Association of Stromal Cell-derived Factor-1 With Diabetic Kidney Disease in Type 2 Diabetic Patients

Background: The present study was designed to explore whether serum stromal cell-derived factor-1 (SDF-1) levels were associated with albuminuria, estimated glomerular ltration rate (eGFR) and diabetic kidney disease (DKD), and detect which clinical parameters might affect serum SDF-1 levels in patients with type 2 diabetes (T2D). Methods: Serum SDF-1 levels were measured by sandwich ELISA. Patients with an eGFR < 60ml/min/1.73m 2 and/or a urinary albumin-to-creatinine ratio (UACR) ≥ 30mg/g who presented with diabetic retinopathy were identied as having DKD. Results: Serum SDF-1 levels in T2D patients were signicantly higher than those in healthy controls (p < 0.05). Urinary albumin and UACR were positively correlated with serum SDF-1 levels (r = 0.216 and = 0.276, respectively, p < 0.01), and eGFR was inversely related with serum SDF-1 levels (r = -0.368, p < 0.001). Moreover, after adjusting for other clinical covariates by multiple linear regression analyses, the serum SDF-1 levels were independently associated with urinary albumin (β = 0.071, t = 2.185, p < 0.05), UACR (β = 0.071, t = 2.077, p < 0.05) and eGFR (β = -3.975, t = -3.375, p < 0.01). Furthermore, receiver operating characteristic analysis indicated that the optimal SDF-1 cutoff value for predicting macroalbuminuria was 5.735 ng/mL (its corresponding sensitivity was 50.00% and specicity was 81.46%), for predicting abnormal albuminuria was 4.321 ng/mL (its corresponding sensitivity was 58.46% and specicity was 70.78%) and for predicting DKD was 3.505 ng/mL (its corresponding sensitivity was 83.33% and specicity was 42.86%). present study was to evaluate whether serum SDF-1 levels were related to albuminuria, eGFR and DKD. We also assessed which factors presumably inuenced serum SDF-1 levels. Our results demonstrated that serum SDF-1 levels might have the ability to be a therapeutic target and marker for DKD. Three multiple stepwise linear regression analyses were used to explore to explore the associations of SDF-1 with lg (urinary albumin), SDF-1 with lg (UACR), and SDF-1 with eGFR, adjusting for age, gender, diabetic duration, blood pressure, BMI, glycosylated hemoglobin A1c (HbA1c), serum lipid and antidiabetic treatment, as these parameters might affect urinary albumin excretion. Furthermore, receiver operating characteristic (ROC) analysis was conducted to analyze the ability of SDF-1 to indicate macroalbuminuria, abnormal albuminuria and DKD cases, and the cutoff values of SDF-1 to indicate macroalbuminuria, abnormal albuminuria and DKD are provided. Data analyses were performed using SPSS statistical software 18.0 (IBM SPSS Inc., USA). A value of p < 0.05 was considered to be statistically signicant. islet and eventually lead to the occurrence of T2D [19]. SDF-1 may accelerate the progression of T2D.


Laboratory examination and calculation
Fasting blood samples were collected to measure laboratory parameters. We also collected fresh morning rst-void urine samples from type 2 diabetic participants for measurement of urinary albumin and urinary creatinine. UACR was calculated as the ratio of urinary albumin and urinary creatinine. According to UACR, normoabluminuria, microalbuminuria and macroalbuminuria were de ned as UACR < 30mg/g, UACR: 30-300mg/g and UACR > 300mg/g, respectively [13]. eGFR was calculated based on the CKD-EPI creatinine-cystatin C equation (2012) [14], and DKD was de ned as an eGFR < 60ml/min/1.73m2 and/or a UACR ≥ 30mg/g who presented with diabetic retinopathy [15]. All blood samples were centrifuged and stored at −80°C. Serum SDF-1 levels were measured by sandwich ELISA (Human SDF-1/CXCL12 Elisa Kit; Elabscience, Wuhan, China). The intra-and interassay coe cients of variation were both less than 10.0%.

Statistical analyses
Clinical variables are shown for normal controls, type 2 diabetic subjects, and for the quartiles of serum SDF-1 levels. The mean ± SD and frequencies (percentages) were adopted to describe normally distributed continuous variables and categorical variables, respectively. Urinary albumin and UACR were log transformed to achieve a normal distribution. We adopted appropriately the one-way analysis of variance (ANOVA) test to compare differences in normally distributed data, the Kruskal-Wallis test to compare differences in skewed distributed data and the chi-square test to compare categorical data among the four subgroups based on the SDF-1 quartiles. The correlations of SDF-1 with lg (urinary albumin), SDF-1 with lg (UACR), SDF-1 with eGFR, and SDF-1 with other clinical parameters were analyzed by pearson's or spearman's bivariate correlation analysis as appropriate. Three multiple stepwise linear regression analyses were used to explore to explore the associations of SDF-1 with lg (urinary albumin), SDF-1 with lg (UACR), and SDF-1 with eGFR, adjusting for age, gender, diabetic duration, blood pressure, BMI, glycosylated hemoglobin A1c (HbA1c), serum lipid and antidiabetic treatment, as these parameters might affect urinary albumin excretion. Furthermore, receiver operating characteristic (ROC) analysis was conducted to analyze the ability of SDF-1 to indicate macroalbuminuria, abnormal albuminuria and DKD cases, and the cutoff values of SDF-1 to indicate macroalbuminuria, abnormal albuminuria and DKD are provided. Data analyses were performed using SPSS statistical software 18.0 (IBM SPSS Inc., USA). A value of p < 0.05 was considered to be statistically signi cant.
ROC analysis to explore the cutoff SDF-1 value to predict macroalbuminuria and abnormal albuminuria ROC analysis was further applied to explore the cutoff SDF-1 value to indicate macroalbuminuria, abnormal albuminuria and DKD cases. The optimal cutoff value of SDF-1 to predict macroalbuminuria was 5.735 ng/mL, to predict abnormal albuminuria was 4.321 ng/mL and to predict DKD was 3.505 ng/mL. The corresponding AUC to predict macroalbuminuria was 0.671 (95% CI 0.626-0.816), and its Youden index was 0.315, its sensitivity was 50.00%, and its speci city was 81.46% (Fig. 2). The corresponding AUC to predict abnormal albuminuria was 0.639 (95% CI 0.551-0.726), and its Youden index was 0.292, its sensitivity was 58.46%, and its speci city was 70.78% (Fig. 3). The corresponding AUC to predict DKD was 0.654 (95% CI 0.536-0.773), and its Youden index was 0.262, its sensitivity was 83.33%, and its speci city was 42.86% (Fig. 4).

Discussion
In the present study, we compared the serum SDF-1 levels between the type 2 diabetic patients and healthy controls, analyzed the associations of serum SDF-1 levels with urinary albumin, UACR and eGFR in Chinese type 2 diabetic patients. The main ndings of this study are as follows: rst, compared with normal controls, serum SDF-1 levels were higher in patients with T2D; second, urinary albumin, UACR and DKD incidence were positively related with serum SDF-1 levels, while eGFR was negatively related with serum SDF-1 levels; third, serum SDF-1 levels were positively associated with HbA1c, D-dimer, ESR, NEU, and negatively associated with APTT; fourth, after adjusting for other clinical covariates, the serum SDF-1 levels were independently and positively associated with urinary albumin and UACR, and inversely associated with eGFR in patients with T2D; and fth, the optimal SDF-1 cutoff value for predicting macroalbuminuria was 5.735 ng/mL (its corresponding sensitivity was 50.00% and speci city was 81.46%), for predicting abnormal albuminuria was 4.321 ng/mL (its corresponding sensitivity was 58.46% and speci city was 70.78%) and for predicting DKD was 3.505 ng/mL (its corresponding sensitivity was 83.33% and speci city was 42.86%).
The present study demonstrated that serum SDF-1 levels were signi cantly higher in type 2 diabetic patients than normal controls, and were positively associated with HbA1c. Similar with our study, R. Derakhshan et al revealed that plasm SDF-1 levels were higher in gestational diabetes mellitus mothers than in normal pregnancy mothers [16], and higher in type 2 diabetic patients than in normal controls [17]. SDF-1 and its receptor CXCR4 are expressed in both islet alpha-and beta-cells [18], and the SDF-1/CXCR4 axis may induce islet in ammation by attracting in ammatory cells to the islet and eventually lead to the occurrence of T2D [19]. Hence, SDF-1 may accelerate the onset and progression of T2D.
T2D is capable to result in chronic in ammation characterized by activated mononuclear phagocyte system and increased secretion of cytokines in vivo [20], and in ammation can promote hypercoagulability through the mechanism that cytokines can stimulate the release and expression of procoagulant molecules and inhibit the expression of anti-coagulant molecules [21]. It was shown that serum SDF-1 levels were positively associated with ESR, NEU, D-dimer and negatively associated with APTT in this study. ESR and NEU are recognized in ammatory biomarkers, while APTT and Ddimer are important indicators of coagulation function. A shorter APTT re ects enhanced endogenous coagulation function [22] and D-dimer is a brin degradation product [1]. In patients with disseminated intravascular coagulation (DIC), the coagulation activation was able to contribute to increased levels of the circulating SDF-1, which in turn enhanced platelet aggregation and then promoted coagulation [23]. Given these results, elevated serum levels of SDF-1 presumably re ected the condition of hyperglycemia, in ammation and hypercoagulability in type 2 diabetic patients.
As DKD is a result of the interaction of hyperglycemia, hemodynamic alterations, in ammation and oxidative stress [24], SDF-1 may play a critical role in the development and progression of DKD, or at least can serve as a potential predictor of DKD. In support of this conclusion, renal biopsy revealed that SDF-1 signi cantly increased in kidneys of diabetic rodents and patients with DKD [25], and blockade of SDF-1 by the speci c inhibitor NOX-A12 signi cantly reduced podocyte loss and glomerulosclerosis, thereby reducing proteinuria [26].
In the early stage of DKD, renal alterations are manifested as glomerular hypertrophy, glomerular basement membrane (GBM) thickening, podocyte loss and tubular damage. In advanced DKD, renal morphological changes include glomerulosclerosis and tubulointerstitial brosis, and corresponding clinical features include renal ltration function declines with or without albuminuria [27]. Our study showed that serum SDF-1 levels were positively associated with urinary albumin and UACR, and negatively associated with eGFR. UACR, an evaluation index of increased urinary albumin, can re ect damage to the basement membrane and endothelium of glomerular capillaries, and represent early changes of DKD [28]. Based on the CKD-EPI creatinine-cystatin C equation, eGFR can accurately estimate renal function [14]. Thus, serum SDF-1 levels may have the potential to be an evaluation indicator of early and advanced DKD. Recent researches strongly suggest that renal tubular damage is a vital component of early DKD and may precede the occurrence of glomerular injury [29]. Serum cystatin C, a non-glycosylated, low molecular weight and basic protein, is completely removed by glomerular ltration and subsequently reabsorbed and degraded by proximal tubular [30]. In addition to the ability to assess GFR, cystatin C can also re ect renal tubular damage of DKD [31]. In this study, serum SDF-1 levels were signi cantly associated with cystatin C (r = 0.330, p < 0.001). Therefore, serum SDF-1 levels possibly and roundly re ect renal alterations of DKD patients, which is consistent with that SDF-1 is expressed in podocytes and distal tubular cells of human kidney [9]. Several limitations of our study should be addressed. First, the present study could not explain the causal relationship between SDF-1 and albuminuria due to the common problem of cross-sectional studies. Second, on account of the small sample size of this study, the correlation between DPP-4 inhibitors use and plasma SDF-1 levels could not be veri ed. Third, all the subjects enrolled in this study were Chinese, which limited the wide applicability of our study. Therefore, further research should be conducted to validate the results of our study and to address the above limitations.

Conclusions
The serum SDF-1 levels were positively associated with urinary albumin, UACR and cystatin C, and negatively associated with eGFR, which indicate that SDF-1 may play a critical role in the onset and progression of DKD.    Figure 1 The proportion of albuminuria types stratifed by SDF-1 quartiles