GCs, (b) alternate-month IV CY for six doses and alternate-day GCs, or (c) alternateday GCs alone. Regimes (a) and (b) were both found to be superior to (c) at 1 year, but, interestingly, while the eff ect was prompter with calcineurin inhibitors, relapses were more frequent than in the CY group [38]. A subset analysis performed in the 84 pure membranous LN cases randomly assigned in two large recent randomized LN trials comparing MMF and IV CY [25,26] revealed that the two drugs were equally eff ective at 24 weeks [39].

Conclusions
Major progress has been achieved in the fi eld of LN over the last decade. Table 1 tentatively summarizes the respec tive indications of the most commonly used immunosuppressants. We now use drugs in a much more patient-friendly way, thereby minimizing side eff ects. MMF is now clearly on board. We have at least two drugs for induction (MMF and IV CY) and maintenance (MMF and AZA) immunosuppression in LN. Together with ever increased standards for optimal global care, this has contributed to lower ESRD rates to fi gures never anticipated before. Recently, a new drug, anti-BLyS belimumab, became the fi rst drug to be approved by the FDA for the treatment of LN since cortisone and hydroxychloroquine were labeled for lupus 52 years ago. Th ese exciting times, however, should not disguise the fact that LN still impacts the survival of patients with lupus, that no cure can be off ered at present, and that further research breakthroughs are still awaited.