After four decades of lupus nephritis trials, is there a “best” therapy?

At some point during their disease course more than half of all patients with systemic lupus erythematosus will have clinically evident kidney disease, generally termed lupus nephritis (LN). The treatment of LN has improved remarkably over the last four decades thanks, in large part, to a consistent record for well-performed, randomized trials evaluating a growing number of available induction and maintenance treatment regimens.¹ The result is a markedly improved prognosis for patients with LN. Whereas kidney survival at 5 years was as low as 20% before 1980, current treatment strategies have improved this rate to as high as 80% in the past decade.^2,3

In this issue of the American Journal of Kidney Diseases, Palmer et al. report the results of a network meta-analysis of randomized trials for proliferative forms of LN – class III or IV lesions, either alone or in conjunction with class V, membranous, lesions – which carry the highest risk for progression to end stage kidney disease (ESKD).⁴ They identified 53 studies involving 4222 participants, dividing their analyses into studies evaluating induction-phase treatment (45 trials with 3623 participants) and maintenance-phase treatment (8 trials with 599 participants). The network meta-analysis design allowed these investigators to compare multiple treatment regimens across trials based on a common comparator: intravenous (IV) cyclophosphamide for induction studies and azathioprine for maintenance studies. As lupus trials have employed a number of different therapeutic regimens, including combinations of therapies (e.g. MMF + calcineurin inhibitors), various modes of delivery (e.g. IV vs. oral cyclophosphamide), and non-pharmacologic interventions (e.g. plasma exchange), the network meta-analysis design is, at least on the surface, an ideal way to evaluate 40-plus years of lupus nephritis studies to determine what is the most effective and safe therapy.

The network meta-analysis found statistically significant higher odds for complete remission with an induction phase regimen of MMF (OR 1.44, 95% CI 1.00–2.06), calcineurin inhibitors alone (OR 1.74, 95% CI 1.09–2.79), or MMF + calcineurin inhibitors (OR 2.69, 95% 1.74–4.16) compared to a traditional regimen of IV cyclophosphamide. Indeed, one of the authors’ main conclusions is that the combination of MMF + calcineurin inhibitors ranked as the “best treatment” to induce remission, and that a trial comparing calcineurin inhibitors alone vs. MMF + calcineurin inhibitors should be a priority. The data on MMF + calcineurin inhibitors, however, come from only two studies, both from China with shared authors.^5,6 Because outcomes in LN are so heavily influenced by race and ethnicity,⁷ it would be premature to extend the results of Chinese studies on MMF + calcineurin inhibitors to international populations. Due to insufficient data, the authors were unable to explore distributions of treatment classes by ethnicity or race.

In addition, the authors themselves note that complete remission was a heterogenous outcome with variable definitions across the studies analyzed. The two studies included here on MMF + calcineurin inhibitors used proteinuria at 6 months as the primary outcome measurement. Because calcineurin inhibitors can lower proteinuria independent of their effect on immune function, a more appropriate assessment of these agents’ efficacy would be at 12 or 24 months, when the drugs have been tapered to lower doses or discontinued, to gauge whether the remission achieved earlier is sustained. Similar pitfalls of relying solely on a proteinuria-based remission endpoint will arise when analyzing the results of the ongoing studies of voclosporin, a new calcineurin inhibitor, as a treatment for proliferative LN, which has shown promising results in a phase 2 study (presented in abstract form) and is now being evaluated in a larger, phase 3 study (NCT03021499).

Alternatively, a non-proteinuria based outcome, such as progression to ESKD, doubling of serum creatinine, or all-cause mortality may be more appropriate for this kind of analysis. Although the data for these outcomes in LN trials is of poorer quality than remission data, mostly due to lack of long-term follow-up, the available results in this meta-analysis suggested similar effects between induction strategies relying on MMF or calcineurin inhibitors, alone or in combination, versus IV cyclophosphamide. An even more robust outcome to evaluate induction phase therapy may be repeat, or protocol, biopsies. Early clinical and histologic outcomes are discordant in proliferative LN,⁸ and progression from active to chronic lesions on a biopsy remains the most consistent predictor of long term risk for ESKD.⁹ These data, of course, were not available for the meta-analysis as repeat biopsies are not routinely performed in LN trials.

Although the data used for maintenance phase comparisons were smaller compared to the data available for induction phase comparisons, the results from this section of the network meta-analysis are likely more reliable given the greater homogeneity in subjects enrolled in such trials and the outcome of interest. Maintenance phase studies enroll patients who have already achieved remission and use an outcome of relapse. In other words, for LN patients, a negative outcome like relapse – easily diagnosed by presence of hematuria, proteinuria, or change in renal function – is a more objective and uniform endpoint than a positive outcome like remission, which is based on a variable and often transient assessment of proteinuria levels. Likewise, the network treatment estimates for adverse effects of the various immunosuppressive regimens under study, which are not influenced by time of assessment and use fairly reproducible criteria, are important data to emerge from this study. Palmer et al.’s finding that MMF was the best treatment to maintain disease remission while also accompanied by lower risks of alopecia, nausea, and vomiting compared to IV cyclophosphamide should lend greater support to using this agent in both induction and maintenance phase. Compared to oral cyclophosphamide, MMF also conferred significantly lower odds of ovarian failure (OR 0.09, 95% CI 0.01–0.99). This difference in rates of ovarian failure did not bear out comparing MMF to IV cyclophosphamide (OR 0.71, 95% CI 0.20–2.45), however. This may be due to the relatively smaller number of adverse events in both groups or from more recent trials on IV cyclophosphamide using a lower dose regimen (i.e. Euro-Lupus dosing)¹⁰ than in earlier studies (i.e. NIH protocol dosing).¹¹

Today, the most consistently used induction and maintenance therapy for LN in everyday clinical practice is MMF. This is based as much on the drug’s performance in clinical trials as on its favorable side effect profile for patients with LN who wish to preserve fertility, an important consideration given that the disease commonly presents in young men and women of childbearing age. The network meta-analysis by Palmer et al. strengthens the case for continuing this practice, as MMF emerges as an effective and safe therapy for proliferative LN against a backdrop of four decades of clinical trials testing a myriad of treatment regimens.