Lupus Nephritis – challenges and progress

Abstract

Purpose of review:

The management of lupus nephritis (LN) remains unsatisfactory due to insufficiently effective treatment regimens and the dearth of reliable predictors of disease onset or progression to guide individualized therapeutic decisions. This review summarizes new findings related to LN over the last 18 months and discusses clinical needs that should be considered in order to advance trials of mechanism based therapeutic strategies.

Recent findings:

Collaborative teams are addressing how to improve disease definitions and are developing predictive models for disease onset, disease response and risk of flare in individual patients. More attention is being paid to clinical trial design. Advanced technologic approaches are allowing the analysis of small amounts of human tissue and urine in unprecedented detail so as to discover new pathogenic mechanisms and identify disease biomarkers. Novel therapies continue to be tested in disease models and include new strategies to protect renal tissue from cell damage and fibrosis.

Summary:

The collaborative efforts of patients, clinical and translational researchers, the pharmaceutical industry and funding sources are needed to advance therapies for LN. Specialized clinical centers can then deliver optimal and more personalized patient care that will improve patient outcomes.

Introduction

Lupus nephritis (LN) is a severe complication of Systemic Lupus Erythematosus that progesses to ESRD in ≈10% of patients within 5 years of onset. Current standards for diagnosis and treatment of LN are unsatisfactory and it is not possible to accurately predict responsiveness to therapy or the long-term outcome of individual patients. Although there has been a recent decrease in the severity of LN in European patients (1), perhaps reflecting a more comprehensive approach to lupus management, the risk of LN-related ESRD has remained unchanged in the US population since the 1990s (2). Immune-mediated inflammation is a major initiator of LN, but pathogenic mechanisms leading to ESRD are poorly understood and cannot be therapeutically addressed in a patient-specific manner. There is as yet no successful biologic therapy for LN and many unsolved problems in clinical trial design impact the interpretation of trial outcomes. In this paper, we will review recent advances in clinical and mechanistic approaches to LN and consider what is needed for translation of new information into successful clinical trials.

Diagnosis and outcome measures

Renal biopsy is the gold standard for diagnosis of LN. Treatment decisions are based on the International Society of Nephrology/Renal Pathology Society classification of glomerular involvement (3) and indices for active inflammation and chronicity. Importantly, although long-term renal outcomes are worse for proliferative disease, better predictive models for risk and outcomes are needed to direct therapeutic decisions. It is still unknown whether preemptive treatment of serologic flare will prevent subsequent LN onset (4) - this needs larger controlled studies and the development of models that predict the risk of renal flare (5, 6). Predictive models are also needed to define a uniform composite short-term treatment response that can predict long-term outcomes and be used either as a surrogate endpoint in clinical trials or as a guide for decreasing maintenance immunosuppression. As a start in this direction, a proteinuria cut-off of <0.7–0.8 g/dl at 12 months after LN onset has been confirmed as a biomarker of good long-term outcome in several studies (7–9) and a set of hazard index tools incorporating clinical data from the first 12 months of treatment, have been shown to predict long-term outcomes (7). A longitudinal study comparing the accuracy of spot urines to 24 hour collections strongly advocates use of 24 hour collections for accurate results (10). Because large patient numbers are needed to test and develop predictive models, the establishment of LN registries with prospectively collected data and biospecimens will be essential to refine current models (BOX1 - Outcome Prediction). An international LN registry would also address the problem of multiple small studies in distinct ethnic and racial groups that lack generalizability.

Box 1 – Outcome Prediction.

Predictors of poor outcome at disease onset

Clinical (1, 7)

• Male gender

• Younger age

• Arterial hypertension

• Increased serum creatinine

• African-American race

Histologic (73, 74)

• Proliferative disease

• High activity and chronicity index

• Glomerulosclerosis and crescents

• Interstitial inflammation

• Tubular injury

• Interstitial fibrosis

Biomarkers that need to be further evaluated (49, 75)

• Markers for tubular injury

• Non-albumin proteinuria

Predictors of poor outcome after treatment

• Lack of access to a specialized center (2)

• Absence of maintenance immunosuppressive therapy (1)

• Failure to reach proteinuria threshold of < 0.7–0.8 g/dl at 12 months (7–9)

• High activity index on second biopsy at 12 months (26)

Controversies in LN management and clinical trial design

Mycophenolate (MMF) or cyclophosphamide combined with high doses of prednisone are standard of care treatment for LN (11, 12). Clinical questions remain about the optimal management of lupus podocytopathy, renal microangiopathy and membranous nephritis (13–15). Because remission rates of LN are low even with optimal management, studies using combinations of standard immunosuppressives (16), or the introduction of a calcineurin inhibitor are being considered. Controlled trials in Asian populations suggest that the combination of low dose MMF with tacrolimus is more effective than MMF alone, but safety and long-term efficacy remain to be established in other populations (11, 17). The new calcineurin inhibitor voclosporin, modified to confer enhanced potency and decreased toxicity, has shown efficacy in combination with low dose MMF and a rapid steroid taper in two phase 2 trials (18, 19) - phase 3 studies are in process. Combinations of immunosuppressive agents with biologic drugs have not yet been successful in randomized clinical trials in LN patients. The addition of belimumab to cyclophosphamide or rituximab or both is a rational approach to prevent the expansion of autoreactive B cells by high levels of BAFF resulting from B cell depletion. However data from the CALIBRATE trial showed no improvement in outcome at 24 or 48 weeks of a regimen of cyclophosphamide, rituximab, prednisone and belimumab compared with the same regimen without belimumab (20). A study of belimumab plus either MMF or cyclophosphamide is ongoing. Given the increasing reports of decreased general SLE flare rates over the long term in patients treated with belimumab (21), longer term follow up will be needed to determine its benefit in LN.

MMF is the preferred treatment for remission maintenance (22, 23) but there are no data regarding the optimal duration of treatment and no definition of a low disease activity state predicting safe treatment withdrawal (24, 25). Two recent studies have addressed this question by performing repeat biopsies at 6–12 months after induction and have shown an alarming discrepancy between clinical and histologic response. A small prospective study showed that activity on a second biopsy performed 12 months after induction predicted subsequent renal flare following maintenance withdrawal, regardless of clinical parameters (26). A second study showed that chronicity at 6 months after induction predicted long term renal outcome regardless of clinical or histologic remission status (27); larger studies are needed to define the utility of second biopsies for guiding personalized treatment based on pathologic or molecular findings. Unfortunately, adherence to maintenance therapy among LN patients is unacceptably low and new approaches are needed to address the complex contributors to this behavior (28). In those patients who develop ESRD, timely referral for transplant is associated with a survival benefit by reducing deaths from comorbidities such as infections and cardiovascular disease (29).

Because of the continuing failure of clinical trials of rational therapies for LN (11), much thought is being given to clinical trial design (30). Limited duration phase 3 trials allow evaluation of remission induction but do not address subsequent flare prevention or long-term renal outcomes. Defining response for clinical care and endpoints in clinical trials remains problematic as there is no consensus on definitions of complete and partial renal response or remission and the utility of second renal biopsies is still unknown. Doubts about the inclusion of high doses of prednisone in LN clinical trials and about the robustness of outcome measures have led to difficulty in the interpretation of LN trial outcomes. Most biologic agents are tested with MMF, without considering whether each drug targets the same or different immune pathways. Design features that consider the mechanism of action of each drug in the context of genetic polymorphisms or biomarkers have not yet been incorporated into LN clinical trials. (Box 2 - Clinical Trial Considerations).

BOX 2 – Considerations in Trial Design.

Problems

• Confounders

  • high placebo response to standard of care therapy leads to small effect sizes, necessitating large patient cohorts
  • use of high dose background steroids may mask the effects of new immune therapies

• Enrolment criteria do not always reflect patients seen in clinical practice

• Optimal outcome measures are still not defined and are not uniform across trials

• Long-term outcomes are currently not measured

• In the absence of informative biomarkers, diagnosis of residual activity may require a second biopsy

Solutions

• When to treat

  • current approach

    • based on clinical evidence of nephritis e.g. proteinuria onset
  • future approaches

    • prevention based on evaluation of risk
    • preclinical initiation of treatment triggered by biomarker change
    • medication choice based on risk stratification (genetic, biomarker or OMICs driven)
• Drug mechanism should drive trial design

  • treatment of active disease vs. flare prevention
  • rational choice of standard of care therapy

• Stratification for individual patient differences should replace post-hoc analyses e.g. ethnicity, genetic polymorphisms, gender

• Uniform trial design may help to compare the effects of agents with similar mechanisms of action

• Measurements of compliance need to be incorporated, especially when multiple drugs are being used

• Improved understanding of disease pathogenesis should result in development of better diagnostic and therapeutic tools

Mechanisms for renal damage in LN

An increased understanding of disease pathogenesis may expand treatment strategies beyond global immunosuppression.

Genetics of LN:

Genetic risk factors for LN are only beginning to be described. In SLE patients of European descent, polymorphisms of PDGFRA, sodium glucose cotransporter Slc5a11, hyaluronan synthase 2, TNIP1 and MHC Class I and II alleles are associated with LN (31). Identification of LN-associated variants of ITGAM that increase its pro-inflammatory properties and genetic polymorphisms that decrease renal kallikrein expression have led to the development of therapies that specifically target these pathways (32, 33). A polymorphism that increases the expression of the adapter Dab2 that mediates TGFβ signaling in epithelial cells is associated with CKD in humans (34). Epigenetic studies have identified differential methylation of genes regulating the response to tissue hypoxia and interferon-mediated signaling in women with LN (35). European ancestry protects against LN (36) and it is therefore important to study genetic risk factors in patients of other ethnicities. APOL1 risk genotypes are associated with poor outcome of most forms of CKD in individuals with African ancestry, with the risk being intrinsic to the kidney. Several pathogenic mechanisms have been suggested, but the relative role of each mechanism is still not known, making the APOL1 risk alleles difficult to target therapeutically. Consideration of ApoL1 status in the kidney transplant setting is now being prospectively studied by the APOLLO Consortium (37, 38).

Cellular composition and gene expression in LN kidneys:

The kidney harbors multiple cell types and infiltrating immune cells add to the complexity of the microenvironment in the LN kidney. Two new technologies are being used to understand the heterogeneity of the renal microenvironment in LN patients so as to develop better diagnostic tools and individualized therapy. The first is two-photon microscopy together with cell distance mapping (CDM) to determine relationships between infiltrating renal cells (39). The combination of CDM with staining of more than 20 different antigens using only small amounts of frozen tissue will yield insights into inflammatory responses by revealing how various cell types interact in the kidneys.

The second technology is single cell transcriptome analysis of renal biopsies (40, 41). While the sequencing depth using this approach is relatively shallow, it allows a full description of the cell types present in individual kidneys and some understanding of cell functions that can be correlated with histologic and clinical variables and outcomes. Phase 1 studies of infiltrating renal cells in 24 LN patents and 10 controls revealed multiple subtypes of B, T and myeloid cells in the LN kidneys. NK cells and CD8 T cells with cytotoxic activity are the major proliferating immune populations; despite the identification of exhausted CD8 T cells in kidneys from MRL/lpr mice (42), no CD8 T cell exhaustion phenotype was identified in LN biopsies although it was readily identified in PBMCs. T follicular helper cells and activated B cells were present, with the accumulation of plasma cells and B cells with an age-associated B cell phenotype. These studies failed to show a predominance of IL-17 or IFNγ-producing CD4 T cells with the caveat that most patients had already been treated at the time of biopsy. There is also evidence for activation of the resident macrophage population and for renal infiltration with CD16⁺ inflammatory macrophages that then appear to transition to a M2-like phenotype that may orchestrate migration of other inflammatory cell subsets (41). Dysregulated repair function of these cells may contribute to their pathogenic potential (43).

A similar single cell transcriptomic analysis of renal stromal cells from 21 LN patients revealed both an interferon signature and a fibrotic signature in the tubular cells, both of which may be associated with poorer response to therapy (40). Analyses of other tissues such as urine and unaffected skin that are more amenable to repeat sampling may yield information that reflects changes in the kidneys (40, 41).

Gut microbial diversity and LN:

An alteration in the composition of the gut microbiota has been associated with the production of antibodies to a particular species, Ruminococcus gnavus only in patients with active LN (44) but not in inactive lupus patients. T and B cell tolerance to the gut microbiota may be lost if disturbance of the gut epithelial barrier allows bacterial translocation to sites where they may elicit an inflammatory response (45, 46). Alternatively, a change in the composition of the microbiota may induce pathogenic cross-reactive anti-self/anti-commensal immune responses. It remains to be determined whether the dysbiosis of LN is causative or reflects homeostatic disturbances associated with inflammation and immunosuppressive medications. Longitudinal studies are now needed to examine the course of gut dysbiosis in LN patients and to test the therapeutic applicability of approaches that restore commensal homeostasis and gut barrier integrity.

LN Biomarkers

Biomarker discovery in LN has progressed from analysis of individual candidate markers to unbiased high throughput methods such as mass spectrometry (47), multiplexed immunoassays, renal imaging (48) and modular transcriptome analyses. Proteomic studies indicate that small panels of biomarkers can distinguish LN from healthy control urine and that of active from inactive disease (49–53). A set of 6 biomarkers, Renal Activity Index for Lupus (RAIL) is associated with nephritis in children and to a lesser extent in adults in cross-sectional studies (54, 55). However, a recent longitudinal study using RAIL and additional biomarkers failed to identify a panel that outperformed GFR or predicted renal functional decline in individual patients (56). Nevertheless it may be possible to develop a home-based assay to be performed between visits to improve early detection of nephritis (57). Transcriptomic analyses have identified a peripheral blood neutrophil signature as a risk factor for LN, although it may not be a robust biomarker of disease response (58, 59). Finally there is still debate about the significance of renal deposition of the terminal Mac complex as a biomarker for complement activation that could be targeted by anti-C5 drugs (60, 61).

New approaches to therapeutic targeting

Mining of existing databases has revealed pathways that could be targeted by available drugs - repurposing of IL12/23 inhibitors, proteasome inhibitors and Jak inhibitors are examples of this approach. While the concordance of mouse and human interventions for LN has historically been poor, a number of new therapeutic targets have been recently tested in murine models. Allogeneic mesenchymal stem cells prevent LN in mouse lupus models and multiple mechanisms for their efficacy have been reported (reviewed (62)). Results in humans are conflicting and await randomized trials. Defective production of IL-2 by conventional T cells in patients with lupus and LN favors the generation of inflammatory T cells. Correction of this defect by low dose IL-2 improves survival in lupus mouse models and approaches to enhance IL-2 are now being developed for human use (63). Immune activation is associated with metabolic changes that favor cell proliferation and differentiation. Inhibition of both glycolysis and oxidative phosphorylation can effectively treat established nephritis in mouse models (64). Other approaches include targeting of the cholinergic inflammatory reflex with galantamine (65), targeting of renal macrophages by introducing deficiency of IL-34 (66), targeting of inflammatory cytokines by introducing deficiency of inactive rhomboid 2 (iRhom2) (67), targeting of Th17 cells and podocyte injury by inhibition of CAMK4 (68) and targeting of inflammasome mediated renal injury by inhibition of NLRP3 (69, 70). Because some of these therapies may have unacceptable systemic consequences, new approaches are needed to direct such therapies only to the inflamed site. Nanogels are specially formulated drug-containing liposomes that can be targeted to specific cells using antibodies. In a recent proof-of-principle study a nanogel containing a CamK4 inhibitor was directed to podocytes using anti-nephrin or anti-podocin antibodies and prevented podocyte injury and proteinuria in a mouse lupus model (68). Development of such site-directed therapeutic agents may enhance our current ability to prevent flares while decreasing the need for chronic global immune suppression.

It is increasingly evident that non-immune pathways contribute to renal injury in LN. One recent question is whether the kidneys can be protected from inflammatory damage by altering the resilience of resident renal cells to oxidative stress (71). Tubular cell injury is a common feature of all forms of CKD. Tubular epithelial cells rely on mitochondrial fatty acid oxidation for their energy supply and can be protected in mice by strategies that conserve this pathway (72). Tubule specific deletion of the TGFβ signaling effector Dab2 in mice also protects from renal injury and fibrosis, suggesting another therapeutic approach especially in genetically susceptible individuals (34). Since multiple cell types are affected during renal injury, it is expected that multi-OMICs discovery experiments will uncover new pathways for targeting.

Conclusions

Coordinated approaches that involve all stake holders are needed to prevent and treat LN, especially in those ethnic groups whose outcome is historically poor. Collaborative interactions can help to identify knowledge deficits and clinical needs so as to design appropriate multi-center studies. Registries and uniform biospecimen collection will allow testing of hypotheses that can only be addressed with longitudinal data. Mechanistic studies involving individual renal cell types from human samples will allow us to further unravel precise mechanisms in appropriate mouse models and to predict new and/or synergistic therapeutic approaches. Because it is easier to prevent than to treat established LN, identification of patients at risk and at home monitoring for early renal changes may improve outcomes. Providing access to care in specialized lupus centers can facilitate early detection, encourage and monitor patient compliance, improve management of ancillary morbidities and enable access to renal transplant so as to improve the long term outcome of all patients with LN.

Key points:

• Lupus nephritis (LN) remains a significant unmet need that causes morbidity and mortality in patients with SLE

• There is a pressing need for the development of non-invasive predictors of LN risk and prognosis that can only be addressed using clinical registries with optimal collection of patient data and biospecimens

• Using advanced technologies, human tissue can now be examined in unprecedented detail, yielding new insights into disease mechanisms

• The successful development of new therapies will need more attention to clinical trial design and the consideration of individual patient features including genetic polymorphisms and tissue characteristics.

• Improvement of disease outcomes will need universal patient access to specialized clinics that monitor disease progression and medication adherence and deliver optimal access to both standard care and new therapies.