Urinary CXCL1: A Novel Predictor of IgA Nephropathy Progression

Background IgA nephropathy (IgAN) is the most common form of primary glomerulonephritis worldwide. In recent years, consistent efforts have been made to develop new non-invasive biomarkers for IgAN progression. In our previous in vitro study we found mesangial derived CXCL1 as a contributor for kidney injury, and observed higher urinary CXCL1 levels in patients with IgAN. It implied that the urinary CXCL1 might be a potential biomarker. Methods In the present study, we enrolled 425 IgAN patients with follow-up data and detected their urinary CXCL1 levels at the time of renal biopsy, to explore the predictive value of urinary CXCL1 in IgAN progression. Urinary CXCL1 levels were measured using enzyme-linked immunosorbent assay. Results Urinary CXCL1 levels were associated with presently well established predictors of IgAN progression, including SBP (r = 0.138, p = 0.004), DBP (r = 0.114, p = 0.019), proteinuria (r = 0.155, p = 0.001), eGFR (r = -0.259, p<0.001) and tubular atrophy and interstitial fibrosis (r = 0.181, p<0.001). After adjusted for them, higher levels of urinary CXCL1 were independently associated with a greater risk of deterioration in renal function (HR, per s.d. increment of natural log–transformed CXCL1: 1.748; 95% CI: 1.222–2.499, P = 0.002). Furthermore, time-dependent receiver operating characteristic (ROC) curve showed that urinary CXCL1, when combined with proteinuria and eGFR, could enhance the prognostic value of these traditional predictors for IgAN progression. Conclusions The results in our present study suggested urinary CXCL1 as a new non-invasive predictor of IgAN progression.


Introduction
Immunoglobulin A nephropathy (IgAN) is the most common form of primary glomerulonephritis (GN) worldwide [1], characterized by mesangial IgA deposition. Patients with IgAN may appear with a variety of clinical manifestations and histological lesions. Moreover, the prognosis of IgAN is highly variable, ranging from sustained asymptomatic hematuria to rapid progression to end stage renal disease [2,3]. Therefore, identification of patients at high risk of progression would be valuable for patient's management in clinical practice. At present, clinical prognostic features indicating the well-known predictors for progression in IgAN include sustained hypertension, impaired renal function and persistent proteinuria. Histologically, renal biopsy is a vital means for IgAN diagnosis and outcome prediction. However, renal biopsy still has some limitations. First, it is an invasive examination, which precludes the repeated testing. Second, it reflects renal lesion only at the time point of biopsy, missed its dynamic changes as the disease progressed. Third, since only tens of glomeruli can be used for examination, it should be more cautious to use it to reflect the whole renal lesion, especially in case of focal other than diffused lesions. Thus, non-invasive bio-markers, which can dynamically reflect the whole renal lesion, are in urgent need.
In recent years, consistent efforts have been made to develop new non-invasive biomarker for acute and chronic kidney diseases. Approximately 70% of the urinary proteins are from the kidney and urinary tract, and therefore urine becomes a promising bio-sample for biomarker identification in patients with kidney disease. Recently, there are several studies devoted in urinary biomarker for IgAN, such as epidermal growth factor (EGF) [4], monocyte chemoattractant protein-1 (MCP-1) [5], EGF/MCP-1 ratio [6], complement (C3a and C5a) [7], Laminin G-like 3 and free κ light chains [8], and interleukin (IL)-1β,-2,-17,-6,-10 and Interferon (INF)-γ [5]. This kind of studies greatly enriched our knowledge about IgAN progression, although most of them need further validation because of limited sample size or their retrospective studies.
We previously reported that IgA1 complexes from IgAN patients could induce the up-regulation of chemokine (C-X-C motif) ligand 1 (CXCL1) in mesangial cells. And further in vitro experiments proved the podocyte injury effect induced by mesangial derived CXCL1. Moreover, we observed significantly higher urinary CXCL1 levels in patients with IgAN than those in healthy controls, which implied urinary CXCL1 as a potential bio-marker for IgAN [9]. In the present work, we enrolled a cohort of IgAN patients with regular follow-up and explored whether it could improve the predictive value for IgAN progression when adding urinary CXCL1 onto presently used markers, in order to evaluate the predictive value of urinary CXCL1 as a non-invasive biomarker in IgAN progression.

Study population
The present study recruited 425 IgAN patients with regular follow-up in Peking University First Hospital. The IgAN diagnosis was based upon the presence of dominant IgA demonstration in mesangial area by immunofluorescence, and confirmed by light microscopy and electronic microscopy, as well as the lack of clinical or serological evidence of other inflammatory conditions, such as systemic lupus erythematosus, vasculitis, or Henoch-Schoenlein purpura. During follow-up, patients received the same therapy regimen that be treated according to the Kidney Disease: Improving Global Outcomes (KDIGO) Guideline. As the proteinuria>1g/d, ACE inhibitors or ARBs were provided and dosage-adjusted according to the changes of blood pressure that optimal control target of < 130/80mmHg. Proteinuria is higher than 1g for three to six months without relief, and estimated glomerular filtration rate (eGFR)>50 ml/min/1.73m 2 , steroids are recommended. Steroids, in combination of other immunosuppressive agents, like cyclophosphamide, mycophenolate mofetil or FK506, were appropriate for crescentic IgAN with rapid progression [10,11].
In the present study, we also enrolled 40 patients with minimal change disease (MCD), 40 patients with membranous nephropathy (MN), 40 patients with primary focal segmental glomerulosclerosis (FSGS), and 40 patients with lupus nephritis (LN) as disease controls. Moreover, 74 healthy subjects whose urinary CXCL1 were detected in our previous study were enrolled as healthy controls [9].
The study protocol was reviewed and approved by the Ethics Committee of Peking University and written informed consent was obtained from all participants.

Detection of urinary CXCL1 by ELISA
Early morning urine samples on the day of renal biopsy from recruited 425 patients, as well as on the day of follow-up visits of 6 patients, were centrifuged immediately at 1800rpm and 4°C for 5 minutes. Then supernatants were stored at -80°C until the time of assays. Urinary CXCL1 were quantified by a standard sandwich ELISA assays using the DuoSet human CXCL1 ELISA kits (R&D Systems, Minneapolis, MN, USA), according to the manufacturer's protocol. Urinary CXCL1 were calibrated against urine creatinine before the levels were compared.

The correlation analysis of urinary CXCL1 levels and patients outcomes
At first, the correlation of urinary CXCL1 levels with clinical and pathological indicators at the time of renal biopsy were explored. And then, survival analysis was applied to evaluate whether urinary CXCL1 levels at biopsy could predict IgAN outcome. A composite end point, defined as 50% eGFR decline, ESRD or death, whichever occurred first, was used in the present study. ESRD was defined as eGFR < 15 ml/min per 1.73 m 2 or need for renal replacement therapy (such as hemodialysis, peritoneal dialysis or renal transplantation), for the purpose of this study.

Statistical analyses
Statistical analyses were performed by SPSS software (version 16.0; SPSS, Chicago, IL, USA). Normally distributed quantitative variables were expressed as mean ± standard deviation. For non-normally distributed variables, we used median and interquartile range (IQR). Categorical data were summarized as absolute frequencies and percentages. For Continuous variables, independent-samples t test was used if the data was in normal distribution, and if not, Mann-Whitney or Kruskal-Wallis test was performed. Categorical variables were compared using χ 2 test. Spearman's correlation was applied for analyzing correlation. Factors correlated with urinary CXCL1 were further explored using multivariable linear regression analysis. We derived cumulative kidney survival curves using the Kaplan Meier method, and analyzed differences between curves by log-rank test.
For the analysis of Cox proportional hazards models and time-dependent receiver operating characteristic (ROC) curve, survival package and suvivalROC package in R (Version number: 3.0.2) were used, respectively. Cox proportional hazards models were used to analyze the association of urinary CXCL1 levels and the composite outcome. Results were presented as hazard ratio (HR) and 95% confidence interval (CI). Urinary CXCL1 were highly skewed to the left in our patients, therefore, natural log transformation was used. Urinary CXCL1 was analyzed as a continuous variable with HRs calculated per s.d. increment of natural log-transformed CXCL1. The relationship between CXCL1 and risk of end point was examined in unadjusted and multivariable-adjusted Cox models. Additionally, time-dependent receiver operating characteristic (ROC) curve analysis was conducted to evaluate the prognostic value of CXCL1 for the composite end. Areas under the curve were calculated for CXCL1, eGFR, proteinuria. A two-tailed P-value less than 0.05 was considered statistically significant.

Urinary CXCL1 levels correlated with severity of IgAN
In order to identify the independent factors correlated with urinary CXCL1, we then applied multivariable linear regression model to our data. After adjusted for gender, age, eGFR, systolic blood pressure, diastolic blood pressure and tubular atrophy and interstitial fibrosis, proteinuria was independently associated with urinary CXCL1 levels (β = 0.128, P = 0.010, Table 2).

Variation tendency of urinary CXCL1 during follow-up
In order to evaluate the effect of treatment on urinary CXCL1, we collected sequential urine samples of 6 patients with IgAN. After treatment, all patients presented with varying degrees remission of proteinuria, while some had controlled blood pressure. However, we found that the variation tendency of urinary CXCL1 and proteinuria levels were not always the same. For patients with high baseline urinary CXCL1 levels (case 1 & case 2, Fig. 2A-B), the variation tendency of urinary CXCL1 was almost the same with that of proteinuria. For patients with middle levels of baseline urinary CXCL1 (case 3 & case 4, Fig. 2C-D), although proteinuria relieved a lot, urinary CXCL1 levels just decreased a little. For patients with low levels of baseline urinary CXCL1 (case 5 & case 6, Fig. 2E-F), decreased urinary CXCL1 levels were not observed accompanied with proteinuria remission. Regarding to blood pressure under treatment, the variation tendency of urinary CXCL1 was not the same with that of SBP (Fig. 3).

Urinary CXCL1 levels correlated with progression of IgAN
In order to evaluate the prognostic effect of urinary CXCL1 for IgAN progression, we divided our patients into two groups according to the median level of urinary CXCL1 (18.29 pg/mg). Patients with CXCL1 levels less than 18.29 pg/mg were classified as group A, while others (CXCL1 levels above 18.29 pg/mg) were classified as group B. During follow-up, 11 patients (5.2%) in group A reached the composite outcome compared with 34 patients (16.0%) in group B (P<0.001; Table 3). Of these patients, 9 patients (4.2%) in group A had a 50% decline in eGFR compared with 23 patients (10.8%) in group B (P = 0.010). Moreover, end-stage renal disease occurred in 8 patients (3.8%) in group A and 26 patients (12.2%) in group B (P = 0.001). All of the 5 deaths were in group B (P = 0.061).
Kaplan-Meier survival analysis indicated that renal survival for the composite outcome was significantly lower in group B than that in group A (Log Rank test, p<0.001; Fig. 4). The renal survival rate for patients at first and fifth year were 100.0% and 96.3% for Group A, while 97.2% and 81.3% for Group B.
In Cox proportional hazards model, we at first tested baseline clinical and pathological variables for association with the composite progression outcome. In univariate analyses, lower baseline eGFR, higher proteinuria, higher systolic blood pressure, severer tubular atrophy and interstitial fibrosis, usage of steroids or other immunosuppressive agents were significantly associated with a poor renal outcome (Table 4). An increase in urinary CXCL1 levels was also significantly associated with reduced long term renal survival (HR, per s.d. increment of natural log-transformed CXCL1: 2.047; 95% CI: 1.487-2.816, P < 0.001). Moreover, after adjusting for well-established risk factors for IgAN (including baseline proteinuria, SBP, eGFR, tubular atrophy and interstitial fibrosis, and steroid or other immunosuppressive therapy), higher levels of urinary CXCL1 persistently showed significantly greater risk of kidney failure (HR, per s.d. increment of natural log-transformed CXCL1: 1.748; 95% CI: 1.222-2.499, P = 0.002, Table 5). We also conducted a separated analysis in which CXCL1 level was treated as a categorical variable. In accordance with the result above, the risk of reaching the composite outcome was significantly higher in group B patients than that in group A patients (HR, per s.d. increment of CXCL1: 2.914; 95% CI: 1.417-5.989; P = 0.004).
To evaluate the operating characteristics of urinary CXCL1 level as a prognostic value for IgAN progression, we conducted a time-dependent ROC analysis for urinary CXCL1 level in comparison with proteinuria and reciprocal of eGFR (1/eGFR) level (Fig. 5). The areas under the ROC curve (AUC) for CXCL1 at 24, 48, and 72 months were 0.770, 0.651, and 0.668, respectively, which were comparable to those for proteinuria and 1/eGFR. Furthermore, after combined urinary CXCL1 level, proteinuria and 1/eGFR together, the AUCs were higher than single clinical parameters (proteinuria or 1/eGFR) or in their combination (proteinuria plus 1/eGFR) (Fig. 5).
In order to further evaluate the additive effect of urinary CXCL1 on traditional predictive markers for IgAN progression, we carried out the survival analysis by using both base model (M0: including our previous well known markers for IgAN progression, including eGFR, proteinuria, blood pressure and pathological features) and a new model, in which urinary CXCL1 were added (M1 = M0+CXCL1). The results showed that with addition of urinary CXCL1, the overall performance of the predictive model for IgAN progression improved (R square: M1 Vs M0: 0.182 Vs 0.164, p = 0.002, Table 6).

Discussion
CXCL1, also known as growth-regulated oncogene-alpha, is one of the CXC chemotactic factors [14][15][16]. It plays a key role in inflammation through its receptors, CXCR2, which is expressed on both inflammatory cells and non-inflammatory cells [17][18][19][20], including kidney podocytes [7]. Besides gastric, colon and skin cancers [21,22], up-regulated CXCL1 has been reported in many different types of kidney diseases. In anti-MPO IgG-induced necrotizing  crescentic glomerulonephritis, increased circulating CXCL1 protein levels were demonstrated [23]. Furthermore, elevated expression of CXCL1 was also reported in glomeruli of patients with focal segmental glomerulosclerosis obtained by laser capture microdissection [24]. However, reports regarding urinary CXCL1 and even its effect in IgAN were limited till today. Our present study is the first clinical study evaluating the prognostic significance of urinary CXCL1 level on renal survival conducted in a cohort of IgAN patients with long-term followup. We found that urinary CXCL1 levels were significantly higher in patients with IgAN and LN than in other GN, including MCD, MN and FSGS. Since usually accompanied with other systemic symptoms, LN can be easily differentiated from IgAN. Therefore, although urinary CXCL1 cannot be used as a fully specific biomarker for IgAN among patients with primary nephritis, those with IgAN showed significantly higher urinary CXCL1 levels. Combined with other clinical parameters, urinary CXCL1 might be used as specific biomarker set for IgAN. Of course, the establishment of the specific biomarker set for IgAN needs further exploration and validation in the future studies.
In this study, we found that urinary CXCL1 levels were positively associated with 24h urinary protein excretion, SBP and tubular atrophy and interstitial fibrosis, and negatively associated with eGFR, all of which were well-established predictive markers for IgAN progression. Interestingly, when combined with baseline 1/eGFR and proteinuria, urinary CXCL1 could enhance the prognostic value over each individual clinical marker. The results were consistent with our previous finding in cross-sectional study, IgAN patients with severe clinical and pathological lesions presented with higher urinary CXCL1 levels [9]. In the next multivariable linear regression analysis, we found that 24h urinary protein excretion was independently associated with urinary CXCL1 level. Using in vitro mesangial cell culture model, we previously reported that mesangial derived CXCL1 could induce podocyte loss. Since increasing evidences proved podocyte damage as a leading cause for proteinuria [25], our present finding verified the important role for CXCL1 in podocyte injury in IgAN.
Following our identification that urinary CXCL1 level was independently associated with 24h urinary protein excretion, we further evaluated the dynamic changes of urinary CXCL1 during treatment, focused on those patients who presented with some degrees of proteinuria remission after treatment. Our observation in follow-up patients with sequential urine samples suggested that only for patients with high levels of baseline urinary CXCL1, the fluctuations of urinary CXCL1 levels were synchronized with those of proteinuria, implying that urinary CXCL1 above a certain level might participate in podocyte injury. Thereby, for those with high baseline levels, urinary CXCL1 could be served as a useful marker for not only IgAN progression, but also treatment efficacy, especially regarding proteinuria remission. However, our sequential urine samples were just available in a small proportion of patients. We presently have no chance to evaluate the correlation of proteinuria and urianry CXCL1 during follow-up in all patients, which awaiting further verification in future studies. In order to evaluate the role of urinary CXCL1 in prediction of IgAN progression, COX regression model was used in our study. We demonstrated that urinary CXCL1 could serve as an independent prognostic factor for composite end point of IgAN, even after adjustment for baseline risk factors including proteinuria, eGFR, SBP and tubular atrophy and interstitial fibrosis. This result implied extra value of urinary CXCL1 over already established risk factors for the prediction of IgAN progression. Furthermore, we evaluated the predictive value of urinary CXCL1 in IgAN progression. With addition of urinary CXCL1 onto the traditional predictive markers for IgAN, the overall performance of the predictive model for IgAN progression improved significantly, which suggested that urinary CXCL1, although not fully specific, still could be served as a valuable non-invasive biomarker in predicting IgAN progression.
Regarding the underlying mechanism of CXCL1 in IgAN progression, we had little evidence, which was the limitation of our present study. CXCL1 was famous for its powerful neutrophil chemoattractant activity. However, in IgAN, the infiltrating inflammatory cells are mainly lymphocytes and monocytes, rather than neutrophils. Our previous study proved that up-regulated CXCL1, derived from mesangial cells under the challenge of pathogenic IgA1 complex, would increase podocyte death and reduce podocyte adhesion, and at last lead to podocyte loss [9]. In accordance with our finding, Lai KN et al. reported the decreased expression of nephrin in podocytes in IgAN [26]. In IgAN patients, Hara M et al. observed that cumulative excretion of urinary podocytes could reflect IgA nephropathy progression [27]. The podocyte was regarded as a central cell in glomerular disease. Damage and detachment of podocytes lead to foot processes retraction, resulting in proteinuria, even in glomerulosclerosis in later period. Immune injury induced podocyte loss would develop proteinuria, glomerulosclerosis, and progressive loss of kidney function [28]. Although it is very likely that CXCL1 contributed to IgAN progression through its induction of podocyte injury, direct evidence is still lacking today.
In conclusion, we showed that high urinary CXCL1 level is associated with severe clinical and histological findings, and poor renal prognosis in IgAN. When combined with well-established clinical risk factors, urinary CXCL1 may be a useful non-invasive biomarker for IgAN progression.