The kidney biopsy in lupus nephritis: time to move beyond histology

The percutaneous kidney biopsy has been the gold standard for the diagnosis of lupus nephritis (LN) since before the original World Health Organization classification of LN in 1974 [1]. The incorporation of immunofluorescence and electron microscopy into kidney biopsy evaluation provided significant insights into the pathogenesis of LN. These and other insights from experimental models and clinical studies were driving forces behind the 2004 International Society of Nephrology (ISN) and Renal Pathology Society (RPS) reclassification of LN histology. An important goal of reclassification was to better align histology to treatment options and prognosis. The ISN/RPS classification introduced the subcategories of global (G) and segmental (S) to proliferative forms of LN and defined composite indices for active and chronic disease to better forecast outcomes [2]. Unfortunately, despite these efforts, very little improvement has occurred in our ability to accurately predict treatment response or determine long-term kidney outcomes by examining histology alone. This may be due, in part, to the fact that the histologic responses of the kidney to injury are limited, whereas the underlying molecular mechanisms of injury are quite heterogeneous in a disease like systemic lupus erythematosus. Much as the addition of immunofluorescence and electron microscopy to light microscopy advanced the field of renal pathology, integration of molecular analysis with kidney histology may better account for the heterogeneity of LN, and ultimately yield a superior classification scheme that will foster targeted therapeutic approaches and provide more reliable prognostic information. Here we consider ‘molecular’ to include gene, transcript and protein expression. In this issue of the journal, Alaiya and colleagues attempt to differentiate Class IV-G from Class IV-S LN through a molecular (proteomic) approach and then correlate the molecular differences with clinical parameters and long-term outcomes. The rationale for their work is to address the controversy of whether Class IV-S and Class IV-G are different entities that require different therapies. Surprisingly, no differences in the proteomes of Class IV-S and Class IV-G were identified. Not unexpectedly, the LN proteome did differ from that of anti-neutrophilic cytoplasmic antibody-associated vasculitis with renal involvement, as well as normal kidney tissue. These findings support previous studies that failed to differentiate clinical outcomes between Class IV-S from Class IV-G [3–5]. This work represents a novel, and we believe necessary approach to interpreting clinical renal pathology in LN and other forms of glomerulonephritis. Although the study did not differentiate Class IV-S from Class IV-G on the basis of renal protein expression, the approach provides insights into the feasibility and challenges of adding molecular-type analyses to the evaluation of kidney pathology in the future. The molecular analysis of kidney biopsies is challenging because the tissue cores are small and then further divided between different fixatives. As in the study by Alaiya and colleagues, frozen tissue has traditionally been necessary for RNA and proteomic studies. However, most of a biopsy core is formalin-fixed and paraffin-embedded (FFPE), thus limiting frozen tissue available for study. To get around this problem, recent protocols have been developed to facilitate molecular analysis of the kidney from FFPE tissue [6–8]. One such example is the use of Nanostring technology (Nanostring Technologies, Seattle, WA) for transcript expression. Nanostring is a sensitive gene expression analysis tool that utilizes molecular ‘barcodes’ which directly hybridize on to target molecules so they can be individually counted and prevents the need for amplification. This technology can be applied to small amounts of tissue and detects hundreds of unique transcripts in a single hybridization reaction [6, 9]. The complex architecture of the kidney also provides a unique challenge to molecular analysis. The kidney is made up of two distinct compartments, glomerular and tubulo-interstitial. In Alaiya's study, whole renal cortex was used for proteomic analysis without differentiation between these compartments. Because the cortex is 90% tubulointerstitium, their proteomic analysis mainly reflects interstitial protein expression; nuances of glomerular protein expression may have been masked. This may explain why the glomerular lesions of Class IV-G and IV-S were found to have similar protein expression profiles. A method using laser capture microdissection (LCM) to isolate the specific compartments, followed by mass spectrometry would allow molecular characterization of the individual compartments. A limitation of tissue proteomics is its inability to capture low molecular weight, low-abundance proteins that often have high biologic activity, such as cytokines and growth factors. Adding transcript analysis to proteomic analysis would yield a more comprehensive molecular picture of LN. In order to attain this comprehensive molecular picture, it will be necessary to apply several molecular techniques to the small amount of kidney tissue available for evaluation. If molecular analysis of kidney biopsies is shown to provide added clinical value, it may become necessary to take an extra tissue core during the biopsy. Despite the challenges described above, data from molecular analysis of kidney biopsies in LN are emerging. LCM was successfully applied to LN kidney biopsies and convincingly demonstrated the molecular heterogeneity of LN [10]. Glomeruli from kidney biopsies of LN and normal controls were isolated by LCM and then analyzed by complementary transcriptional DNA phenotyping [10]. Four gene clusters were significantly activated in LN compared with normal kidney tissue and included type I interferon-inducible genes, fibrosis-related genes, B-cell genes and myelomonocytic gene clusters. Although all patients in this study had Class III or IV LN histology, there was significant patient-to-patient variation in transcript expression. Interestingly, however, glomerular transcript expression was similar within a patient despite intra-glomerular histologic variability. For example, fibrosis-related gene expression was increased in all the glomeruli within an individual kidney regardless of whether or not they all demonstrated glomerulosclerosis [10]. The take away point from these observations is that there is discordance between histology and underlying molecular pathology. Thus only molecular analysis of biopsy tissue will truly provide a gateway to personalized therapeutics in LN. Figure 1 illustrates one approach of how molecular analysis of the kidney biopsy could be utilized at the time of flare to guide treatment and improve outcomes. Using molecular analysis to guide treatment of lupus nephritis: this schema outlines one approach to utilize the molecular analysis of kidney tissue to guide treatment decisions. Patients with clinical evidence of renal flare undergo kidney biopsy as per standard of care. The present: Patients are treated with a standard of care regimen as suggested by kidney pathology with the goal of achieving a complete renal response as defined by clinical metrics. If the patient has an incomplete response, repeat kidney biopsy is done and therapies are adjusted to address residual disease. The future: In the future, we propose that molecular analysis will be performed on kidney biopsy tissue at flare. Biomarkers of treatment response will be identified allowing adjustment of standard of care therapy based on the predicted response. Or, novel therapies that target the specific immune/inflammatory pathways activated in that kidney will be applied in lieu of, or in addition to standard of care treatment. The goal will be to attain complete renal remission, which in this paradigm is defined as achieving both clinical and histologic remission. If either histologic or clinical activity persists, repeat biopsy will again be analyzed at the molecular level. In this situation, metabolomics has been included in the molecular analysis, because after initial immunosuppression, it appears that metabolic pathways play a prominent role in progressive kidney damage, even if immune pathways have been suppressed (see [11]). Therapy will then be tailored to target the active mechanisms so that complete renal remission can be achieved and progressive kidney disease halted. With respect to using the initial kidney biopsy to predict future outcomes or response to therapy, previous studies have shown a poor correlation between clinical markers of LN and renal histology. As an example, among patients considered to have complete renal response by clinical metrics, approximately one-third had an activity index of two or greater [12]. Additionally, 63% of complete responders were also found to have a chronicity index of four or greater on repeat biopsy [12]. Besides demonstrating the discordance between clinical and histologic variables, these findings show that despite aggressive immunosuppression, significant chronic renal damage may continue to occur even in situations where clinical markers would suggest otherwise. A microarray analysis of the kidneys of lupus-prone mice from different stages of disease provides some insights into the histologic-clinical discordance [11]. At onset of proteinuria, there was a significant increase in inflammatory gene expression that improved with treatment and returned to baseline levels in mice with clinical remission. However, in mice with prolonged disease activity the molecular signature was significant for down-regulation of metabolic stress and mitochondrial pathways. These changes were persistent even in mice that eventually achieved clinical remission, suggesting that routinely used clinical markers that reflect active inflammation may miss non-immune mechanisms that are responsible for progressive renal deterioration. Preliminary work from our group examined kidney gene expression patterns in LN, followed by tissue proteomics to identify molecular pathways that differentiate early responders from non-responders [13] (Parikh, ASN 2014). Diagnostic biopsies from proliferative forms of LN were segregated by renal response status after an average of 11 months of standard induction and maintenance immunosuppression. The expression of several hundred immune response genes was evaluated using Nanostring technology [6, 14]. By principle component analysis the samples clustered by response status (complete response, partial response, no response). Genes were differentially expressed in each group compared with normal controls and to each other. Specifically, the interferon-alpha, T- and B-cell signaling and IL-10 pathways were highly activated in flares that ended in no early response, while apoptotic and T-cell regulatory pathways were activated in flares that ended in complete response. These data suggest that the initial molecular profile of a responder differs from a non-responder and that gene expression analysis of diagnostic kidney biopsies may help distinguish responders from non-responders prior to initiating treatment. Alternative therapy may then be considered for those with an unfavorable molecular profile. Because therapeutic targets are often proteins, proteomic evaluation of clinical kidney biopsies should complement gene expression analyses and help elucidate therapeutic targets from activated pathways identified through transcript analysis. We previously compared the proteome of proliferative LN to normal controls using LCM followed by mass spectrometry [15]. LN glomeruli showed increased expression of complement proteins and decreased apoptotic and podocyte proteins compared with normal glomeruli. Additionally, the proteomic analysis showed increased expression of interferon-alpha inducible proteins consistent with what was found by gene expression analysis [15]. This suggests that transcript and proteomic profiling can be used together to more completely characterize molecular pathways active in LN. This may be especially useful for identifying biomarkers and developing novel therapeutics. The present work by Alaiya and colleagues represents an important step toward the incorporation of molecular analysis into the pathologic evaluation of LN. It is now evident that more information resides within the kidney biopsy than what can be appreciated from histology alone. There seems to be discordance between both molecular findings and histology and clinical findings and histology. With the application of novel molecular techniques, we hope to find concordance between the clinical and molecular events that occur in LN. Molecular characterization of renal lesions present in LN will provide clinicians the tools necessary to more accurately classify disease, predict outcomes and appropriately apply novel therapeutics. None declared. (See related article by Alaiya et al. Proteomic analysis of Class IV lupus nephritis. Nephrol Dial Transplant 2015; 30: 62–70.)