Redefining lupus nephritis: clinical implications of pathophysiologic subtypes

Redefining lupus nephritis: clinical implications of pathophysiologic subtypes 1 Renal Division, Department of Medicine, Peking University First Hospital, Institute of Nephrology, Peking University, Key Laboratory of Renal Disease, Ministry of Health of China, Beijing 100034, P. R. China. 2 Department of Nephrology, Peking University International Hospital, 1 Zhongguancun Life and Science Street, Changping District, Beijing 102206, P. R. China. 3 Department of Pathology and Laboratory Medicine, Cedars Sinai Medical Center, Los Angeles, 90048 California, USA. 4 Department of Medicine, David Geffen School of Medicine at UCLA, 8 Bethany, Laguna Niguel, 92677 California, USA. 5 Peking-Tsinghua Center for Life Sciences, 5 Summer Palace Street, Haidian District, Beijing 100871, P. R. China. Correspondence to M.-H.Z. mhzhao@bjmu.edu.cn doi:10.1038/nrneph.2017.85 Published online 3 Jul 2017 Feng Yu 1,2, Mark Haas 3, Richard Glassock 4 and Ming-Hui Zhao 1,5 Abstract Systemic lupus erythematosus (SLE) is associated with a broad spectrum of clinical and immunologic manifestations, of which lupus nephritis is the most common cause of morbidity and mortality. The development of nephritis in patients with SLE involves multiple pathogenic pathways including aberrant apoptosis, autoantibody production, immune complex deposition and complement activation. The 2003 International Society of Nephrology/Renal Pathology Society (ISN/RPS) classification system for lupus nephritis was widely accepted with high intraobserver and interobserver concordance to guide therapeutic strategy and provide prognostic information. However, this classification system is not based on the underlying disease pathophysiology. Some additional lesions that contribute to disease presentation, including glomerular crescents, podocyte injury, tubulointerstitial lesions and vascular injury, should be recognized. Although outcomes for patients with lupus nephritis have improved over the past 30 years, treatment of this disease remains challenging and is best approached on the basis of the underlying pathogenesis, which is only partially represented by the various pathological phenotypes defined by the ISN/RPS classification. Here, we discuss the heterogeneous mechanisms involved in the pathogenesis of lupus nephritis and how improved understanding of underlying disease mechanisms might help guide therapeutic strategies. Systemic lupus erythematosus (SLE) is a chronic autoimmune disease that affects multiple organs and tissues, of which the development of kidney disease is the most important predictor of morbidity and mortality. Renal involvement is indicated by the presence of haematuria, proteinuria or decreased renal function, and requires confirmation by renal biopsy. The purpose of renal biopsy is to confirm the diagnosis of lupus nephritis, establish pathologic patterns, activity and chronicity of renal injury, guide therapeutic strategy, and provide prognostic information including the likelihood of response to treatment and of progression to end-stage renal disease (ESRD). A well-established histopathologic classification system that is predictive of therapeutic response and outcomes is therefore required. Three major guidelines for the management of lupus nephritis exist the American College of Rheumatology (ACR) guideline, the Kidney Disease: Improving Global Outcomes (KDIGO) guideline, and the Joint European League Against Rheumatism/ European Renal Association European Dialysis and Transplant Association (EULAR/ERA EDTA) guideline all of which base their recommendations on the 2003 International Society of Nephrology/ Renal Pathology Society (ISN/RPS) classification system 1 4 (BOX 1). Despite the availability of these guidelines, remission is achieved in only 50 70% of patients and 10 20% of patients will progress to ESRD within 5 years of diagnosis 1 3. These poor outcomes represent an enormous challenge to the use of therapeutic strategies based on the current classification system 5 9. Moreover, several clinical trials of immunomodulatory therapies, including trials of mycophenolate mofetil 10, ciclosporin 11, and rituximab 12, failed to demonstrate superiority over conventional therapies. Although the failure of these trials might be in part attributable to the selection of incorrect end points 13, their failure might be also partly attributed to the molecular heterogeneity and complex nature of the disease, suggesting that targeted approaches might be needed to improve therapeutic efficiency in lupus nephritis. In support of this notion, a 2016 study that profiled the blood transcriptome of 158 paediatric patients with SLE confirmed a common type I interferon signature and identified a plasmablast signature as the most robust biomarkers of disease activity. Importantly, gradual NATURE REVIEWS NEPHROLOGY VOLUME 13 AUGUST 2017 483

Key points The main purposes of renal biopsy in patients with systemic lupus erythematosus are to confirm the diagnosis of lupus nephritis, to assess disease activity and/or chronicity, guide therapeutic strategy, and provide prognostic information Although the 2003 International Society of Nephrology/Renal Pathology Society (ISN/RPS) lupus nephritis classification system has been widely accepted, some additional lesions reflect the underlying disease pathogenesis and should be included or otherwise recognized The ISN/RPS classification system is based on histology and not necessarily on the underlying pathogenesis, which needs to be the focus of treatment The presence of glomerular crescents, podocyte injury, tubulointerstitial lesions and thrombotic microangiopathy, in particular, are indicative of underlying disease processes and should be considered when assessing patients with lupus nephritis The current management of lupus nephritis is based on steroids and non-selective immunosuppressive drugs; novel agents that specifically interfere with the underlying pathophysiologic mechanisms of lupus nephritis are needed to improve disease outcomes enrichment of neutrophil transcripts was detected during the development of active nephritis and distinct signatures in response to treatment were identified based on different pathological subtypes, enabling stratification of the patients into seven groups according to disease activity and the underlying molecular signature 14. This study indicates that the pathologic classification of lupus nephritis could be improved by incorporating immunopathological and/or molecular markers of disease processes with the currently used histological parameters 15,16, a suggestion that is supported by other researchers 17,18. A good association between clinical indices and pathological features is needed to evaluate the treatment response of patients with lupus nephritis 19. In this Review, we summarize potential approaches by which the ISN/RPS classification system could be improved through consideration of underlying disease processes characterized by the presence of glomerular crescents, podocyte injury, tubulointerstitial lesions and vascular injury. Particular emphasis is given to discussion of the potential clinical relevance of histopathologic findings in predicting disease outcome and in facilitating the choice of therapeutic interventions. Classification of lupus nephritis The value of renal biopsy Renal biopsy is crucial to the diagnosis and management of lupus nephritis. The KDIGO, ACR and EULAR/ERA EDTA guidelines recommend that a renal biopsy is performed to confirm the diagnosis, assess disease activity and/or chronicity and guide treatment 1 3. A repeat renal biopsy should be considered in some circumstances such as in patients who relapse, or to reconcile discordance among clinical parameters such as persisting immunological activity despite complete or partial remission of proteinuria or renal function 3,4,20. In general, renal biopsy should be considered when a patient with SLE develops proteinuria (>500 mg per day or >3+ on urine dipstick), active urine sediments or evidence of renal insufficiency (defined by an estimated glomerular filtration rate <60 ml/min/1.73 m 2 ). Some clinical analyses have also demonstrated histological variants in patients with lupus nephritis and acute kidney injury (AKI), which might influence clinical decision making 1 3. For example, a 2011 analysis of a large cohort of patients with lupus nephritis found evidence of pathological changes such as endocapillary proliferative glomerulonephritis, extracapillary proliferative glomerulonephritis (also known as crescentic glomerulonephritis), membranoproliferative glomerulonephritis, thrombotic microangiopathy and acute tubular interstitial nephritis in renal biopsy samples from patients with AKI 21. More importantly, the different renal histopathologic injuries associated closely with clinical implications and long-term outcomes beyond serum creatinine level, highlighting the importance of renal biopsy. The above clinical studies illustrate the value of renal biopsy in providing insights into disease processes that are not apparent by clinical or laboratory data alone. A valuable histopathologic classification system for lupus nephritis should have certain features: it should have clear and precise definitions of disease pathology, have reproducible criteria with low intraobserver and interobserver variation, and should provide prognostic information for response to therapy and renal outcomes. The 2003 ISN/RPS classification The aim of the 2003 ISN/RPS classification of lupus nephritis 4 was to accommodate new insights into pathologic lesions and pathogenesis of lupus nephritis and eliminate inconsistencies and ambiguities present in the 1995 WHO classification 22. Updates included a new definition for class I lupus nephritis, an emphasis on the diagnosis of combinations of membranous and proliferative glomerulonephritis (class III and V, or class IV and V), the separation of class IV lupus nephritis into segmental (IV S) and global (IV G) variants, and clearer definitions for all classes 4 (BOX 1). Since its publication multiple validation studies have compared the ISN/RPS classification with the older WHO classification. One comparative study reported higher reproducibility using the ISN/RPS classification, but noted that the interobserver agreement was far from ideal and that reproducibility of the assessment of activity and chronicity scores was disappointing 23. Similarly, a 2008 study reported that intraobserver and inter observer agreement was higher using the ISN/RPS classification than the WHO classification, although interobserver agreement was found to be low for both classification systems 24. Studies have also evaluated the prognostic value of the 2003 classification system; for instance, one retrospective study of long-term outcomes of 60 patients with lupus nephritis reported good correlation between disease classification according to the ISN/RPS system and longterm renal outcomes in terms of predicting response to therapy, and progression to ESRD and/or death 25. Many other studies since 2005 have further documented the importance of renal biopsy and the ISN/RPS system in managing patients with lupus nephritis 19,26 32. Although the ISN/RPS classification is an improvement on previous systems and has been widely accepted by renal pathologists and clinicians, it relies heavily on pathology as observed by light microscopy, and not the 484 AUGUST 2017 VOLUME 13 www.nature.com/nrneph

Box 1 The 2003 ISN/RPS lupus nephritis classification system 4 Class I: Minimal mesangial lupus nephritis Normal glomeruli by light microscopy, but mesangial immune deposits by immunofluorescence. Class II: Mesangial proliferative lupus nephritis Purely mesangial hypercellularity of any degree or mesangial matrix expansion by light microscopy, with mesangial immune deposits. A few isolated subepithelial or subendothelial deposits may be visible by immunofluorescence or electron microscopy, but not by light microscopy. Class III: Focal lupus nephritis * Active or inactive focal, segmental or global endocapillary or extracapillary glomerulonephritis involving <50% of all glomeruli, typically with focal subendothelial immune deposits, with or without mesangial alterations. Class IV: Diffuse lupus nephritis Active or inactive diffuse, segmental or global endocapillary or extracapillary glomerulonephritis involving 50% of all glomeruli, typically with diffuse subendothelial immune deposits, with or without mesangial alterations. This class is divided into diffuse segmental (IV S) lupus nephritis when 50% of the involved glomeruli have segmental lesions, and diffuse global (IV G) lupus nephritis when 50% of the involved glomeruli have global lesions. Segmental is defined as a glomerular lesion that involves less than half of the glomerular tuft. This class includes cases with diffuse wire loop deposits but with little or no glomerular proliferation. Class V: Membranous lupus nephritis Global or segmental subepithelial immune deposits or their morphologic sequelae by light microscopy and by immunofluorescence or electron microscopy, with or without mesangial alterations. Class V lupus nephritis may occur in combination with class III or IV in which case both will be diagnosed. Class V lupus nephritis may show advanced sclerosis. Class VI: Advanced sclerotic lupus nephritis 90% of glomeruli globally sclerosed without residual activity. Indicate and grade (mild, moderate, severe) tubular atrophy, interstitial inflammation and fibrosis, severity of arteriosclerosis or other vascular lesions. *Indicate the proportion of glomeruli with active and with sclerotic lesions. Indicate the proportion of glomeruli with active and with sclerotic lesions, and the proportion of glomeruli with fibrinoid necrosis and/or cellular crescents. Modified with permission from the International Society of Nephrology Weening, J. J. et al. Kidney Int. 65, 521 530 (2004). Segmental lesions A lesion that involves less than half the glomerular tuft. Global lesions A glomerular lesion that affects more than 50% of the glomerular tuft. underlying disease pathophysiology. Although small studies are available, whether the classification system can accurately predict clinical outcomes and therapeutic responses when combined with clinical and laboratory data has not been thoroughly assessed 17. Prospective, evidence-based validation studies similar to those used to validate the Oxford MEST system for IgA nephropathy 33 41 are still needed. Proposed modifications In 2015, a group of nephropathologists highlighted some difficulties with the ISN/RPS classification, mostly relating to uncertainties in the evaluation of particular disease classes and inconsistencies in the definitions of histologic parameters 16. The researchers also suggested ways in which the classification could be improved, in particular, by including thrombotic microangiopathy and glomerular crescents in the classification system because these lesions might adversely affect renal outcomes 16. Other researchers have suggested incorporating components of the complement system, such as C1q, C4d, C3 and the membrane attack complex, in the classification 17,18, given evidence for the involvement of complement in disease pathogenesis. Here, we outline some of the key issues with the current classification system and suggest additional lesions that should be considered in any new classification system of lupus nephritis. Cut-off thresholds for segmental versus global diffuse nephritis. The ISN/RPS classification divides class IV lupus nephritis into IV S and IV G subclasses, depending on whether the glomerular lesions are primarily segmental or global 4. This subclassification of class IV lupus nephritis was initially proposed on the basis of findings showing that diffuse segmental lesions had more vasculitic like features than global lesions and were associated with a worse renal prognosis 42. However, subsequent validation studies that have compared features of IV S with IV G have failed to show statistically significant differences in renal outcomes between the two subclasses 24,25,43 46, a finding supported by a 2012 meta-analysis of eight studies 47. The subclassification of ISN/RPS class IV nephritis, which includes all biopsy samples with 50% glomerular involvement, into IV S and IV G lesions differs from the original 1982 WHO classification of lupus nephritis, which classified biopsy samples according to the type of lesion: those with severe segmental glomerulonephritis with 50% glomerular involvement were classed as WHO class III 50%, and those with diffuse global glomerulonephritis were classed as WHO class IV 48. A 2008 study in which biopsy samples with WHO class IV or class III >50% lupus nephritis were reclassified according to ISN/RPS criteria led to the reclassification of 39 WHO class IV samples as ISN/RPS class IV-G 46. Of the 44 samples originally classified as WHO class III >50%, 22 were reclassified as ISN/RPS IV S. The remaining 22 were considered to be ISN/RPS class IV G lupus nephritis, but were categorized separately as class IV query (IV Q). Of the 61 patients with class IV G, renal survival was worse for those with class IV Q (that is, those who had been reclassified from the WHO class III >50% category) than for those originally classified as WHO class IV (33% versus 77% at 10 years, P = 0.001). 10 year renal survival was almost twice as high for patients in class IV S than for those with class IV Q, but did not quite reach statistical significance (62% versus 33%, P = 0.057). The researchers concluded that although the ISN/RPS IV S and IV G categories were created to be analogous to severe segmental and diffuse glomerulonephritis in the WHO classification and are considered to be equivalent by some pathologists, clearly they are not. Grouping patients who were originally classified as having WHO class III >50% (that is, those with IV Q) together with those with the best prognosis (WHO class IV) within subclass IV G, obscures prognostic and pathogenic differences, and explains the lack of difference in outcomes between ISN/RPS IV S and IV G lesions in the studies mentioned above. The threshold of 50% glomerular involvement to define classes III and IV and the use of segmental and global lesions to subdivide class IV lupus nephritis in the NATURE REVIEWS NEPHROLOGY VOLUME 13 AUGUST 2017 485

Pauci-immune A pattern associated with minimal evidence of staining for immunoglobulins on glomeruli by immunofluorescence. ISN/RPS system were predominantly selected to bring the classification in line with traditional pathology terminology, rather than based on actual data relating to the clinical relevance of these features; however, some studies published since publication of the 2003 ISN/RPS classification support the notion that pathological differences exist between IV G and IV S lesions. A 2005 study demonstrated that class IV S lesions had more glomerular fibrinoid necrosis and fewer immune deposits than IV G lesions, whereas IV G lesions behaved as a typical immune complex-mediated glomerulonephritis, suggesting pathogenic differences 44. A study of autoantibodies supports this hypothesis. We demonstrated a higher prevalence of anti-neutrophil cytoplasmic antibodies (ANCA) in sera of patients with class IV S lupus nephritis than in those with class IV G nephritis and a higher prevalence of anti C1q IgG1 and IgG3 subclasses in sera of patients with class IV G nephritis 45. Together, these data suggest that some forms of lupus nephritis might resemble ANCA-associated pauci- immune necrotizing and crescentic glomerulonephritis (as is discussed in further detail below). However, studies 27,43,44 have also shown that class IV S and IV G lupus nephritis can transform between each other, potentially challenging the concept that different pathogenic mechanisms underlie the two subclasses. Re evaluation of the ISN/RPS classification should consider a more evidence-based approach to define cut-off thresholds for focal versus diffuse lesions as well as the pathophysiological and clinical relevance of segmental versus global forms of lupus nephritis. Crescentic lupus nephritis in class IV G. Rapidly progressive glomerulonephritis (RPGN) is characterized by pathological features of crescentic glomerulonephritis and is a severe clinical syndrome associated with progressive loss of renal function. Crescents can be observed in association with a number of different renal pathologies, and are fairly common in biopsy samples from patients with lupus nephritis 49. In one cohort of 62 patients with RPGN and biopsy-proven crescentic glomerulonephritis, 32 patients (51.6%) had lupus nephritis 50. The clinical relevance of these crescents is, however, unclear as no studies have examined the clinical relevance of extensive (or true) crescentic lupus nephritis (which we define as the presence of crescents in 50% of glomeruli). In a cohort of 152 Chinese patients with class IV G lupus nephritis, we found that patients with extensive crescents (affecting 50% of glomeruli) have worse outcomes than those of other patients with class IV G lupus nephritis 51. Although biopsy samples with true crescentic glomerulonephritis are classified as diffuse class IV lupus nephritis according to the ISN/RPS schema, in our histopathological evaluations 51, we found that biopsy samples with crescentic IV G lesions have significantly higher levels of interstitial inflammation, interstitial fibrosis and necrosis than IV G biopsy samples without crescentic glomerulonephritis 51. More interestingly, we identified lower levels of IgG, IgA, IgM, C3, C1q and fibrin in crescentic biopsy samples than in non-crescentic biopsy samples, suggesting further similarities between crescentic IV G samples and ANCA vasculitis. These findings suggest that differences in underlying pathogenic mechanisms and disease outcomes might exist within the same subclass of lupus nephritis, and leads us to recommend that the presence of true crescentic glomerulonephritis (defined as the presence of crescents in 50% of glomeruli) is included as part of a new lupus nephritis classification system (FIG. 1). Podocyte injury. Renal biopsy samples from patients with lupus nephritis can demonstrate injury to nearly any cell type, including the podocyte. Evidence for the clinical relevance of podocyte injury comes from several studies that have described patients with lupus nephritis, nephrotic-range proteinuria and extensive effacement of podocyte foot processes, without evidence of immune complex deposition in peripheral glomerular capillaries or endocapillary proliferation 52,53. Morphologically, these podocyte lesions resemble minimal change disease or focal segmental glomerulosclerosis (FSGS), and researchers involved in these studies have proposed using the term lupus podocytopathy to describe this entity 54 56. A retrospective study of 19 patients with collapsing glomerulopathy and SLE or SLE-like disease reported massive proteinuria in 95% of patients; segmental and/or global collapsing glomerulopathy was seen in 11 77% of glomeruli, and extensive foot process effacement was seen in 82% of patients. Seven of 13 patients with follow up data progressed to ESRD 57. A large Chinese study of 3,750 biopsy samples from patients with lupus nephritis identified 50 cases (1.33%) of lupus podocytopathy, including 13 cases of minimal change disease, 28 instances with evidence of mesangial proliferation, and nine cases of FSGS 58. Extensive foot process effacement was evident in all 50 biopsy samples and electron-dense deposits present exclusively in mesangial cells were evident in 47 biopsy samples. All 50 patients presented with nephrotic syndrome, and 34% had AKI. Most notably, patients with FSGS had a higher risk of AKI and severe tubulointerstitial injury and a lower remission rate than patients with minimal change disease or mesangial proliferation, although no patient died or developed ESRD over a median follow up of 62 months. On the basis of the differences in AKI incidence, severity of tubular injury and response to treatment, the researchers proposed that lupus podocytopathy is divided into two different subtypes: minimal change/mesangial proliferation and FSGS 58. A 2014 study 59 of renal biopsy samples from patients with lupus nephritis and proteinuria found that immuno histochemical staining for podocyte markers was generally preserved in patients with membranous (class V) lupus nephritis. By contrast, <10% of proliferative forms of lupus nephritis (class III or IV) showed preserved expression of podocyte markers, suggesting the presence of structural podocyte damage in proliferative forms of lupus nephritis. This loss of podocyte integrity in patients with proliferative forms of lupus nephritis might account for the worse prognosis of these patients compared to those with membranous 486 AUGUST 2017 VOLUME 13 www.nature.com/nrneph

lupus nephritis, despite similar levels of proteinuria, at least in the patients in this cohort. Our study measured foot process width (FPW) to assess the podocyte foot process effacement and studied correlations between podocyte damage and clinico-pathological parameters in 202 Chinese patients with lupus nephritis. FPW correlated with proteinuria, and a threshold FPW was identified that could differentiate nephrotic proteinuria from non-nephrotic proteinuria 60. On the basis of these studies, we suggest that the presence of lupus podocytopathy, a b Podocyte injury Crescentric glomerulonephritis c Mesangial hypercellularity Podocyte Mesangial cell Endothelial cell d Diffuse lupus nephritis Parietal epithelial cell Bowman capsule IV-G IV-S e Vascular lesions Blood vessel Blood clot Plaque Plaque ICDs f Tubulointerstital lesions ICDs Interstitial inflammation Tubulitis Arteriosclerosis TMA T cell Macrophage Figure 1 Pathological features of lupus nephritis subtypes. a The proliferation of extracapillary epithelial cells and infiltration of inflammatory cells leads to the formation of cellular crescents. Periodic acid silver methenamine staining (PASM) 400. b Podocyte injury and foot process effacement can also occur in lupus nephritis. Original magnification of electron microscope image 6,000. c Mesangial hypercellularity is defined by the proliferation of mesangial cells, matrix expansion and mesangial deposition of immune complexes. Periodic acid Schiff (PAS) staining 400. d Diffuse (class IV) lupus nephritis can be subsclassified into segmental (IV S) and global (IV G) subclasses. Class IV S exhibits segmental fibrinoid necrosis with segmental endocapillary hypercellularity (Masson trichrome staining 400). Class IV G exhibits global endocapillary and mesangial hypercellularity with infiltration of inflammatory cells (PAS staining 400). e Vascular lesions include the presence of immune complex deposits (ICDs) in the vasculature, demonstrated by granular staining for IgG along arteriolar walls (immunofluorescence 400), arteriosclerosis characterized by arterial intimal fibrosis (PASM staining 400), and thrombotic microangiopathy (TMA) with arteriolar thrombi and swelling of endothelial cells (PASM staining 400). f Tubulointerstitial lesions include ICDs in the tubular basement membrane demonstrated by granular staining for IgG (immunofluorescence 400), interstitial inflammation with severe interstitial infiltration of mononuclear inflammatory cells (haematoxylin and eosin staining 400), and tubulitis with infiltration of lymphocytes between tubular epithelial cells (PAS staining 400). NATURE REVIEWS NEPHROLOGY VOLUME 13 AUGUST 2017 487

Renal flares Defined by an increase in urine sediment, protein excretion, and serum creatinine value from baseline. defined by the presence of nephrotic-range protein uria and podocyte lesions, might represent an extreme form of podocyte damage. We propose that the degree of podocyte injury in different types of lupus nephritis should be given more attention (FIG. 1), as the presence of specific injuries might have therapeutic implications. Tubulointerstitial lesions. The ISN/RPS classification focuses primarily on glomerular pathology, although it indicates that the biopsy report should include a description of tubulointerstitial injuries. An increasing body of evidence, however, suggests that the importance of tubulo interstitial damage requires greater emphasis than is given in the current classification. We investigated the impact of tubulointerstitial damage in a multicentre cohort of 313 lupus nephritis patients from North China, and found that glomerular and tubulointerstitial lesions do not always coexist, and that tubulointerstitial inflammation, tubular atrophy, and interstitial fibrosis were independent risk factors for renal outcomes 61. In another cohort of 68 patients from the USA, Hsieh et al. found that 72% of biopsy samples had evidence of moderate or severe tubulointerstitial inflammation 62. The researchers found that severity of tubulointerstitial inflammation, but not severity of glomerular injury, identified patients at greatest risk of ESRD. Similar results have been reported by several other studies 63 67. Hsieh et al. also suggested that patients with lupus nephritis and tubulointerstitial inflammation should receive intensive therapeutic intervention 62. To this end, a 2013 cohort study of 73 patients with lupus nephritis identified a correlation between interstitial inflammation at repeat biopsy and renal survival (P = 0.005), despite the absence of a correlation between interstitial inflammation at the baseline biopsy and worsening of renal function (P = 0.17) 63. Furthermore, resolution of interstitial inflammation in lupus nephritis as evident on the repeat biopsy correlated with a favourable outcome in patients with interstitial inflammation at baseline (P = 0.047). Thus, thorough assessment of tubulointerstitial damage provides information about disease prognosis and should be included in the lupus nephritis classification system (FIG. 1). Renal vascular lesions. The ISN/RPS classification requests that pathologists indicate the severity of arteriosclerosis or other vascular lesions, but overall pays very little attention to vascular lesions, although these are common in biopsy samples from patients with lupus nephritis 68. Two studies from 2012 and 2013 showed that adding an assessment of vascular damage to the ISN/RPS classification increased its prognostic value 69,70. At least five renal vascular lesions are commonly observed in biopsy samples from patients with lupus nephritis: vascular immune complex deposits (ICDs), arteriosclerosis, thrombotic microangiopathy (TMA), non-inflammatory necrotizing vasculopathy and true renal vasculitis 68 (FIG. 1). In a cohort of 341 patients with stable lupus nephritis, we found renal vascular lesions in 81.8% of samples, including vascular ICDs (74.3%), arteriosclerosis (24.0%), TMA (17.6%), necrotizing vasculopathy (3.8%), and true renal vasculitis (0.6%); 37.6% of the patients presented with more than two types of vascular lesions 69. Patients with TMA had the poorest renal outcome; surprisingly, those with purely vascular ICDs (that is, with no other vascular lesion) had the second-to worst outcomes, although vascular ICDs are traditionally thought to be a benign phenotype 69. A series of other studies indicate that renal vascular lesions are closely associated with clinical disease activity and renal outcomes, and that renal TMA is an independent risk factor for long-term renal outcomes in patients with lupus nephritis 69,71 74. The presence of TMA lesions in renal biopsy samples from patients with lupus nephritis might not always be associated with haematological features of microangiopathy, such as haemolytic anaemia, schistocytosis and thrombocytopaenia). Thus, we propose that concise descriptions and grades of different renal vascular lesions are added to the classification system. In another study of 79 patients with biopsy-proven lupus nephritis, we found that 50 (63.3%) had arteriosclerosis lesions on renal biopsy whereas 29 patients had no evidence of vascular changes 75. Patients with arteriosclerosis presented with more severe echocardiographic indices, including larger left atrial diameter, left ventricular end-diastolic diameter and interventricular septum thickness, than patients without vascular changes, highlighting the clinical relevance of vascular lesions 75. Disease pathogenesis and treatment Despite the use of histopathology to guide therapeutic decisions, the morbidity and mortality of lupus nephritis remains high. The current approach to the management of lupus nephritis is based on steroids and other nonspecific immunosuppressive drugs. Dysregulation of the immune system is fundamental to the pathogenesis of lupus nephritis, and targeting multiple aspects of the immune response through the combined use of multiple immunosuppressants (multitarget therapy) proved superior to a standard regimen of steroids and cyclophosphamide (intravenous or oral) as induction therapy for lupus nephritis in a 2015 Chinese study 76. Nonselective immuno suppressants are, however, associated with potentially life-threatening complications including an increased risk of infection. In addition, they are often associated with incomplete renal remission and a high rate of renal flares, which might be more prominent in patients with features such as vasculopathy, crescentic glomerulonephritis, and tubulointerstitial lesions 1 3. Thus, treatment of lupus nephritis remains a challenge and we propose that it should target the pathogenesis of the disease, as determined by the assessment of the histopathological disease phenotypes. Novel drugs that interfere specifically with the pathogenic mechanisms of lupus nephritis, with fewer adverse effects and higher efficacy than nonspecific immunosuppressive agents, are needed. Here, we briefly discuss traditional therapeutic approaches guided by the ISN/RPS classification system, and how consideration of additional lesions might facilitate the identification of more targeted therapies 77 83. 488 AUGUST 2017 VOLUME 13 www.nature.com/nrneph

Tertiary lymphoid organs Highly organized lymph node-like structures. CD19 Autoantigens Endothelial cell GBM Podocyte Complement Epithelial cell Anti-dsDNA antibodies BAFF BAFF-R CD20 Plasma cell Autoantibodies B cell TNF BCL-2 Current therapeutic approaches As mentioned earlier, the three major guidelines, ACR, KDIGO and EULAR/ERA EDTA 1 3, base their recommendations primarily on the underlying histologic lesions as defined by the ISN/RPS classification. Although some differences exist in the recommendations CD40 CD22 C4d MHC class II B7 Immune complexes T cell TCR Macrophage IFNα Dendritic cell CD40L CD28 T cell CTLA4 Tubular epithelial cell HMGB1 histones Neutrophil T H 17 IL-6 Interstitial endothelial cell IL-23 TLR Immunostimulatory nucleic acids T H 17 T reg IL-17 Autoantigenspecific T cell Antiphospholipid antibodies Blood vessel Figure 2 Cell-mediated disease mechanisms of lupus nephritis. The sensing of immunostimulatory nucleic acids by dendritic cells drives B lymphocyte and T lymphocyte activation and the production of autoantibodies and autoantigenspecific T cells, leading to glomerular endothelium, podocyte, tubulointerstitial and vascular injury. Specific leukocyte subsets, including IL 17 producing T helper type 17 (T H 17) cells, drive inflammation and contribute to renal immunopathology. B cellactivating factor (BAFF) might increase the generation of new autoreactive B cells, inhibitb cell apoptosis and stimulate the differentiation of B cells into immunoglobulinproducing plasma cells. Infiltrating leukocytes might also provide a source of nuclear antigens and cytokines, such as interferon (IFN) α, tumour necrosis factor (TNF), IL 1 and IL 6. High mobility group box 1 (HMGB1), biglycan and histones released from immune cells can bind Toll-like receptors (TLRs) expressed by endothelial cells, leading to an inflammatory response. Inaddition, anti-double-stranded DNA (anti-dsdna) antibodies can induce fibronectin secretion in proximal renal tubular epithelial cells, leading to TGFβ activation and collagen synthesis. Activation of the apoptosis regulator BCL 2 might also contribute to the development of tubulointerstital inflammation, whereas antiphospholipid antibody-induced thrombosis might contribute to the development of interstitial vascular inflammation. GBM, glomerular basement membrane; TCR, T cell receptor; T reg, regulatory T cell. of these three guidelines, the general recommendations are outlined below along with brief discussion of the underlying pathogenesis of the different subtypes. Class I and Class II lupus nephritis. Class I and Class II lupus nephritis result from immune complexes that form within the mesangium by binding of antibodies to autoantigens or following the release of antigens from mesangial cells 84. The deposition of immune complexes in the mesangium and expansion of the mesangial matrix do not usually cause irreversible glomerulosclerosis, possibly owing to the regenerative capacity of mesangial cells 84. Thus, Class I and Class II lupus nephritis might imply a more favourable prognosis than other classes of lupus nephritis, and aggressive immunosuppressive therapy is not indicated unless the proteinuria is >3 g per day or extrarenal clinical manifestations are prominent. Class III and IV lupus nephritis. Class III and Class IV lupus nephritis result from the deposition of immune complexes in the subendothelial space of the glomerular capillaries, which causes endothelial cell activation via the same complement Fc receptor and nucleic acid sensing mechanisms that operate in mesangial cells (FIG. 2). The development of tertiary lymphoid organs in the tubulointerstitium leads to clonal expansion and ongoing somatic hypermutation of B cells and plasma cells that can lead to intrarenal autoantibody production, local inflammation, and tissue pathology, implying that B cell-targeted therapies with heightened tissue penetration might be beneficial for patients with class III and IV lupus nephritis 85,86. The recommended therapy for class III and IV lupus nephritis includes a sequence of induction and main tenance phases. Induction therapy usually involves a combination of high-dose parenteral and oral corticosteroids with either cyclophosphamide (oral or intravenous) or mycophenolate mofetil (MMF), whereas maintenance therapy involves azathioprine or MMF and low-dose oral corticosteroids. Studies have reported average remission (complete and partial) rates of nearly 70% after 6 months initial therapy 10,87,88, but higher remission rates can be achieved with more prolonged treatment. A 2012 systematic review found that MMF in combination with corticosteroids was as effective as cyclophosphamide plus corticosteroids in achieving remission in patients with proliferative lupus nephritis. However, these standard treatments are all associated with considerable adverse effects, therapy failure and relapse, highlighting the need for more effective, less toxic options. Interest is growing in the potential use of calcineurin inhibitors (CNIs) in severe class III and IV lupus nephritis 1 3. A 2015 meta-analysis reported that tacrolimus was more effective than cyclophosphamide at inducing complete remission in patients with class III and IV lupus nephritis, but was not more effective than MMF 89. However, randomized clinical trials of tacrolimus in lupus nephritis have so far been small 90,91. Extended follow- up of the CYCLOFA-LUNE trial (median 7.7 years) indicated that the CNI ciclosporin NATURE REVIEWS NEPHROLOGY VOLUME 13 AUGUST 2017 489

Nonselective immunosuppressants CNIs Epithelial cell Tabalumab Belimumab Podocyte Complement Vascular rarefaction Loss of capillaries. CD19 CD20 BAFF BAFF-R Plasma cell Autoantibodies C4d Infliximab Anti-IFN-γ T cell B cell CD22 Rituximab Eculizumab/ CCX-168 TNF IFNα CD40 Autoantigens MHC class II B7 TCR Epratuzumab CD40L CD28 Neutrophil T cell CTLA4 Abatacept Immune complexes T H 17 Autoantigenspecific T cell BCL-2 IL-23 IL-6 Sirukumab Endothelial cell GBM IL-17 ABT-199 Anti-IL-23 Macrophage Figure 3 Therapeutic targets in systemic lupus erythematosus (SLE) and lupus nephritis. Greater understanding of the pathogenic processes underlying lupus nephritis has led to the identification of new therapeutic targets. B cells have a critical role in the pathogenesis of SLE, and B cell depletion therapy therefore remains an attractive therapeutic option. The anti CD20 antibody, rituximab, can deplete autoreactive B cells and thereby attenuate the production of autoantibodies involved in disease manifestations. The B cell-activating factor (BAFF)-neutralizing antibody, belimumab, was shown to reduce renal flares and proteinuria in patients with SLE. Other promising B cell-depleting monoclonal antibodies, such as tabalumab and epratuzumab, require further testing. Abatacept is a co-stimulatory inhibitor that targets B7 1 (CD80) on the surface of dendritic cells or B cells, and blocks co-stimulation of CD28 on T cells. Inhibitors of the complement system, such as eculizumab and CCX 168 could have therapeutic value in patients with concurrent thrombotic microangiopathy or anti-neutrophil cytoplasmic antibody-associated vasculitis. Calcineurin inhibitors (CNIs) have immune modulatory effects but also direct effects on podocytes, and might be useful in patients with lupus-associated podocyte injury. Inhibition of the apoptosis regulator BCL 2 prevented the development of tubulointerstital inflammation in a mouse model of lupus nephritis. IL 23 activation might have a role in the development of glomerular crescents, suggesting that anti IL 23 antibodies might be beneficial. Other promising therapies that need further exploration include infliximab, which targets the inflammatory tumour necrosis factor (TNF), sirukumab, which targets inflammatory IL 6, and monoclonal antibodies to interferon (IFN). GBM, glomerular basement membrane; T H 17, T helper type 17 cell. achieved similar results to those achieved with cyclophosphamide in patients with class III and IV lupus nephritis 11. Although these small trials did not find any obvious evidence of acute or chronic nephrotoxicity with use of tacrolimus or ciclosporin, concerns still exist with regard to the use of CNIs in patients with lupus nephritis, especially in those with renal vascular lesions such as TMA, which has been attributed to the use of drugs such as CNIs. Large-scale, multicentre clinical trials are needed to further confirm the safety and efficacy of CNIs in lupus nephritis: the ongoing AURA LV study 92 will hopefully provide insights into the safety and efficacy of the new CNI, voclosporin in patients with class III and IV lupus nephritis 93. Of note, and as discussed in further detail below, lupus nephritis remission is largely defined on the basis of a decline in proteinuria, and all CNIs act directly on podocytes to reduce proteinuria independent of their immunosuppressive effects 2 (FIG. 3). Thus, whether the disease remissions observed with CNI therapy result from true remission of the immunological disease processes or are the result of the antiproteinuric effects of CNIs is not clear. Repeat renal biopsies are needed to make this distinction. Class V lupus nephritis. Class V lupus nephritis (also known as membranous lupus nephritis) is caused by the deposition of immune complexes in the subepithelial compartment of the glomerular tuft, which leads to complement activation and ultimately podocyte injury 94 (FIG. 2). Patients with class V lupus nephritis are usually treated with antiproteinuric and antihypertensive medications such as renin angiotensin system blockers, and can receive corticosteroids and immunosuppressants as required depending on the presence of persistent nephrotic proteinuria or extrarenal manifestations of SLE. As noted above, CNIs can reduce proteinuria through direct actions on podocytes. CNIs inhibit dephosphorylation and degradation of the actin-associated podocyte protein, synaptopodin, and thus help to stabilize the podocyte actin cytoskeleton. Preliminary data indicate that combination CNI and corticosteroid therapy is effective in the treatment of class V lupus nephritis, especially in reducing proteinuria 89,94 97 ; however, the initiation and duration of CNI therapy remains a matter of debate, as relapses are very common once CNIs are stopped 2. Moreover, CNIs are associated with an increased risk of TMA, indicating that alternative therapeutic approaches are required. Class VI lupus nephritis. The development of sclerotic lesions in lupus nephritis results mainly from insufficient cellular regeneration following the injury and loss of epithelial cells 98, and leads to irreversible glomerulosclerosis and nephron loss. Vascular rarefaction, tubulointerstitial ischaemia, and inflammation all contribute to the progression of renal fibrosis and sclerosis, which is usually associated with a progressive decline in glomerular filtration rate and ultimately the development of ESRD. Patients with class VI lupus nephritis should be treated with renin angiotensin system blockers. Corticosteroids and immunosuppressive agents should be used only as dictated by the presence of extrarenal manifestations. 490 AUGUST 2017 VOLUME 13 www.nature.com/nrneph

Perinuclear-ANCA A category of ANCA originally described on the basis of their immunofluorescence patterns around the nucleus. In situ adaptive immunity A process in which infiltrate organizes into wellcircumscribed aggregates of B cells and T cells (in germinal centres). These germinal centres contain follicular dendritic cells and enable intrarenal B cells to undergo clonal expansion and somatic hypermutation. Plasmablast foci Plasmablast aggregates in an organ. Treatments guided by additional lesions As described above, the presence of features beyond those included in the current classification system of lupus nephritis including the presence of glomerular crescents, podocyte injury, vasculopathy and tubulointerstitial changes provide insights into the underlying pathogenic mechanisms of lupus nephritis and might aid the identification of new therapeutic targets. Here we discuss the underlying pathogenic mechanisms of each of these lesions and how they might be targeted therapeutically. Crescentic lupus nephritis. The mechanisms that drive the formation of crescents in lupus nephritis are not fully understood. A 2009 study showed that depletion of immature invariant natural killer T cells accentuated disease severity in a model of crescentic glomerulonephritis 99. Thus, targeting CX3CL1/CX3CR1 pathway, which regulates the activation of invariant natural killer T cells, might represent a promising treatment approach with fewer adverse effects than general immunosuppressive therapies (FIG. 3). Animal models studies have also demonstrated that the IL 23/IL 17 pathway contributes to renal injury in experimental glomerulonephritis 100 ; thus, blocking IL 23 production and/or the subsequent recruitment of IL 17A producing effector γδt cellreceptor-expressing CD3 + CD4 CD8 NK1.1 T cells, which are thought to have a role in crescent formation, might also prove beneficial in limiting glomerular damage in part through limiting the recruitment of myeloid cells. Cell lineage-tracing studies indicate that podocytes and parietal epithelial cells also participate in crescent formation 101,102. Importantly, in one study of 10 patients with lupus nephritis and glomerular crescents, all patients were posi tive for serum ANCA and five had antimyeloperoxidase (MPO) antibodies 103. Similarly, another study reported that 10 of 33 Chinese patients with lupus nephritis and crescents in 50% glomeruli were ANCApositive and seven had MPO-ANCA 51. In fact, one study reported a correlation between the serum ANCA status and the presence of crescents in patients with lupus nephritis 104, and others have reported an association between atypical perinuclear ANCA, such as anti-cathepsin G antibodies, and the development of crescentic lupus nephritis 105,106. These findings provide strong evidence for a role of ANCA in crescentic lupus nephritis, and suggest that therapeutic approaches that are used to treat vasculitis, including intensive plasmapheresis and the new anti C5aR antibody, should be assessed in this subgroup of patients 51,103,107,108 (FIGS 3,4). Podocyte injury. Proteinuria is a common feature of lupus nephritis and reflects podocyte injury. Furthermore, the presence of nephrotic-range proteinuria in patients with lupus membranous nephropathy might represent the presence of podocyte dysfunction. Decreased expression of the slit diaphragm proteins, nephrin and podocin, in kidneys of mice and patients with lupus nephritis suggests that immune-mediated processes contribute to disruption of the slit diaphragm 109. In patients with lupus nephritis and severe podocyte effacement, we found that those who received CNIs had better remission rates and long-term renal outcomes than those treated with other regimens, suggesting that therapies that facilitate podocyte stability might be beneficial in a subgroup of patients with lupus nephritis 60 (FIGS 3,4). Tubulointerstitial lesions. The mechanisms underlying the development of tubulointerstitial lesions in lupus nephritis remain unclear, and thus no specific treatment options exist. Severe tubulointerstitial inflammation is associated with in situ adaptive immunity and the development of tertiary lymphoid organ like structures with aggregates of T cells and B cells, plasmablast foci, and germinal centres 110. Vimentin, an antigenic feature of inflammation, has been identified as a dominant autoantigen that drives clonal B-cell selection in lupus tubulointerstitial inflammation, suggesting that vimentin might represent a therapeutic target for tubulointerstitial lesions in lupus nephritis 111. Anti-double-stranded DNA (anti-dsdna) antibodies can induce fibronectin secretion in proximal renal tubular epithelial cells, leading to TGFβ activation and collagen synthesis, again highlighting potential therapeutic targets 112. Thus, explorations into the pathogenesis of tubulointerstitial lesions in lupus nephritis will likely identify targeted therapeutic interventions for future study 113. This point is illustrated by a study that used novel computational approaches to quantify the expression of apoptosis regulators in infiltrating lymphocytes in biopsy samples from patients with lupus nephritis and tubulointerstital inflammation. The apoptosis regulator, BCL 2, was frequently expressed in these biopsy samples, indicating a role for dysregulated apoptosis in the immune response. Moreover, treatment of NZB/WF 1 lupus nephritis mice with a selective oral inhibitor of BCL 2 prevented the development of tubulointerstitial inflammation 114. Renal vascular lesions. Vascular lesions are a common feature of SLE, and renal vasculopathy might represent an extreme form of such lesions. Key pathogenic mechanisms include endothelial cell activation and dysfunction, and dysregulation of the immune system, particularly through immune complex-induced vascular inflammation and antiphospholipid antibody-induced thrombosis 74,115 (FIG. 2). In addition to antiphospholipid antibodies, renal TMA in lupus nephritis can be caused by thrombotic thrombocytopenic purpura, malignant hypertension, pregnancy, scleroderma or drugs. Interestingly, studies have demonstrated a strong relationship between the intensity of glomerular C4d staining and the presence of renal microthrombi in lupus nephritis 116,117. This finding suggests a role for the classical complement pathway in lupus nephritis- associated renal TMA, similar to the mechanisms involved in antiphospholipid antibody-induced thrombosis and humoral rejection in transplanted kidneys 116,117. We have demonstrated that the presence of glomerular C4d deposits and decreased serum levels of the complement control protein, factor H are associated with poor renal outcomes in NATURE REVIEWS NEPHROLOGY VOLUME 13 AUGUST 2017 491