Biomarkers and Updates on Pediatrics Lupus Nephritis

INTRODUCTION

Systemic lupus erythematosus (SLE) is a multiorgan autoimmune disease with increasing mortality that often targets young women and children of United States minorities. Childhood-onset SLE (cSLE)¹ has manifestations similar to those of SLE in adults, but earlier disease onset is accompanied by more severe multiorgan involvement, including lupus nephritis (LN) in up to 80% of pediatric patients. Treatment of LN in children continues to lack support from large randomized clinical trials. Instead, medication regimens for pediatric LN are deduced from studies in adult SLE and pediatric solid-organ transplants, or are based on consensus reached by associations of health care providers.

The criterion standard for the diagnosis and monitoring of LN remains histologic evidence from a kidney biopsy. Conversely, to reduce cost and avoid invasive procedures, monitoring of LN in clinics is achieved by measures that consider changes in certain blood and urine tests. Because such traditional testing for LN has limited responsiveness to change, it is ill suited to capture worsening or improvement of LN in a timely manner. Recently, promising LN biomarkers have been discovered that accurately reflect LN activity and chronicity as seen on kidney biopsy, and can forecast LN flares. In the future, such biomarkers are expected to facilitate the monitoring of LN in daily clinical care and the conduct of research studies in support of evidence-based therapies for LN in children.

EPIDEMIOLOGY, COURSE, AND ECONOMIC IMPACT

Given the phenotypic differences of cSLE around the world, the prevalence of kidney involvement with cSLE likely also varies with racial background and environmental exposures. The incidence of SLE is thought to have increased 10-fold during the preceding 50 years in industrialized Western countries,² which could indicate that cSLE in general, and LN in children in particular, also are becoming more frequent. Using information available in administrative databases and an algorithm that correctly identifies 80% of LN cases, Hiraki and colleagues³ report that in the United States 37% of children with cSLE and who are enrolled in Medicaid have renal disease. Based on this study, the risk of developing LN is independent of gender but is higher among teens than younger children. Compared with Caucasians, Asians have almost 5 times and African Americans a nearly 3 times higher risks of developing LN. Overall, the annual incidence of LN is 0.72 cases per 100,000 children in the United States. This figure may be a conservative estimate of the frequency of LN, as higher estimates are reported by population-based studies from tertiary pediatric rheumatology centers and a recent meta-analysis.^4,5

Recent 5-year renal survival rates in children with cSLE have ranged from 77% to 93%,^6–8 with marked improvement over the preceding decades.⁹ Nonetheless, adults with LN have an 8-times higher mortality and children with LN a 19-times higher risk of dying compared with age-matched general populations.^6,7,10 The poor prognosis of children with end-stage renal disease from LN is particularly troublesome. There is 22% mortality during the 5-year period since the initiation of renal replacement therapy, with cardiopulmonary compromise and infections accounting for 47% of all causes of death.⁶

Associated with the higher mortality is the need of more intensive therapy for LN in children. Among the almost 7400 cSLE-related hospitalizations in the United States in 2006, 57% noted the presence of LN¹¹ with an average charge of $43,100 per admission.¹¹ Based on this and an earlier study, LN accounts for 11% to 28% of cSLE-associated medical costs in the United States.¹² Taken together, the cost of therapy for LN in children likely exceeds $350 million annually in the United States.^3,11–13

DIAGNOSIS OF LN AND CLASSIFICATION

Kidney biopsies are required to establish the diagnosis of LN. Despite considerable variation in practice, there is consensus that reproducible daily proteinuria of at least 0.5 g, especially in the setting of an active urinary sediment, warrants a kidney biopsy in a child with cSLE who has not yet been diagnosed with LN.^14,15 Although clinically relevant biopsy findings are more common in the presence of significant proteinuria, the current approach results in at least 50% of newly diagnosed patients already being found to have proliferative LN, rendering them at a higher risk of end-stage renal disease.^16–19 A lower threshold for performing a kidney biopsy arguably is warranted in cSLE patients, including those with persistent isolated glomerular hematuria and new-onset low-grade proteinuria.

When interpreting a kidney biopsy specimen it is important to ensure that an adequate sample with sufficient numbers of glomeruli is available, namely, a minimum of 8 glomeruli that can be examined under light microscopy.^20,21 The International Society of Nephrology/Renal Pathology Society (ISN/RPS) Classification replaced in 2004 the previously used World Health Organization (WHO) Classification for LN.²⁰ The ISN/RPS Classification is based on light microscopy, rather than electron microscopy, as a tool for interpreting LN histology, even though it has been shown that electron microscopy greatly enhances the interpretation and classification of kidney biopsies.²⁰

The ISN/RPS Classification was introduced to standardize and clarify the interpretation of LN histology findings.²⁰ Six classes of LN are described with focus on changes concerning the renal glomeruli, and the National Institutes of Health (NIH) Histology Score is often used to quantify the degree of LN activity and chronicity (Table 1).²² The maximum score of the NIH-AI (activity index) and the NIH-CI (chronicity index) is 24 and 12, respectively, because scores from “(fibro)cellular crescents” and “fibrinoid necrosis/karyorrhexis” are given a weight of 2 in the NIH-AI (see Table 1). Pathologic changes of the kidney interstitium, are not well considered in the ISN/RPS Classification, although they are considered critical for the course of LN.²³ However, it is recommended to report the extent, severity, and type of tubulointerstitial (tubular atrophy, interstitial inflammation, and fibrosis) and vascular disease (vascular deposits, thrombi, vasculitis, sclerosis).²⁰

Table 1.

Classification and interpretation of lupus nephritis biopsy findings

ISN/RPS Lupus Nephritis Classification Criteria
Class I	Minimal mesangial lupus nephritis
Class II	Mesangial proliferative lupus nephritis
Class III	Focal lupus nephritisa
Class IV	Diffuse segmental (IV-S) or global (IV-G) lupus nephritisb
Class V	Membranous lupus nephritisc
Class VI	Advanced sclerosing lupus nephritis

NIH Activity and Chronicity Indexd
Active Lesions	Chronic Lesions
1. Endocapillary hypercellularity, with or without leukocyte infiltration and with substantial luminal reduction	1. Glomerular sclerosis (segmental, global)
2. Karyorrhexis (fibrinoid necrosis)e	2. Fibrous adhesions
3. Rupture of glomerular basement membrane	3. Fibrous crescents
4. Crescents (cellular or fibrocellular)e	4. Tubular atrophy
5. Subendothelial deposits identifiable by light microscopy (wireloops)
6. Intraluminal immune aggregates (hyaline thrombi)
NIH Activity Index 0–24	NIH Chronicity Index 0–12

^aIndicates the proportion of glomeruli with active and with sclerotic lesions.

^bIndicates the proportion of glomeruli with fibrinoid necrosis and cellular crescents.

^cClass V may occur in combination with class III or IV, in which case both will be diagnosed.

^dEach item scored from 0 to 3 depending on degree of involvement: 0 = no lesions; 1 = <25% of glomeruli; 2 = 25%–50% of glomeruli; 3 = >50% of glomeruli.

^eThese items scores have a weight of 2.

RISK FACTORS TO POOR OUTCOME OF LN

Clinical research has identified, albeit inconsistently, several risk factors for poor LN outcome^{3,6,16,24–33}; these include male gender, non-Caucasian race, nonadherence to treatment, presence of antiphospholipid or anti-dsDNA antibodies, persistent hypocomplementemia or proteinuria, nephrotic syndrome at presentation, failure to adequately respond to therapy by 3 months,³⁴ flare of LN,³⁵ or diagnosis with proliferative LN, especially in the setting of a high degree of histologic activity and damage. Given the multitude of the proposed risk factors for LN, close monitoring of any child patient with LN seems to be warranted in achieving the best possible control of LN.^15,34

MONITORING OF LN IN CLINICAL CARE

There are no studies that directly compare the clinical features of the various classes of LN between children and adults with SLE. However, the presentation of children with LN varies considerably, ranging from mild abnormalities on urinalysis, to anasarca caused by marked proteinuria, to posterior reversible encephalopathy owing to uncontrolled hypertension with nephritic syndrome.³⁶

Proteinuria

Abnormally elevated excretions of albumin and total protein in the urine are highly sensitive indicators of glomerular disease. Albumin is a small-sized molecule, and one of the first proteins able to pass through the kidneys. The value of monitoring microalbuminuria for the early diagnosis of LN has not been well established, and mesangial LN can be present without proteinuria.³⁷ A prompt and significant decrease in proteinuria after 3 and 6 months of therapy is an important prognostic factor for good long-term renal outcome.³⁸ Proteinuria furthers the development of tubulointerstitial inflammation and injury, and thereby a decline in renal function in the long term.³⁹

Traditionally proteinuria is quantified by a 24-hour urine collection. Conversely, and despite its common use, urine dipstick is poorly suited to quantify the degree of proteinuria.⁴⁰ There is now sufficient evidence that the protein-to-creatinine ratio in a random urine specimen, best from first morning urine,^41–43 is adequate to estimate daily proteinuria in cSLE. Whether 12-hour overnight urine collection is more accurate than estimation of proteinuria using spot urine will need further study.⁴⁴

Urine Sediment

The presence of cellular casts on urine-sediment examinations, for example, the microscopic examination of the cellular components at casts seen in centrifuged urine, is supportive of glomerulonephritis. Accuracy of urine-sediment interpretation requires timely processing of the urine, as lysis of leukocytes and erythrocytes occurs even within the first hour after collection, especially when low specific gravity and high urine pH are present. Presence of mucus in the urine can entrap both cells and casts, and sometimes repeated assessment of urine sediment is necessary to detect cellular casts.⁴⁵

Glomerular Filtration Rate

The reference method for assessing the “true” glomerular filtration rate (GFR) is to measure the renal clearance of inulin, ethylenediaminetetraacetic acid, and iohexol; that is, markers freely filtered through the glomerulus, neither secreted nor reabsorbed by the tubule. Because such techniques are complex and costly to perform, alternative means to estimate the GFR in a clinical setting have been developed.⁴⁶ In pediatrics, the 2009 modification of the Schwartz Formula and serum cystatin C–based methods seem reasonably accurate and easy to use in a clinical setting (Table 2).⁴⁶ Despite its appeal, the use of serum cystatin C to estimate the GFR of patients with LN will need further evaluation, as levels of cystatin C seem positively correlated with general SLE activity, even in the absence of LN or changes in renal function.^47,48

Table 2.

Estimation of GFR (eGFR) in children in comparison with the reference standard of inulin clearance (iGFR)

Comparators	Modified Schwartz Formulaa	Le Briconb
	eGFR = 36.5 × Height (cm)/Cr	eGFR 5 (78/Cys) + 4
eGFR means ± SD (mL/min per 1.73 m2)	109 ± 44c	99 ± 26
iGFR-eGFR means ± SD (mL/min per 1.73 m2)	−8 ± 29	2 ± 19
Accuracy 10% (%)d	38	46
Accuracy 30% (%)	84	90
Accuracy 50% (%)	96	98
Correlation between eGFR and iGFR	0.779e	0.784e

^aSchwartz GJ, Muñoz A, Schneider MF, et al. New equations to estimate GFR in children with CKD. J Am Soc Nephrol 2009;20:629–37.

^bLe Bricon T, Thervet E, Froissart M, et al. Plasma cystatin C is superior to 24-h creatinine clearance and plasma creatinine for estimation of glomerular filtration rate 3 months after kidney transplantation. Clin Chem 2000;46;1206–7.

^cP<.05, Wilcoxon paired test, in comparison with inulin clearance.

^dInterpretation: 38% of the patient’s eGFR is within 10% of the reference standard, ie, inulin clearance.

^eP<.001; Spearman correlation coefficient.

SHORTCOMINGS OF TRADITIONAL MEASURES OF LN

Whereas blood urea nitrogen and creatinine often stay in the normal range in cSLE, even if with profound histologic pathology, the urinary sediment and urinalysis are generally abnormal in untreated LN. Conversely, in pretreated patients only minor abnormalities on urinalysis, including mild proteinuria or hematuria, may be present in patients with severely active biopsy-proven LN. This finding is supported by the research of Christopher-Stine and colleagues,⁴⁹ who reviewed 25 LN patients undergoing serial kidney biopsies. At diagnosis proteinuria, hematuria, hypoalbuminemia, and hypertension were all associated with a worse LN class. By contrast, none of these parameters correlated with the LN class on follow-up biopsy, raising the possibility that normal urinalyses do not necessarily ensure the absence of active LN.⁴⁹

With LN, there is a balance between complement activation via the classical pathway, which facilitates the removal of immune complexes, and activation of the alternative pathway, which promotes kidney injury.⁵⁰ The literature is inconsistent at best as to whether the concentration of complement and anti-dsDNA antibodies can serve as useful markers of concurrent SLE activity or future flares.⁵¹ In 98 patients who experienced 146 flares, Ho and colleagues⁵² showed that hypocomplementemia and anti-dsDNA antibodies accompanied SLE relapse in only 54% and 27% of patients, respectively.

Research in adults with LN suggests that less than 25% of LN patients with low C3, C4, or anti-dsDNA levels have a concurrent flare of LN, and only 50% of LN flares are preceded by a drop in the levels of C3 and C4 or an increase in anti-dsDNA antibodies, respectively.^51,53 In other words, these tests are not much better than the flip of a coin in helping clinicians anticipate LN flare. These reports from adults with LN have been confirmed in children with LN.⁵⁴

Like the immunologic markers C3, C4, and anti-dsDNA antibodies that are traditionally used to assess the course of LN, kidney biopsies have their pitfalls. In a recent study, 5 experienced nephropathologists rated 126 renal biopsy specimens of 87 patients with proliferative LN.⁵⁵ These experts demonstrated significant variation in agreement when rating the various histologic aspects of biopsy specimens as part of the ISN/RPS Classification. Excellent agreement (>60%) was reached only for the number of glomeruli seen in the biopsy, the overall activity index score, and the presence of proliferative features. Conversely, agreement was less than 40% (interclass correlation coefficient <0.4) for the presence of mesangial proliferation, tubular necrosis, and, notably, the overall ISN-RPS class designation.⁵⁵

The aforementioned shortcomings of kidney biopsies, as well as the limitations of currently available urine and blood laboratory tests, support a need for potent biomarkers to help accurately diagnose LN and to determine the response of LN to therapy in a clinical setting.

BIOMARKERS AND ASSESSMENT OF THEIR QUALITY

In its simplest definition, a biomarker is anything that can be measured to extract information about a biological state or process. The NIH Biomarkers Definitions Working Group has defined a biological marker (biomarker) as “A characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.”⁵⁶ Biomarkers are the essential tools for the implementation of personalized medicine. The biomarker development process, also sometimes referred to as biomarker qualification, has typically been divided into 5 phases,⁵⁷ as shown in Table 3. In recent years, the ready availability of powerful tools to scan both the genome and the proteome of an organism have revolutionized and greatly accelerated biomarker discovery.

Table 3.

Phases of biomarker discovery, translation, and validation

Phase	Terminology	Action Steps
Phase 1	Preclinical discovery	Discover biomarkers in tissues or body fluids Confirm and prioritize promising candidates
Phase 2	Assay development	Develop and optimize clinically useful assay Test on existing samples of established disease
Phase 3	Retrospective study	Test biomarker in completed clinical trial Test if biomarker detects the disease early Evaluate sensitivity, specificity, receiver-operating characteristic
Phase 4	Prospective screening	Use biomarker to screen population Identify extent and characteristics of disease Identify false referral rate
Phase 5	Disease control	Determine impact of screening on reducing disease burden

From Devarajan P. Proteomics for biomarker discovery in acute kidney injury. Semin Nephrol 2007;27(6):637–51; with permission.

For biomarker discovery, microarrays are used to screen messenger RNA (mRNA) levels. This approach has yielded several biomarkers of kidney disease, such as neutrophil gelatinase-associated lipocalin (NGAL). Microarrays can be combined with other techniques, such as laser-capture microdissection, to target specific areas of diseased tissue to give mechanistic clues not possible just a decade ago. Even with this level of specificity, a daunting amount of biomarker candidates will be identified with these approaches, and the usefulness of such candidates must be sifted through for relevance. Another shortcoming of transcriptomic profiling approaches is that direct measurement in biological fluids is not possible and that mRNA levels do not always correlate with protein levels or enzyme activity. Hence, larger validation studies are necessary that measure protein levels to confirm the biological relevance of mRNA biomarkers.

Focusing on peptides and actual proteins, proteomics allow one to go beyond simple translation of mRNA into protein. Instead protein regulation, posttranslational modifications such as glycosylation and methylation, and even disease-specific fragmentation of proteins are assessed. Proteomic techniques are capable of identifying and quantifying proteins and peptides in exceedingly large numbers.⁵⁸ The urinary proteome itself is quite large, with laboratories having identified more than 1500 proteins to date.^59,60 The blood proteome is even larger, with more than 3000 nonredundant proteins identified in the plasma alone.^61–63 Adding the proteome of the cellular component of blood will yield thousands more.^64,65 To this end, we have entered what has been termed an “open loop,”⁶⁶ or unbiased, approach to biomarker discovery, in stark contrast to the hypothesis-driven approach of our past. With such a vast pool of potential biomarkers from readily available, noninvasive sources, one must take care to plan and design the proper experimental approach to ensure parsimony.

There are universal characteristics important for any biomarker: (1) they should be noninvasive, easily measured, inexpensive, and produce rapid results; (2) they should be from readily available sources, such as blood or urine; (3) they should have a high sensitivity, allowing early detection, and no overlap in values between diseased patients and healthy controls; (4) they should have a high specificity, being greatly upregulated (or downregulated) specifically in the diseased samples and unaffected by comorbid conditions; (5) their levels should vary rapidly in response to treatment; (6) their levels should aid in risk stratification and possess prognostic value in terms of real outcomes; and (7) they should be biologically plausible and provide insight into the underlying disease mechanism.^56,57

The most readily available sources of biomarkers are urine and blood. Urine is an excellent source of biomarkers produced in the kidney,⁶⁷ and thus may give better mechanistic insight into specific renal abnormalities. Urine is less complex than serum, and thus is easier to screen for potential biomarkers. Urinary biomarker studies typically adjust for urine creatinine to account for differences in urine concentration resulting from hydration status and medications such as diuretics. However, the utility of urine creatinine in biomarker correction has been questioned because of its variable excretion throughout the day and its dependence on normal renal function.

Serum biomarkers are considered more stable, as they are less prone than urine biomarkers to bacterial contamination. However, serum biomarkers are more likely to represent a systemic response to disease, rather than an organ response. There are exceptions, such as the troponins in cardiac disease. The real problem with serum as a source of biomarkers lies in the discovery phase. Serum has a wide range of protein concentrations across several orders of magnitude, with a small number of proteins (such as albumin) accounting for a large percentage of the volume; this can be akin to trying to spot a single strand of cotton in a large tapestry. Although assays do exist to remove these high-abundance proteins from serum, many potential biomarkers have been shown to bind to albumin.⁶⁸ Thus, albumin depletion to help identify relevant biomarkers risks erroneous removal of proteins relevant to LN.

The sensitivity and specificity of a biomarker go hand in hand. The receiver-operating characteristic (ROC) curve is a binary classification test, based on the sensitivity and specificity of a biomarker at certain cutoff points. ROC curves are often used to determine the clinical diagnostic value of a biomarker.^57,69 The area under the ROC curve (AUC_ROC) is a common statistic derived from ROC curves. An AUC_ROC of 1.0 represents a perfect biomarker, whereas an AUC_ROC of 0.5 is a result that is no better than expected by chance. An AUC_ROC of 0.75 or greater is generally considered a good biomarker while an AUC_ROC of 0.90 is considered an excellent biomarker.⁵⁷ However, even a sensitive biomarker with what experimentally would be considered an excellent specificity of 90% would still yield a false-positive rate of 10%, which may be unacceptably high for clinical use as a stand-alone marker.⁶⁶ As a result, the best approach clinically may be to find multiple biomarkers that can be combined as part of a panel to achieve even higher specificity.

TYPES OF LN BIOMARKERS

Traditional measures of LN have limited responsiveness to change, and are unsuited to capture worsening or improvement of LN in a timely manner. This lack of early response measures to verify the effectiveness of LN therapies hinders clinical care, requires clinical trials of new medications for LN to study large populations and follow them over several years, and increases the risk of negative trials. In addition, traditional measures of kidney function, such as creatinine clearance or protein-to-creatinine ratio, reflect significant loss of kidney function such that major renal damage can occur before it is detected by these traditional methods. Thus, novel biomarkers that can rapidly detect lupus renal involvement and severity, predict flares, and monitor treatment response and disease progression are greatly needed, and have been the subject of intense research.

The advent of new technologies to rapidly screen the genome and proteome over the last few decades has led to an explosion in the identification of novel biomarkers for many disease states. Ann immense number of biomarkers has been investigated in recent years, far too many to discuss in this article. The authors therefore focus the discussion on the most promising investigational biomarkers for LN discovered over the last several years.

Urine MCP-1

Monocyte chemoattractant protein-1 (MCP-1) is a leukocyte chemotactic protein involved in the mediation of inflammation and renal injury in LN.⁷⁰ Animal models of LN have demonstrated direct involvement of MCP-1 in renal abnormality, as blockade of MCP-1 through the use of an antagonist or an RNA oligonucleotide specifically designed to bind to and sequester MCP-1 (also known as a spiegelmer) led to marked improvement in LN and lupus-like inflammatory skin lesions.^71,72 Several cross-sectional studies have demonstrated that urine MCP-1 levels are concurrently higher in those patients with active LN than with nonactive LN.^73–75 The AUC_ROC of MCP-1 for distinguishing active LN from inactive LN⁷⁶ or nonrenal flares is 0.76.⁷⁷ Urine MCP-1 also seems to have promise in helping to distinguish certain classes of LN. Urine MCP-1 levels are significantly higher with ISN/RPS Classes III and IV than with other classes of LN (P = .01).^78,79 Both children and adults with Class IV LN have the highest glomerular expression of MCP-1.⁴⁶ There are some differing findings regarding the potential of urine MCP-1 to predict renal flares. A study by Rovin and colleagues⁷³ reported increases in urine MCP-1 as early as 2 to 4 months before the clinical diagnosis of a renal flare. However, a similar study by Tian and colleagues,⁸⁰ while demonstrating elevated MCP-1 during renal flares, did not find MCP-1 levels to be an independent predictor of flare.

Similar results were found by Chan and colleagues⁸¹ when examining chemokine mRNA from urine sediment of LN patients. MCP-1 mRNA levels were elevated during active LN in comparison with inactive LN and healthy controls. However, in this study urine MCP-1 mRNA levels were found not to be useful predictors of LN flares. It should be noted that the best use for MCP-1 as it relates to SLE is as part of a broader panel of markers, as elevated urine MCP-1 can also signal chronic fibrosis^82,83 and has presented in other glomerular disorders.⁸⁴ Thus a combinatorial approach may lead to additional specificity for LN.

Urine NGAL

NGAL is expressed in several cell types, including neutrophils, specific epithelia, and renal tubular cells. NGAL is markedly upregulated in the distal tubules in response to many types of kidney injury. It has garnered significant attention as a promising early marker for acute kidney injury,^85–91 but recent studies have also shed light on NGAL’s potential as a biomarker for chronic kidney disease, such as diabetic nephropathy^92,93 and focal segmental glomerulosclerosis,⁹⁴ as well as LN.^95,96 Two cross-sectional studies investigated NGAL as a biomarker for LN in pediatric patients⁹⁵ and adults.⁹⁷ In children, elevated urine NGAL levels had a high sensitivity and specificity for active biopsy-proven LN (AUC_ROC 0.94). In adults the specificity was still high (91%), but sensitivity was lower (50%) for LN. This thread is a common one in biomarker studies, as adults typically have more concurrent confounding physiologic conditions, which leads to higher variability in biomarker measurements. NGAL was not correlated with extrarenal SLE disease activity in either population. More recent longitudinal studies in the pediatric population have shown that urine NGAL as well as plasma NGAL levels are significantly higher in SLE patients than those with juvenile idiopathic arthritis (JIA) or healthy controls, unrelated to physiologic factors such as height, weight, and age.⁹⁸ Levels of urine NGAL, but not plasma NGAL, correlated well with LN activity scores.^96,98 Urine NGAL rose 3 to 6 months before worsening renal disease activity, demonstrating value in predicting flares.^96,98 One study demonstrated a lesser, though significant, increase in plasma NGAL as early as 3 months before flare.⁹⁶ In addition, in patients with a biopsy, urine NGAL levels were greater in patients with diffuse proliferative than membranous nephropathy, indicating, along with MCP-1, the possible use of NGAL in a panel to distinguish LN classes.⁹⁸ Similar to MCP-1, urine NGAL is not specific to LN and thus must be used in a context-specific setting.

Hepcidin

Hepcidin is a small peptide hormone mainly produced in the liver, and has a role in iron homeostastis. Hepcidin is upregulated in response to high iron levels and inflammation, and decreases during anemia and iron deficiency. Proteomic evaluation by surface-enhanced laser desorption-ionization time-of-flight mass spectrometry (SELDI) revealed the 25- and 20-amino-acid isoforms of hepcidin as potential biomarkers for LN.⁷⁹ Zhang and colleagues⁷⁹ prospectively analyzed 24 LN flare cycles in 19 patients, and demonstrated an increase in hepcidin-20 4 months before flare, which then decreased to baseline levels by 4 months after flare. An opposing pattern was discovered for hepcidin-25, which decreased during renal activity then returned to baseline along with hepcidin-20 after flare. It will be interesting in future studies to evaluate the physiologic role of hepcidin in LN because it is regulated in part by inflammatory cytokines, such as interferon-α and interleukin (IL)-6, which are known to play a role in modulating tissue damage in SLE,^99,100 and have been shown experimentally to induce monocyte expression of hepcidin in vitro.¹⁰¹ It has been speculated that monocyte infiltration of the kidney may be the source of urine hepcidin in LN.

Urine Protein Signature

Also using SELDI, Suzuki and colleagues¹⁰² discovered and subsequently validated⁵⁴ a protein signature that identified active LN in children. After removal of 4 albumin fragments from the signature, the panel included transferrin (Tf), orosomucoid (or α-1 acid glycoprotein [AGP]), ceruloplasmin (CP), and lipocalin-type prostaglandin D synthase (L-PDGS, or β-trace protein). Using enzyme-linked immunosorbent assay or immunonephelometry, all 4 proteins were found to be significantly higher in patients with active LN than in those with nonrenal SLE or JIA controls. Urine L-PDGS, AGP, and Tf all increased as early as 3 months before renal flare, but Tf did so most consistently, demonstrating increased sensitivity to renal changes in SLE in comparison with L-PDGS or AGP. Urine CP did not demonstrate the ability to predict flares. Combining this panel with other markers such as NGAL and MCP-1 may demonstrate enhanced predictive and diagnostic value in comparison with individual markers alone.

Complement Component C4d

C4d is a breakdown product of the activated complement factor C4b, a critical component of the C5 convertase. In a controversial pilot study using an alternative approach, Batal and colleagues¹⁰³ evaluated cellular deposition of the immune complex C4d on circulating erythrocytes, reticulocytes, and platelets as a potential biomarker for LN activity. Previous studies had linked peritubular capillary and glomerular staining of C4d with severity of LN and development of renal thrombotic microangiopathy, respectively.^104,105 The investigators found higher circulating levels of erythrocyte-bound C4d (EC4d) and reticulocyte-bound C4d (RC4d) in LN patients than in both nonrenal SLE patients and patients with renal disease without SLE. Moreover, EC4d levels correlated with the NIH renal activity index. There has been some level of skepticism¹⁰⁶ regarding the ability of these markers to distinguish renal from nonrenal SLE, as higher levels can also observed in SLE patients without LN,^107,108 and there have been no scientific findings to date that dispute the results. An additional study lends credence to the finding in this study, indicating higher levels of certain C4d-positive circulating T cells in LN patients than in those without LN.¹⁰⁹ Further prospective investigations of circulating C4d are needed for it to rise to the levels of the previously discussed biomarkers for LN, but the novel approach warranted mention in this review.

TWEAK

Tumor necrosis factor–like weak inducer of apoptosis (TWEAK) is a member of the tumor necrosis factor (TNF) superfamily, and is involved in modulating cell survival and induction of several proinflammatory chemokines through its receptor fibroblast growth factor–inducible protein 14 (Fn14).¹¹⁰ In human kidney, TWEAK acts on multiple Fn14-expressing cells types, including podocytes, tubular cells, and mesangial cells, and is responsible for induction of several mediators of inflammation, including MCP-1, interferon-γ–inducible protein 10 (IP-10), intercellular cell adhesion molecule 1, vascular cell adhesion molecule 1 (VCAM-1), matrix metalloproteinases 1 and 9, and macrophage inflammatory protein α.^111,112 During periods of inflammation, Fn14 expression is upregulated, which lends itself to enhancing a positive feedback loop. The major source of TWEAK in LN is thought to be infiltrating monocytes and macrophages. Cross-sectionally, urinary TWEAK levels are significantly higher in active LN; levels are significantly higher in patients with LN flare than in those with stable disease.^113,114 In a multicenter longitudinal analysis, Schwartz and colleagues¹¹⁵ discovered that whereas urinary TWEAK levels peaked at the height of renal flare, urinary TWEAK was significantly elevated 4 to 6 months before and following renal flare. Performance of urinary TWEAK in distinguishing LN patients from SLE patients without kidney involvement was better than that of anti-dsDNA levels and complement C3 or C4 levels. The study also demonstrated a strong association between urinary TWEAK levels and LN activity over time. Conversely, serum levels of TWEAK were not associated with LN activity. TWEAK is intriguing as a biomarker for LN, and has a biologically plausible role in LN pathology.

Other Chemokines, Receptors, and Adhesion Molecules

Space does not permit in-depth discussion of all biomarkers under investigation for LN, but several cytokines, chemokines, and their receptors deserve some mention. Chemokine C-X-C motif ligand 10 (CXCL10, also known as IP-10) and its receptor CXCR3 promote T-cell migration to areas of inflammation and are upregulated in SLE.^116,117 CXCL10 and CXCR3 mRNA levels collected from urine sediment were highly specific for identifying Class IV LN (AUC_ROC 0.89 for CXCL10 and 0.79 for CXCR3), and also demonstrated reduction in response to successful treatment signified by clinical remission.¹¹⁸ FOXP3 (forkhead box P3) mRNA collected from urine sediment of LN patients has been found to be significantly higher in LN patients,¹¹⁹ despite FOXP3 levels in regulatory T cells having been found to be lower in patients with active lupus than in healthy controls.¹²⁰ Research has also shown that a reduced number of circulating FOXP3⁺ T cells and serum transforming growth factor β levels inversely correlated with LN activity as measured by SLE disease activity index renal domain score (P = .0013 and 0.0005, respectively).¹⁰⁹ Collection of mRNA from urine sediment presents several technical difficulties, such as stability, which may limit the clinical utility of urine mRNAs as biomarkers. So although there may be a link between FOXP3 and LN, additional study must be completed to solidify its role and usefulness as a biomarker for LN.

VCAM-1 demonstrates reliability as an indicator of renal disease activity in LN. VCAM-1 has been shown to be induced in mice by inflammatory cytokines such as IL-1 and TNF.¹²¹ VCAM-1 plays a role in tethering leukocytes, which are drawn to sites of inflammation, to endothelial cells.¹²² Urinary VCAM-1 has been shown in several studies of human disease to be strongly correlated with LN activity and severity^77,123,124 in LN. Serum levels of VCAM have previously been shown to correlate with the severity of LN, being highest in WHO Class III and IV, versus inactive or mild nephritis (WHO Class I or II),¹²⁵ and levels diminished with treatment. Singh and colleagues¹²⁶ compared urine levels of VCAM-1, MCP-1, and CXCL16 (another potential LN biomarker) with pathologic features of LN on biopsy collected concurrently with the urine sample. Urine VCAM and MCP-1 were highly predictive of LN when compared with healthy controls (AUC_ROC 0.92 and 0.89, respectively). Surprisingly, urine MCP-1 was also significantly higher in African American subjects than in persons of other ethnic origins. Of the 3 markers, urine VCAM-1 was most highly correlated with LN activity, with none shown by CXCL16. CXCL16 and urine VCAM-1 were significantly higher in patients with WHO Class IV LN compared with other Classes, as determined by concurrent biopsy analysis. It should be noted that this association with Class IV proliferative nephritis may not be specific to pathology, but a may be a result of these patients having a high degree of renal disease activity. These findings provide a great deal of support for urine VCAM-1 as a biomarker for LN, but these studies have all been cross-sectional. Longitudinal studies are needed to determine the utility of VCAM-1 in monitoring disease progression and detecting flares. It should also be noted that, like NGAL and MCP-1, elevated VCAM-1 is not exclusive to LN. Increased levels of VCAM-1 have been found in other glomerular diseases such as membranous nephropathy and focal segmental glomerulosclerosis.¹²⁶

CURRENT TREATMENT OF LUPUS NEPHRITIS IN CHILDREN

The novel biomarkers introduced in the preceding sections are not used to support efficacy in clinical trials at present, although validation studies are ongoing to achieve biomarker qualification by regulatory bodies. Qualification would allow for the use of biomarkers in clinical care and research.^127,128 In addition, there is no known biomarker at present that a priori would support the choice of therapeutics for the treatment of LN. However, it seems reasonable to assume that novel biomarkers will become available for clinical use within the next 5 to 7 years.

No medication has likely improved the prognosis of LN more than systemic glucocorticoids (GC), especially if combined with immunosuppressive medications. Nonetheless, use of GC is a concern, given the often devastating short-term and long-term side effects. There is a lack of systematic studies in support of the most appropriate dose of GC in patients with LN. Based on consensus among pediatric rheumatologists in the United States, three GC dosing regimens for the treatment of proliferative LN in children have been proposed,¹²⁹ but data are lacking to determine which regimen is the most appropriate for a given patient. Of note, the Joint European League Against Rheumatism and the American College of Rheumatology consider much lower GC exposure sufficient for mainly adults with LN.^14,130

Unless commanded by cSLE activity in other organ systems, hydroxychloroquine and GC are considered sufficient for the treatment of ISN/RPS Class I and, often, Class II LN.^14,131 For proliferative LN Class III or IV with or without membranous features, treatment with cyclophosphamide or mycophenolate mofetil (MMF) for induction therapy, and maintenance therapy using MMF or azathioprine are proposed.^14,129 Based on a Cochrane review of studies of adults with LN,^132,133 compared with intravenous cyclophosphamide, MMF was as effective in achieving stable kidney function and complete remission of proteinuria. No differences in mortality or major infections were observed. In maintenance therapy, the risk of LN flare was significantly higher with azathioprine or cyclophosphamide compared with MMF. Based on small studies, children and adolescents have a response to MMF and cyclophosphamide similar to that of adults with LN.¹³⁴ Whether MMF is as effective in children as it is in adults¹³⁵ or whether cyclophosphamide might have a better risk/benefit profile in children than in adults owing to lower frequency of clinically relevant ovarian injury and lower risk of nonadherence is not supported by high-level scientific evidence.¹²⁹ In addition, the pediatric correlate of the “Euro Lupus Regimen” for the dosing of intravenous cyclophosphamide has not been developed or systematically studied.¹⁴

There is mounting evidence that individualized dosing of MMF based on pharmacokinetic profiling will increase the likelihood of achieving remission of LN.^136,137 Target exposure between 60 and 90 mg/h/L is more often associated with LN improvement, with the highest exposures being reserved for the most severe cases because of the increased frequency of adverse effects.¹³⁶ Given high interindividual differences, weight-based or body-surface–based dosing of MMF does not suffice to reliably achieve such a target exposure.¹³⁸

Pure membranous lupus glomerulonephritis (ISN/RPS Class V) seems rarely the initially diagnosed type of LN, and typically the other forms of LN develop into Class V over time. Treatment of Class V probably should not differ from that of idiopathic membranous nephropathy. Depending on the degree of proteinuria, only angiotensin-inhibiting medications, or GC with MMF or other immunosuppressives are the preferred initial therapy.¹³⁹

Despite favorable reports mostly from observational studies,^140–144 the clinical trial of the anti-CD20 antibody rituximab (Rituxan, Mabthera) failed to show clinically relevant improvement of LN.¹⁴⁵ The anti–B-lymphocyte stimulator antibody belimumab (Benlysta) has recently been approved for the treatment of active SLE,^146,147 but its benefit or detrimental effects on LN will require further study.

There are currently several ongoing studies of LN, some including younger patients, which explore the efficacy of various combination therapies of GC with regimens including various combinations of cyclophosphamide, cyclosporin, azathioprine, tacrolimus, MMF, fludarabine, azathioprine, rituximab, abatacept, etanercept, and leflunomide, as well as mesenchymal stem cells. It is hoped that these studies consider the genetic differences of patients and include potent LN biomarkers when assessing the benefits of these therapies under investigation.

It is plausible to assume that the use of novel biomarkers will yield better stratification of patient populations for the purpose of clinical trials, and enable researchers to determine the response to LN therapy earlier and more accurately. This approach would necessitate smaller sample sizes for clinical trials, and ultimately make possible adequately powered studies in children with LN.

KEY POINTS.

• Lupus nephritis is frequently diagnosed in children with systemic lupus erythematosus and warrants close medical attention to avoid progression to end-stage renal disease.

• Diagnosis of lupus nephritis requires at present a kidney biopsy.

• Current laboratory tests used to monitor lupus nephritis lack accuracy, making appropriate management difficult.

• Novel urine biomarkers hold promise for improving the approach to the surveillance of lupus nephritis and interpretation of patient response to therapy.

• Despite the lack of adequately powered clinical trials, standardized approaches to the therapy for children and adolescents with lupus nephritis are now available.