Background
Nephritis and its treatment constitute one of the major causes of morbidity and mortality in systemic lupus erythematosus (SLE). Approximately 50–60 % of SLE patients will develop lupus nephritis (LN), with 30 % of these patients developing significant renal impairment, culminating in end-stage renal disease in 15 % of patients [
1,
2]. The clinical course of LN is highly variable, being marked by unpredictable flares and variable responses to treatment [
3,
4]. As treatment of LN results in significant immediate (e.g., infection) and delayed (e.g., avascular necrosis and cardiovascular disease) onset morbidity, the clinical care of LN patients seeks to establish a balance between optimal control of inflammation and tissue injury, while limiting exposure to the side effects of immunosuppressive therapies [
4]. One of the challenges to this approach is the lack of biochemical and serologic tests that accurately reflect the extent and type of renal inflammation.
Currently, monitoring of LN relies on serological biomarkers and measures of renal dysfunction (e.g., proteinuria and serum creatinine) [
5]. Although elevated anti-dsDNA antibody levels and hypocomplementemia associate with disease activity in cross-sectional analyses, longitudinal studies indicate that these traditional biomarkers do not distinguish between active SLE patients with and without LN, and are inconsistent at predicting impending flares [
3]. Proteinuria and other measures of renal function also falter as accurate markers of immune-mediated renal injury. As these measures reflect kidney damage, reliance on proteinuria as a marker of renal inflammation leads to delayed initiation of treatment. Conversely, it may also lead to unnecessary prolongation or premature tapering of immunosuppressive therapy due to the persistence/resolution of urinary and/or functional abnormalities that may not reflect resolution of the inciting immunologic insult. Indeed LN-associated proteinuria frequently persists for years after renal injury, normalizing in less than 50 % of patients within 2 years [
6], and often a renal biopsy is the only way to distinguish between persistent activity and a chronic inactive lesion. As a result, there has been tremendous interest in the identification of biomarkers that accurately indicate the extent, type, and course of renal inflammation in LN.
Of the potential compartments that can be surveyed to identify LN-specific biomarkers, the urinary compartment may hold the most promise. Several previous studies have compared and contrasted levels of various cyto/chemokines within serum and urine samples of patients with active LN. These studies indicate that urine levels most accurately reflect renal status as compared with serum levels of these markers [
7‐
13]. While these studies have identified some potential urinary biomarkers for diagnosis and management of LN, such as MCP-1 [
7,
9,
14‐
16], NGAL [
8,
12,
17‐
21], TWEAK [
11,
14,
22‐
24], and sVCAM-1 [
13,
25‐
27], there has been no comprehensive survey of urinary proteins to evaluate whether the above proteins constitute the best potential markers of active renal disease. Furthermore, analyses are limited on the association between urinary analyte levels and renal disease activity indices since, in the majority of studies, paired renal biopsies were not performed. In the current study, a Luminex-based proteomics approach was used to contrast levels of 128 urinary proteins found in the urine of SLE patients with active LN (ALN) and active SLE patients without LN (ANLN) to identify potential novel urinary biomarkers.
Methods
Subjects and data collection
Two independent cohorts were used for this study. The discovery cohort consisted of 85 active SLE patients satisfying four or more of the revised 1997 ACR classification criteria for SLE [
28] and 24 age- and sex-matched healthy controls (HC). Patients were recruited from the University Health Network and Mount Sinai Hospitals. Sixty patients had ALN, confirmed by renal biopsy performed within 2 weeks of urine sampling for 88 % of patients (mean 7.5 days from sample accrual), with the remainder having active disease (score >0 on the clinical SLE Disease Activity Index-2000 (SLEDAI-2 K) components [
29]) and no clinical evidence of LN. The validation cohort consisted of 33 patients with ALN (with ≥1 of the renal SLEDAI-2 K indices scoring positive and requiring changes in therapy), 16 patients with ANLN, and 30 patients with a prior history of biopsy-proven LN who were in remission with no urinary abnormalities (except one patient with fixed proteinuria (0.82 g/day) not requiring treatment).
Measurement of urinary analyte concentrations
All urine samples were spun to remove cellular debris and frozen at –80 ° C. To avoid repeated freeze/thaws, samples were thawed once on ice, sub-aliquoted, re-frozen at –80 ° C, and then individual aliquots thawed immediately prior to use. The urinary concentrations of 128 analytes were measured by coupled bead assay (Luminex using MILLIPLEX® Map Kits (EMC Millipore Corporation), Eve Technologies Inc.) Further information regarding the sensitivity and dynamic range of the assays can be found on the company website (
http://www.emdmillipore.com/). For the majority of assays, the urine samples were run undiluted except: KIM-1 and renin (diluted 1/2); albumin, beta-2-microglobulin, clusterin, cystatin C, and osteopontin (diluted 1/50); and TIMP-1 and TIMP-2 (diluted 1/5). For the discovery phase, all analytes were measured in duplicate for ALN patient samples, with a single sample on each of two separate plates, and averaged. ANLN and HC samples were measured in singles and randomly assigned to one of the two plates with equivalent numbers for each group per plate. As the duplicates run on separate plates were very reproducible, samples in the validation phase were run in singles on a single plate. All results were normalized to urinary creatinine prior to analysis.
Renal histopathology scoring
ISN-RPS histopathological class [
30] and activity and chronicity scores [
31,
32] were determined by an individual renal pathologist (CA-C), blinded to the results of the urine protein determination.
Statistical analyses of discovery data
Normalized protein data were loaded into the R statistical environment (v3.1.1) for analyses. Data were scaled using the standard deviation for each variable and hierarchical clustering performed using divisive analysis (DIANA) with a Pearson’s correlation as a similarity metric. The Adjusted Rand Index, available in the mclust package (v4.4), was used to corroborate the clustering. Normalized abundance data were correlated with clinical variables across all patients using Spearman's correlation, followed by false discovery rate (FDR) adjustment of the
p values to correct for multiple testing. A Venn diagram was generated using the VennDiagram package (v1.6.9) to visualize the overlap of significantly correlated genes among clinical variables [
33]. Data were log
2 transformed and linear modeling was performed with the limma package (v3.20.9) in R to identify proteins with significant differences in abundance between groups. An empirical Bayes moderation of the standard error [
34], followed by FDR correction, was employed [
35]. Coefficients (i.e., log
2 fold-changes relative to control samples) that were determined to be significantly different from 0 following FDR correction (
p
adj < 0.01) were carried forward. A Venn diagram was generated (as described above) to visualize overlap of significantly altered proteins among clinical variables, with hypergeometic testing performed to determine if overlap was greater than expected by chance alone. Finally, results (magnitude, direction, and significance of change) were visualized in a dotmap using the lattice (v0.20-29) and latticeExtra (v0.6-26) packages for R.
A receiver operating characteristic (ROC) analysis was performed to evaluate the ability of various analytes to discriminate between active patients with or without nephritis, or with and without proliferative changes on renal biopsy. A total of 9 analytes were evaluated using data from 85 active lupus patients for the analysis of nephritis and 8 analytes on 60 patients with paired renal biopsies for analysis of proliferative nephritis. The pROC package (v1.8) for R (v3.2.1) was used to calculate the true positive (sensitivity) and false positive (1 – specificity) rates across various analyte level thresholds, along with the area under the curve (AUC). ROC curves were created using the lattice (v0.20-33) and latticeExtra (v0.6-26) packages for R.
To assess the ability of selected analytes alone or in combination with conventional biomarkers to discriminate between proliferative and non-proliferative/chronic nephritis, all possible combinations of conventional biomarkers (C3, anti-dsDNA, albuminuria) and the top performing univariate analytes associated with the activity index (vWF, IP-10, PDGF-BB, IL-16, adiponectin) were evaluated. Samples were divided into four balanced folds (such that each fold contained a similar number of active proliferative (1) and non-proliferative (0) nephritis patients). For each combination of analytes, a linear model was fit using disease status (1/0) as the response to be predicted by the indicated analytes (i.e., response ~ analyte.1 + analyte.2 …). Each model was trained using data from three of the four folds, with the fourth fold used for testing. This process was repeated four times such that each fold was used for testing only once. Mean sensitivity, specificity, and accuracy were calculated across the four repetitions.
Statistical analysis of validation data
Statistical significance of differences between groups (e.g., ALN and ANLN) was determined using the Mann-Whitney U test with a Bonferroni correction for multiple comparisons.
Discussion
In this study we used an unbiased proteomics-based approach to compare the levels of 128 proteins in the urine of active SLE patients with and without LN. We show that many urinary proteins are elevated in patients with ALN, with approximately one-third of the analytes tested being at least 2-fold increased in the urine of patients with ALN as compared to ANLN. These proteins represent a broad array of molecules, many of which have been implicated in the pathogenesis of nephritis through diverse mechanisms including: cyto/chemokines and their receptors (e.g., IL-15 and PF4), metalloproteinases and their regulators (e.g., TIMP-1 and MMP-9), growth factors (e.g., GM-CSF), markers of endothelial injury/repair (e.g., adiponectin, PAI-1, and vWF), and markers of kidney damage (e.g., KIM-1 and cystatin C).
Although elevated levels of multiple proteins were seen in the urine of patients with ALN, these elevations did not appear to result solely from the decreased filtering capacity of the kidney in LN. For those urinary analytes that demonstrated >4-fold increase in the urine of ALN patients, there was no or only a weak correlation with albuminuria (Fig.
2c, and data not shown). Furthermore, consistent with previously published reports [
7‐
13], we have previously observed a poor correlation between serum and urine levels of MCP-1, adiponectin, and sVCAM-1 (unpublished observations). These findings suggest that the elevated levels of urinary proteins in LN reflect increased elaboration within the kidney as a consequence of active inflammation. In support of this concept, increased kidney expression of many of the proteins identified in the current study has been described in various nephritis models [
10,
13,
36‐
39].
A highlight of our study was the identification of a number of biomarkers for ALN with markedly elevated urinary levels as compared to ANLN and remission LN (>4–8-fold). Of the nine urinary proteins in this group that replicated in the validation cohort, four have been previously reported to be elevated in ALN. These include adiponectin [
10], albumin [
40], sVCAM-1 [
13,
25‐
27], and IL-6 [
41,
42]. The remaining five, including PAI-1, IL-15, PF4, TIMP-1, and vWF, represent novel potential urinary biomarkers that have not previously been described in LN. Notably, a number of these biomarkers appeared to outperform previously proposed biomarkers such as NGAL, MCP-1, and TWEAK, with improved sensitivities and specificities at optimal cut-offs. These differences do not appear to reflect differences in the ability of the Luminex system to detect these previously reported biomarkers, because the fold-increases and sensitivity and specificity at optimal cut-offs were very similar in our study to those observed in previous studies [
7,
11,
15,
18,
23].
While several of the biomarkers that demonstrated the greatest fold-differences between ALN and ANLN demonstrated very high sensitivities (up to 93 %) and specificities (up to 96 %), no one biomarker appeared to be sufficient to diagnose ALN with 100 % accuracy, and it is likely that a panel of biomarkers may more optimally discriminate LN from other disease states. Based upon our ROC results, it is probable that a panel consisting of a combination of a subset of the five analytes demonstrating >8-fold difference between ALN and ANLN would offer the best discriminative ability; however, this will require testing in an independent unselected cohort of SLE patients with both active and inactive disease.
Previous work has shown that renal biopsies demonstrating diffuse proliferative (class IV) LN and increased glomerular activity indices are associated with a poorer long-term prognosis [
31,
32,
43,
44], an increased likelihood of re-flare [
45], and increased chronicity scores on subsequent renal biopsies with glomerulosclerosis and consequent renal functional deterioration [
46]. Given the invasive nature of renal biopsies, there has been considerable interest in the identification of urinary biomarkers that accurately reflect renal disease activity scores and/or proliferative nephritis. In this study, we identified 10 urinary analytes that showed a strong correlation with the renal biopsy activity score. Although several urinary proteins, including sVCAM [
47], NGAL [
8,
48], MCP-1 [
48,
49], adiponectin [
48], KIM-1 [
48], and TWEAK [
24], have been previously reported to correlate with the activity index on renal biopsy, with the exception of adiponectin none of these analytes were replicated in our study. Thus, we have identified nine novel proteins that are more closely associated with the activity index on renal biopsy than all previously reported associated proteins except adiponectin. Notably, addition of various combinations of the five top performing analytes associated with the activity index to conventional biomarkers of renal disease activity (C3, anti-dsDNA, and albuminuria) resulted in an improved ability to discriminate between active proliferative and non-proliferative or chronic nephritis, supporting their potential clinical utility for the diagnosis of active proliferative nephritis.
Recently a composite urinary biomarker activity index was reported based upon the urinary levels of six proteins, including adiponectin, NGAL, MCP-1, KIM-1, ceruloplasmin, and hemopexin, that can differentiate pediatric LN patients with high activity indices on renal biopsy from those with low or moderate activity indices with >92 % accuracy [
48]. Our findings raise the possibility that further improvements in this accuracy might be possible through the use of alternate biomarkers that are more closely associated with renal disease activity, a possibility that can be addressed in further studies.
Given the large number of urinary analytes examined in our discovery cohort, it was important to replicate these findings. Although only 18 of the biomarkers examined were replicated, it is likely that this results from several differences between our discovery and validation cohorts. Firstly, our validation cohort was significantly smaller in size (approximately half the size of our discovery cohort) which may have restricted our ability to detect less robust biomarkers. Additionally, the majority of patients in our validation cohort with ALN were re-flaring, usually after several years of treatment, whereas the majority of the patients in the discovery cohort had new-onset LN. As previous work suggests that repeat flares are associated with an increased likelihood of chronic rather than active changes [
50], this might have affected our ability to replicate urinary analytes that are associated predominantly with active proliferative lesions, such as IL-16, PDGF-BB, sgp130, and eotaxin. Finally, there may have been significant differences between the cohorts in the distribution of renal biopsy classes (e.g., more class V in the validation cohort), which may also have affected our ability to replicate these differences. If the inability to validate these activity-associated analytes is related to these cohort differences, then these biomarkers may still be clinically useful as biomarkers that are specific for active class III/IV LN.
The majority of the validated urinary analytes were found to normalize in patients who had achieved renal remissions. These findings suggest that measurement of these urinary proteins will have clinical utility for monitoring responses to therapy and prediction of flare, as has been shown for some of the previously published urinary biomarkers, such as NGAL [
12,
18‐
21] and TWEAK [
23]. Longitudinal studies are ongoing to define which combination of the previously published biomarkers and novel biomarkers identified in this study best reflects changes in LN activity and response to therapy, with the goal of defining an optimal biomarker panel for the monitoring of LN.
Acknowledgements
The authors would like to thank Drs. Murray Urowitz, Dafna Gladman, Jorge Sanchez-Guerrero, and Arthur Bookman for provision and care of study patients.