Early, Noninvasive Clinical Indicators of Kidney Prognosis in Primary Nephrotic Syndrome: A Retrospective Exploratory Study

This retrospective exploratory study aimed to identify early clinical indicators of kidney prognosis in primary nephrotic syndrome (NS). Univariate Cox proportional hazards regression analysis identified clinical parameters in the 2-month period after initiating immunosuppressive therapy (IST); it predicted 40% reduction in the estimated glomerular filtration rate (eGFR) in 36 patients with primary NS. Time-dependent receiver operating characteristic curve analysis was used to evaluate the performance of the predictors for the cumulative incidence of 40% reduction in the eGFR up to 8 years after initiating IST. The mean follow-up period was 71.9 months. The eGFR was reduced by 40% in four patients. Significant predictors for time to 40% reduction in the eGFR were as follows: an increase in the serum soluble urokinase plasminogen activator receptor (s-suPAR) 2 months after initiating IST (Δs-suPAR (2M); hazard ratio (HR) for every 500 pg/mL increase: 1.36, P=0.006), s-suPAR at 2 months after initiating IST (s-suPAR (2M); HR for every 500 pg/mL increase: 1.13, P=0.015), urinary protein-to-creatinine ratio (u-PCR) (u-PCR (2M); HR for every 1.0 g/gCr increase: 2.94, P=0.003), and urinary liver-type fatty acid-binding protein (u-L-FABP) (u-L-FABP (2M); HR for every 1.0 μg/gCr increase: 1.14, P=0.006). All four factors exhibited high predictive accuracy for cumulative incidence of 40% reduction in the eGFR up to 8 years after initiating IST, with areas under the receiver operating characteristic curve of 0.92 for Δs-suPAR (2M), 0.87 for s-suPAR (2M), 0.93 for u-PCR (2M), and 0.93 for u-L-FABP (2M). These findings suggest that Δs-suPAR (2M), s-suPAR (2M), u-PCR (2M), and u-L-FABP (2M) could be useful indicators of initial therapeutic response for predicting kidney prognosis in primary NS.


Introduction
e status of nephrotic syndrome (NS, i.e., urinary protein concentration) at 6 months after initiating immunosuppressive therapy (IST) is an important factor for predicting kidney prognosis in NS cases [1][2][3]. In a large cohort study on NS in Japan, patients in whom urinary protein could not be reduced to <1 g/day even after 6 months of IST had significantly poorer kidney prognosis than those in whom urinary protein was reduced to <1 g/day. Furthermore, patients with uncontrolled urinary protein levels were at higher risk of developing end-stage kidney disease (ESKD) [3]. e Japanese Society of Nephrology defines intractable NS as NS in which urinary protein levels cannot be reduced to <1 g/day even after 6 months of IST [4]. A recent epidemiological research study in Japan found death from infectious diseases to be a critical survival prognostic factor for primary NS [5]. is finding highlights the necessity of predicting kidney prognosis before or at the earliest period after the start of initial treatment to establish individualized treatment protocols based on each patient's kidney prognosis and risk of infectious disease.
In this regard, indiscriminately administering primary treatment for 6 months only to identify intractable NS can be insufficient. In clinical practice, kidney prognosis should be predicted using valuable biomarkers at 1-2 months after commencing treatment to determine the effectiveness of each patient's management plan [4]. However, there are no validated biomarkers of treatment effectiveness within the first 2 months of IST in relation to kidney prognosis. erefore, this study aimed to investigate the feasibility of using baseline clinical parameters to predict kidney prognosis before initiating IST. Furthermore, we aimed to determine whether changes in these clinical parameters 2 months after initiating IST can be used to predict kidney prognosis and ultimately clarify whether response to initial treatment can be predictive of kidney prognosis.

Study Design and Participants.
In this retrospective exploratory study, we evaluated 47 patients with primary NS who were diagnosed using kidney biopsy at our hospital between January 2006 and December 2017. After excluding 11 patients who did not receive IST (all of whom had membranous nephropathy (MN)), 36 patients who received IST were included in the final analysis. Of these, 18 had minimal-change NS (MCNS), seven had focal segmental glomerulosclerosis, nine had MN, and two had membranoproliferative glomerulonephritis. e patients' clinical data were obtained from their medical records and laboratory test reports, all of which were stored in an electronic medical record system. is study was approved by the Kanazawa Medical University Ethics Committee (No. I285) and conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. All patients provided written informed consent for participation.

Estimation of Daily Urinary Protein and Definition of Treatment Response.
Daily urinary protein level was estimated as the urinary protein-to-creatinine ratio (u-PCR) in voided urine. Proteinuria remission was categorized as complete remission (CR, i.e., u-PCR <0.3 g/gCr), incomplete remission type 1 (ICR-1, i.e., u-PCR <1.0 g/gCr), and incomplete remission type 2 (ICR-2, i.e., u-PCR <3.5 g/gCr). Intractable NS was defined as NS in which u-PCR does not decrease to <1 g/gCr despite treatments, including steroids and immunosuppressors, for 6 months. ese definitions were based on the NS clinical practice guidelines of the Japanese Society of Nephrology [4].

Statistical Analysis.
For the main analysis, the time to 40% reduction in the eGFR, which is a surrogate marker of ESKD [7][8][9][10], was set as the outcome variable. Meanwhile, explanatory variables included clinical parameters at the baseline and at 2 months after initiating IST and the amount of change over 2 months (Δ). ese were used in univariate Cox proportional hazards regression analysis to identify significant predictors of the time to 40% reduction in the eGFR. e following clinical parameters were examined at the baseline: age, sex, kidney tissue diagnosis (MCNS or non-MCNS), u-PCR, selectivity index, serum (s-) and urinary (u-) suPAR, u-L-FABP, eGFR, s-albumin, total cholesterol, and use of CyA in initial treatment. Two months after initiating IST, changes in the eGFR (ΔeGFR (2M)), u-PCR (Δu-PCR (2M)), u-L-FABP (Δu-L-FABP (2M)), and s-and u-suPAR (Δs-suPAR (2M), and Δu-suPAR (2M), respectively) were examined. Additional factors such as eGFR (2M), u-PCR (2M), u-L-FABP (2M), s-suPAR (2M), and u-suPAR (2M) were also evaluated. Timedependent receiver operating characteristic (ROC) curve analysis [11,12] using the area under the ROC curve (AUC) was performed to evaluate the predictive accuracy of the parameters for the cumulative incidence (CI) of 40% reduction in the eGFR by 2, 3, 4, 5, 6, 7, and 8 years after initiating treatment. Cutoff values were selected to maximize the sum of sensitivity and specificity.
In the secondary analysis, the outcome variable set was the achievement of ICR-1 (i.e., presence/absence of intractable NS) by 6 months after initiating IST. e same clinical parameters used in the main analysis were used as explanatory variables in univariate logistic regression analysis to identify significant predictors of intractable NS. e performance (AUC) of the predictors for intractable NS identified in the secondary analysis was evaluated using ROC curve analysis, and the cutoff values were set to maximize the sum of sensitivity and specificity.
Continuous variables are presented as means with standard deviation (SD) for normally distributed data and as the medians with interquartile range (IQR) for nonnormally distributed data. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [13]. All P values were two-sided, and P < 0.05 was considered statistically significant.
e P values reported in this manuscript are nominal values.

Baseline Characteristics.
e mean patient age was 53.8 years, and the male-to-female ratio was 5 : 4. e baseline patient characteristics are summarized in Table 1. ere were an equal number of patients with respect to the renal tissue type (18 patients each with MCNS and non-MCNS). Hypoalbuminemia (i.e., mean serum albumin <1.8 g/dL) due to hyperproteinemia (mean u-PCR, 10.9 g/gCr) was observed in all patients. e mean eGFR was 59.8 mL/min/ 1.73 m 2 . CyA was used for initial treatment in 29 (80.6%) patients.

Predictive Factors of Time to 40% Reduction in eGFR.
e mean follow-up period in the study cohort was 71.9 months. ere were four patients (11.1%) who showed 40% reduction in the eGFR; this was achieved at 80 months in a male patient in his 70s, at 17 months in an MCNS male patient in his 80s, and at 67 months in an MN male patient in his 70s and in an MN female patient in her 50s. In the main analysis, univariate Cox proportional hazards regression analysis showed that Δs-suPAR (2M), s-suPAR (2M), u-PCR (2M), and u-L-FABP (2M) were significant predictors of 40% reduction in the eGFR ( Table 2). No baseline clinical parameter showed significance. Figure 1 shows the results of the time-dependent ROC curve analysis of the predictive accuracy (AUC) for the CI of 40% reduction in the eGFR by X years after initiating IST (X � 2, 3, 4, 5, 6, 7, and 8). e cutoff value for each parameter is shown in Table 3, and u-PCR (2M) and u-L-FABP (2M) showed stable and high predictive accuracies. e accuracies of Δs-suPAR (2M) and s-suPAR (2M) for predicting the CI of 40% reduction in the eGFR by 2-5 years after initiating IST were lower than those of the u-PCR (2M) and u-L-FABP (2M). However, the accuracies of Δs-suPAR (2M) and s-suPAR (2M) for predicting the CI of 40% reduction in the eGFR by 6-8 years after initiating IST were closer to those of the u-PCR (2M) and u-L-FABP (2M). e cutoff values of u-PCR (2M) that predict the CI of 40% reduction in the eGFR were 3.9 g/gCr (nearly ICR-2 level), 0.9 g/gCr (nearly ICR-1 level), and 0.4 g/gCr (nearly CR level) by 2, 3-5, and 6-8 years after initiating IST, respectively.
e cutoff values of s-suPAR (2M) that predict the CI of 40% reduction in the eGFR ranged from 3000-4000 pg/mL at all time points.

Predictive Factors for Intractable NS.
Using the same clinical parameters used in the main analysis, Δs-suPAR (2M), s-suPAR (2M), u-PCR (2M), and u-L-FABP (2M) were found to be significant predictors of intractable NS (secondary analysis, Table 4). Table 5 shows the predictive accuracies (AUC) and cutoff values of these clinical parameters for intractable NS. Particularly, s-suPAR (2M) and u-PCR (2M) had high predictive accuracies (AUC). s-suPAR, which did not improve to at least 3000 pg/mL, and u-PCR, which did not improve to ICR-1 levels after 2 months of IST, were found to be predictive factors for intractable NS.

Discussion
Valuable early clinical indicators of kidney prognosis in primary NS are limited. is study found that a reduction of u-PCR (2M) values to ICR-2, ICR-1, and CR levels could suppress the CI of 40% reduction in the eGFR by 2, 3-5, and 6-8 years after initiating IST, respectively. ese findings suggest that reducing u-PCR (2M) to a lower level will reduce the long-term risk of kidney function deterioration. To the best of our knowledge, this study is the first to report on the feasibility of using Δs-suPAR (2M), s-suPAR (2M), u-PCR (2M), and u-L-FABP (2M) to define initial treatment response and predict long-term kidney prognosis in primary NS.
Interestingly, this study revealed that a 3000 pg/mL cutoff value of s-suPAR (2M) predicted the occurrence of 40% reduction in the eGFR within 8 years of initiating IST, which was at the same level as the threshold of s-suPAR that induced podocyte injury in a previous study [14]. s-suPAR induces β3 integrin-dependent podocyte injury and is involved in the onset of proteinuria in NS [14,15]; it has also attracted attention as a molecule involved in the onset and progression of chronic kidney disease (CKD) [16,17]. suPAR is involved in CKD progression through the following mechanisms: it causes podocyte injury, which in turn causes glomerulosclerosis and proteinuria that lead to tubulointerstitial injury [14][15][16][17]. In addition, suPAR that crosses into the urine via glomerular filtration directly causes tubulointerstitial injury [18], which is a common pathway of CKD progression [19] that causes renal function deterioration. erefore, increasing suPAR levels would affect tubulointerstitial injury directly and/or indirectly.
words, we believe that s-suPAR levels reflect both podocyte damage caused by s-suPAR and tubulointerstitial injury caused by u-suPAR, while u-suPAR levels reflect only the latter. Further, we found that the u-PCR and u-L-FABP were superior to s-suPAR as predictors of kidney prognosis. is could be because multiple humoral factors [20][21][22][23], including s-suPAR and autoantibodies [24][25][26], are involved in the kidney injury observed in primary NS. is indicates that clinicians should focus on urinary protein (u-L-FABP) levels (and not on suPAR) as they directly reflect the renal damage caused by multiple humoral factors and autoantibodies, enabling more accurate predictions of kidney prognosis. is study has several strengths. First, different statistical methods, namely, Cox proportional hazards regression analysis for 40% reduction in the eGFR [7][8][9][10] and logistic regression for intractable NS [3], identified the same predictors for the two surrogate endpoints of ESKD. is demonstrates the robustness of our results. Second, to the best of our knowledge, this study is the first to compare longitudinal changes in the performance (AUC) of suPAR, L-FABP, and u-PCR as predictors of kidney prognosis in primary NS using time-dependent ROC curve analysis.
ird, the IST protocols used for initial treatment were fairly uniform, minimizing any confounders from differences in the immunosuppression protocols.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.