Diagnostic and Prognostic Value of Soluble Urokinase-type Plasminogen Activator Receptor (suPAR) in Focal Segmental Glomerulosclerosis and Impact of Detection Method

The plasma soluble urokinase-type plasminogen activator receptor (suPAR) is a biomarker for focal segmental glomerulosclerosis (FSGS), but its value is under discussion because of ambiguous results arising from different ELISA methods in previous studies. The aim of this study was to compare diagnostic performance of two leading suPAR ELISA kits and examine four objectives in 146 subjects: (1) plasma suPAR levels according to glomerular disease (primary, secondary and recurrent FSGS after kidney transplantation, other glomerulonephritis) and in healthy controls; (2) suPAR levels based on glomerular filtration rate; (3) sensitivity and specificity of suPAR for FSGS diagnosis and determination of optimal cut-offs; (4) suPAR as prognostic tool. Patients with FSGS showed significant higher suPAR values than patients with other glomerulonephritis and healthy individuals. This applied to subjects with and without chronic kidney disease. Although both suPARnostic™ assay and Quantikine Human uPAR ELISA Kit exerted high sensitivity and specificity for FSGS diagnosis, their cut-off values of 4.644 ng/mL and 2.789 ng/mL were significantly different. Higher suPAR was furthermore predictive for progression to end-stage renal disease. In summary, suPAR values must be interpreted in the context of population and test methods used. Knowing test specific cut-offs makes suPAR a valuable biomarker for FSGS.

Study population and recruitment. We measured plasma suPAR in 146 adults over 18 years. Overall 27 patients with primary FSGS, 21 subjects with secondary FSGS and 6 patients with recurrence of FSGS after renal transplantation with clinical and histologic data recorded at the Medical University of Vienna were included in this study. The primary sample collection was carried out between 06/2013 and 04/2015 and follow-ups were performed until 12/2018. Clinical and histological data were obtained at the date of presentation. Moreover, 60 patients with other forms of GN and 32 healthy individuals were used as age-and gender-matched controls ( Table 1). Patients with "Other GN" included subjects with membranous GN, minimal change GN, immune complex GN, anti-GBM GN, IgA GN, Lupus GN, interstitial GN and patients with unspecified GN as given in Supplementary Table S1.
Exclusion criteria were age <18 years, active inflammatory disease, chronic liver disease, malignancy, pregnancy and any illicit drug use. Patients with FSGS or other GN whose renal biopsy specimen contained less than 10 glomeruli were also excluded.
We first examined the association between baseline suPAR levels and measures of kidney function (eGFR, urine protein/creatinine ratio and albumin/creatinine ratio) in all study subjects. We also investigated the association of suPAR and change in eGFR and further clinical parameters and outcomes during follow-up in all 54 FSGS patients during a follow-up period of 5 years.
Histopathology. The kidney biopsy was performed at the time of diagnosis. Biopsy specimens were formalin-fixed, paraffin embedded, cut at 2 µm and histologically evaluated by light microscopy, indirect immunohistochemistry and electron microscopy following institutional guidelines. All of the biopsy slides were reviewed by two pathologists blinded to clinical data of the patients.

Sample collection.
Fasting venous blood samples were collected of 54 FSGS patients and samples from 92 age-and gender matched patients with other forms of GN or healthy individuals as normal controls. Plasma Plasma suPAR measurement. Concentration of plasma suPAR in human subjects were detected using the suPARnostic ™ assay (Virogates, Copenhagen, Denmark) and the Quantikine Human uPAR Immunoassay (R&D Systems, Minneapolis, MN, USA) 1 , following the manufacturers protocol. Both kits are measuring suPAR concentrations and minimum detectable levels are 100 pg/mL and 33 pg/mL, respectively. A schematic illustration of both kits is shown as Supplementary Fig. S1. suPARnostic ™ assay (Virogates, Copenhagen, Denmark). The suPARnostic kit (ViroGates A/S, Birkerød, Denmark) represents a double monoclonal antibody sandwich assay that measures total suPAR, including both full-length and cleaved forms. Plates are coated with catching rat monoclonal antibody. Horseradish peroxidase (HRP)-labeled detection mouse monoclonal antibody is mixed in a volume of 225 μL of Peroxidase Conjugate solution together with 25 μL of plasma. Out of this, 100 μL was transferred (in duplicates) to the pre-coated immunoassay plate and incubated for one hour. After washing procedures, 100 μL of the TMB substrate (3,3′,5,5′ tetramethylbenzidine) was added to each well. The substrate/chromogen reaction was stopped after 10 minutes by the Stop Solution (0.45 M sulphuric acid -H 2 SO 4) . All incubations were performed at room temperature in the dark.
Plasma sample concentrations were calculate according to the standard curve included in each plate. The intra-assay variation of the kit is given as 2.8% and the inter-assay variation as 9.2% 33 . Samples were randomly distributed between kits of five different lot numbers and were all measured in duplicates. All samples had duplicate coefficients of variants (CVs) <10%.
Quantikine Human uPAR Immunoassay (R&D Systems, Minneapolis, MN, USA). With regard to the Quantikine Human uPAR Immunoassay, 100 µL assay diluent was applied to every well of 96-well polystyrene microplates that were precoated with a mouse monoclonal antibody, followed by 50 µL plasma and incubated for 2 hours at ambient temperature. After incubation and washing procedure, horseradish peroxidase-conjugated polyclonal antibodies against uPAR were admixed and incubated for 2 hours at ambient temperature. After repeated washing, a substrate solution was applied to every well and incubated for further 30  www.nature.com/scientificreports www.nature.com/scientificreports/ while protecting it from light. Finally, 50 µL of stop solution was admixed in every well and the absorbance was detected at 450 nm over 30 minutes using an ELISA reader 1,34 . The intra-assay variation of the kit is given as 7.5% and the inter-assay variation as 5.9%. Samples were randomly distributed between kits of five different lot numbers and were all measured in duplicates. All samples had duplicate coefficients of variants (CVs) <10%.

Study procedures.
Laboratory diagnostic for routine parameters. All laboratory parameters were determined at the Department of Laboratory Medicine, Medical University of Vienna, which has a certified (ISO 9001:2008) and accredited (ISO 15189:2008) quality management system. The eGFR was calculated using the chronic kidney disease EPI equation 35 .
Outcomes. The current study investigated the association of FSGS status and suPAR levels in plasma measured by different ELISA kits. Furthermore, a prognostic assessment of suPAR was performed for eGFR dynamics, ESRD and death.
Statistical analyses. Categorized data are presented as absolute counts and relative frequencies (percent).
We presented continuous data as means ± standard deviation (SD). To compare baseline variables between study groups we used the Mann-Whitney U test, the median test or the Fisher's exact test, as appropriate. To compare difference in suPAR levels from different test kits we used the t-test with H 0 : delta = 0, calculating disease group specific estimates with respective 95% confidence intervals.
We calculated standard metrics of diagnostic test accuracy including sensitivity and specificity of suPAR levels (index tests) for identifying patients with FSGS (reference standard) independently. We estimated the optimal cut-off by maximizing the sensitivity-specificity product according to Liu's method 36 using the adjustment method suggested by Fluss et al. 37 . The confidence interval for the cut-off value was estimated by bootstrapping.
For prognostic assessment of suPAR (main co-variable) for eGFR dynamics, ESRD, and death (individual outcomes) multivariable regression analyses were employed. For eGFR dynamics linear regression, for ESRD and death logistic regression were used. Co-variables were sex (male vs female), age (years), eGFR (mL/min) and initial urine protein/creatinine ratio (mg/g). We report the estimates from the regression analysis with a 95% confidence interval and the corresponding P-values from the Wald test. We furthermore conducted a sensitivity analysis in which differences in eGFR, urine protein/creatinine ratio and albumin/creatinine ratio values were set to zero after loss of renal function.
Statistical analysis and data management was carried out with Stata 14 for Mac (Stata Corp, College Station, TX, USA), GraphPad Prism (© GraphPad Prism version 7.00 for Windows, GraphPad Software, La Jolla California, USA) and MS Excel (© Microsoft, Redmont, WA).
All tests conducted were two-sided and a P-value below 0.05 was considered significant. Plasma suPAR level depending on kidney function. The mean eGFR was significantly lower in FSGS patients compared to patients with other GN or healthy volunteers, which might partially account for the difference of suPAR levels between groups. Multiple regression analysis showed that baseline suPAR level was significantly correlated with eGFR (p = 0.001) which remained significant after multivariable adjustment as shown in Table 2 Table 3). Characteristics and laboratory parameters of FSGS patients with eGFR above 60 mL/min and below are presented in Supplementary Table S2.

Study population.
Sensitivity and specificity of plasma suPAR for FSGS diagnosis. We performed a receiver operating characteristic (ROC) analysis to determine whether suPAR is a potent marker to diagnose and differentiate FSGS.
By identifying the point at which the difference between sensitivity and 1-specificity is maximal, we determined that the optimal cut-off value is 4.644 ng/mL of the suPAR concentration measured by the suPARnostic ™  (Fig. 2b).
These results suggest that both kits measuring suPAR can be used to distinguish FSGS from non-FSGS, with the suPARnostic ™ assay performing a higher differentiation in our analysis.
An illustration of suPAR cut-off values and possible false negative test results in patients with FSGS depending on the test method used is shown in Supplementary Fig. S2.
The ROC analysis of a subanalysis of subjects with eGFR ≥60 mL/min is presented in Supplementary Fig. S3. www.nature.com/scientificreports www.nature.com/scientificreports/ CI 1.01 to 1.77), p = 0.046. Corresponding results obtained with the Quantikine Human uPAR ELISA Kit are also given in Supplementary Table S3. A graphical depiction of increasing plasma suPAR levels and progression to ESRD is presented in Supplementary Fig. S4.

Prognostic assessment of plasma suPAR levels.
Similar results were obtained in a model used as sensitivity analysis in which differences in eGFR, urine albumin/creatinine ratio and protein/creatinine ratio values were set to zero after loss of renal function compared to the original model using missing data. (Supplementary Table S3).

Discussion
This study was conducted to test the diagnostic and prognostic value of suPAR as a biomarker for FSGS. Since conflicting results about the clinical role of suPAR in the literature are derived from different ELISA methods measuring suPAR plasma levels, two of the leading assays were used in a side-by-side comparison and their results were compared in this study.
Our study analyzed plasma suPAR levels in patients with different forms of FSGS (primary and secondary FSGS as well as recurrence after transplantation), other types of glomerulonephritis and healthy subjects. Furthermore, suPAR was assessed according to estimated glomerular filtration rate in patients with and without chronic kidney disease. The prognostic evaluation was performed over a period of five years with respect to renal retention parameters, proteinuria, disease progression and mortality. Our data show that FSGS patients have significantly higher suPAR plasma levels when compared to patients with other forms of GN or healthy controls indicating the significance of suPAR tests in FSGS diagnosis. Furthermore, elevated suPAR values in FSGS patients were found to have a predictive value for disease progression to ESRD. Our results also demonstrate that both commercially available suPARnostic ™ assay and the Quantikine Human uPAR Immunoassay can be used to differentiate FSGS from non-FSGS based on suPAR plasma levels, with the suPARnostic ™ assay being more FSGS specific. Novel test specific cut-offs of 4.644 ng/mL for the suPARnostic ™ assay and 2.789 ng/mL for the Quantikine Human uPAR Immunoassay were established for FSGS diagnosis in the entire study population.   A number of FSGS studies have used different test methods for the quantitative measurement of suPAR and came to different conclusions. It is noteworthy that fragments of suPAR are not recognized equally by the currently available suPAR ELISA test methods. The separation of uPAR from its GP anchor releases suPAR and further suPAR fragments are generated by additional cleavage between the D1, D2/D3 domains 17 . Furthermore, depending on the amount of glycosylation, the size of the cleaved suPAR fragments ranges from 25 to 50 kDa 17 . This is of central importance: suPAR fragments differ in regard to their biological relevance and are not equally detected in different suPAR assays leading to varying results which are hard to compare among studies 5,38,39 . In our study two ELISA tests were used, the suPARnostic ™ assay and the Quantikine Human uPAR immunoassay. While the suPARnostic ™ assay represents a double monoclonal antibody sandwich assay, the Quantikine Human uPAR immunoassay uses a monoclonal capture antibody and a polyclonal detection antibody. Furthermore, the suPARnostic ™ assay measures both full and cleaved forms of suPAR, which might explain variations in test readings. Thus, although both ELISA kits tested in our study were capable to differentiate FSGS from non-FSGS, test results were not completely identical: in the entire study population of 146 subjects the suPARnostic ™ assay outperfomed the Quantikine Human uPAR-Immunoassay in differentiating FSGS from non-FSGS. This was based on higher suPAR levels in patients with FSGS and lower suPAR levels in healthy volunteers when compared with the Quantikine Human uPAR-Immunoassay. This difference between the two test kits used was attenuated in subjects with eGFR above 60 mL/min.
Our study demonstrates the importance of suPAR testing in the diagnosis of FSGS. In previous studies, the correlation of suPAR with renal insufficiency was repeatedly discussed critically 40 . This led us to analyze suPAR values in patients with chronic kidney disease (according to the chronic kidney disease (CKD) definition of GFR below 60 mL/min) as well as in patients without CKD. The inverse association of suPAR with eGFR was confirmed in our study; moreover the significance of suPAR in FSGS assessment independent of renal function could be emphasized.
SuPAR may not only be a diagnostic biomarker for FSGS, but also a pathogenic factor for progression of disease 26 . Previously, it has been shown that proteinuria can be reduced by eliminating suPAR from the circulation www.nature.com/scientificreports www.nature.com/scientificreports/ using extracorporeal methods, e.g. plasmapheresis and immunoadsorption. Recently, also hemoadsorption by means of polymeric adsorber columns have been used to remove suPAR 38 . As suPAR has a high affinity to protein A immunoapheresis columns, it is removable from the circulation, at least in part, but also rebounds within days 16 . An established therapy for recurrence of FSGS after transplantation is plasmapheresis or immunoadsorption, in combination with steroids, cyclosporine A and rituximab or alone 15,38 .
Recurrence of FSGS after transplantation manifests in about 30% of FSGS patients receiving their first kidney transplant 2 and in up to 80% of patients after second transplantation with a history of transplant loss due to recurrence of FSGS in the first graft 41 . This can even manifest within hours post transplantation 42 . Since the recurrence of disease is a serious problem, there are attempts to reduce recurrence rates by using induction therapies such as rituximab prior to transplantation in patients with FSGS as underlying disease with high suPAR values 43 . Accordingly, suPAR values in kidney transplant candidates where a high recurrence rate can be assumed could serve as guide to a patient-tailored immunosuppressive therapy, but randomized controlled trials are needed to support this hypothesis.
Furthermore, suPAR measurement could be particularly useful in other clinical situations, where a kidney biopsy for diagnosis is not possible. With the aid of other established biomarkers such as anti-PLA2R antibodies 44 , different forms of GN (membranous nephropathy versus FSGS) can be distinguished, which has in turn therapeutic consequences. Thus, suPAR is of manifold interest in FSGS pathogenesis, diagnosis, therapy and prognosis.
The main limitation of this study is its retrospective design. In addition, our sample size was relatively small due to the monocentric nature of the study. Furthermore, no urine samples were taken that would have allowed further potentially significant analyses 34,45 . Our study was strengthened by a well-characterized cohort of patients with primary FSGS, secondary FSGS, recurrence of FSGS after transplantation, different forms of glomerulonephritis and healthy controls with long follow-up resulting in clear results: plasma suPAR levels do not distinguish between patients with primary and secondary form of FSGS and recurrence of disease after renal transplantation, which is in line with results from previous studies 46 . But suPAR levels are useful to differentiate FSGS patients from patients with other forms of glomerulonephritis and healthy controls, e.g. when renal biopsies are difficult. The independence of suPAR from eGFR has been demonstrated and both the suPARnostic ™ assay and the Quantikine Human uPAR-Immunoassay are capable for FSGS identification. Furthermore, the prognostic value of suPAR for the progression to end-stage renal disease was identified.
The scope of this study was not to establish the role of suPAR in the pathogenesis of FSGS. The main aim was to examine the significance of suPAR in FSGS diagnosis and different courses of disease. Hereby, we describe new cut-offs for suPAR using two different suPAR ELISA assays. The comparison of these two assays, showing relevant differences, may explain varying results obtained in previous studies using different detection assays. Therefore, suPAR values have to be interpreted in the context of population and the test method used. Knowledge of test specific cut-offs makes suPAR a valuable biomarker for FSGS.

Data Availability
The datasets generated in the context of this paper are made available by the corresponding author on request.