Background
Diabetic kidney disease (DKD) resulting from systemic microvascular impairment is traditionally referred to as diabetic nephropathy (DN), remains the leading cause for end-stage renal disease, and is a significant risk for cardiovascular morbidity and mortality [
1,
2]. Within the diabetic kidney, the glomeruli and tubules are subject to damage from hyperglycemia, advanced glycosylation products, activation of inflammatory cytokines, and microalbuminuria, which ultimately develop into renal fibrosis with renal failure [
3,
4]. Once an irreversible lesion occurs in the diabetic kidney, it is unlikely that renal function can recover. Understanding the precise pathogenesis underlying DKD and exploring novel biomarkers for early diagnosis play crucial roles in developing more effective preventive and therapeutic strategies [
5].
At present, the diagnosis of DKD mainly relies on the Kidney Disease Outcome Quality Initiative (KDIGO) guidelines, the standards of which include duration of diabetes (6–10 years) and an increase in the threshold value of the urinary albumin excretion rate [
6]. The utility of the two KDIGO criteria cannot effectively realize the goal for the early diagnosis of DKD. The rate of awareness regarding the duration of diabetes in new cases is not high, especially in less developed areas. Proteinuria, a hallmark of renal injury, has insufficient power to predict and categorize DKD due to low sensitivity and specificity [
7]. A large proportion of renal impairment in diabetes patients is associated with non-proteinuria [
8,
9]. The predictive value and existing threshold for diagnosing abnormal levels of albuminuria have come under question [
10,
11]. Hence, there is a critical need to identify reliable biomarkers for individuals who are at risk for developing DKD that maybe progress to end-stage renal disease [
7].
A growing body of evidence indicates that some novel tubular biomarkers have excellent predictability for chronic kidney disease (CKD) or acute kidney injury (AKI), and may exhibit a promising perspective on the early diagnosis for DKD, such as neutrophil gelatinase-associated lipocalin (NGAL), cystatin C (CysC), and clusterin (CLU) [
12‐
16]. NGAL, a glycoprotein belonging to the lipocalin superfamily, is one of the most commonly studied novel biomarkers for renal impairment. NGAL can be expressed in various cells. The stimuli that induce epithelial damage lead to NGAL high expression and then increase the baseline serum level. NGAL filtered by the glomerulus will be captured by proximal tubules and only a minimal amount is excreted in the urine. Tubular injury results from ischemia, inflammation, and hyperglycemia leads to a decrease in NGAL reabsorption and an increase in NGAL secretion by tubular cells [
17]. CysC is produced in all nucleated cells, and is a 13.4-kDa cysteine protease inhibitor. CysC is freely filtered by glomeruli and is completely taken up by proximal tubular cells without tubular secretion. CysC is not normally present in the urine in significant amounts. Fluctuation in the urinary CysC level is mainly due to decreased reabsorption from injured/dysfunctional tubules [
18]. CLU is a 75 kD-disulfide-linked heterodimeric glycoprotein with multiple biologic functions involving sperm maturation, lipid transportation, complement inhibition, tissue remodeling, and membrane recycling. CLU protein is highly expressed by the TGF-β signaling pathway in renal tubular epithelia after renal injury and can be deposited in the kidney as a component of immune deposits [
19]. Thus, it can be assumed that the level of each marker in urine may change significantly, given that tubular impairment plays an important role in the pathogenesis underlying DKD [
20,
21]. Previous studies have demonstrated that markedly increased levels of biomarkers are positively correlated with albuminuria and reflect the severity of renal damage in diabetic cohorts [
22‐
24] or animal models [
13,
25,
26]. Renal impairment that is mainly caused by hyperglycemia, however, has a different and more complicated pathogenesis compared with renal ischemia, hypoxic damage, or toxic injury, and thus the feasibility and efficacy of these biomarkers warrants further confirmation in clinical studies. To date, the clinically comprehensive performance of novel biomarkers has rarely been reported for the diagnosis of DKD and diabetic albuminuria. Moreover, it is unclear whether or not these markers have sufficient changes for clinical detection before the onset of microalbuminuria.
The goal of this study was to evaluate the diagnostic capacity of urinary NGAL, CLU, and CysC for DKD and diabetic albuminuria in T2DM patients. Furthermore, the temporal profile of the three biomarkers was determined to illustrate the potentially additional superiority in the diagnosis of DKD and diabetic albuminuria.
Discussion
Tubular impairment plays an important role in the pathogenesis underlying DKD [
20,
21]. Microalbuminuria activates renal proximal tubular epithelial cells to induce tubulointerstitial inflammation. In contrast, high glucose levels and diabetic substrates, including advanced glycation end-products, carbonyl intermediates, and ultra-filtered growth factors, trigger a number of signaling pathways to promote tubular cell hypertrophy and the interstitial deposition of chemokines, cytokines, growth factors, and adhesion molecules, which are capable of accelerating further inflammation and fibrosis. The extent of tubulointerstitial injury may ultimately determine the attrition rate of renal function in DKD patients [
20,
30]. Therefore, tubular biomarkers may be crucial as glomerular markers for early diagnosis and stratification of DKD or renal impairment in T2DM.
Previous studies have demonstrated that urinary NGAL, CLU, and CysC are promising biomarkers for tubular injuries in CKDs [
14,
15], AKIs [
12,
31], or nephrotoxic lesions [
32], specifically reflecting proximal and distal, distal, and proximal tubular injuries, respectively; however, the diagnostic performance for renal impairment in T2DM and the temporal characteristics of appearance in the urine have rarely been investigated in clinical subjects. The results of the present study provide new evidence supporting the potential role of novel biomarkers as candidates for DKD diagnostic tools. The values of biomarkers were higher in the DKD group than the non-DKD group (Table
1), all of which were positively correlated with the current golden criteria for UACR (Table
2) and as independent factors for DKD. The levels of biomarkers were significantly increased in patients with renal insults and normoalbuminuria, microalbuminuria, and overt nephropathy (Fig.
1). ROC curve analyses affirmed that candidates exhibited moderate diagnostic performance for DKD and macroalbuminuria, of which uNGAL and uCysC performed better (AUC = 0.816 and 0.803 for DKD, respectively; AUC = 0.855 and 0.894 for macroalbuminuria, respectively; Figs.
2 and
3). In a study comprised of 70 patients with diabetes, the AUC of urinary NGAL was 0.848 for predicting macroalbuminuria, with 70.6% sensitivity and 83.3% specificity at the cut-off value [
33]; the findings are nearly consistent with the present AUC (0.816). A study reported the AUC of uNGAL reached 0.956 for normoalbuminuric DN [
22], but the AUC was not comparable to the present AUC because of the different diagnostic objectives and study cohort. Although uCysC and uCLU are promising biomarkers for diabetic damage of renal tubules, the capability to diagnose DKD has rarely been reported. As shown in Figs.
2 and
3, both biomarkers had moderate accuracy in diagnosing DKD and macroalbuminuria.
The capability of a biomarker to diagnosis a renal insult early depends on whether or not the alteration in the level of the biomarker detected by current methods precedes the appearance of urinary microalbumin. With respect to being an early biomarker, Table
3 shows that the levels of uNGAL and uCysC were elevated in T2DM and DKD patients with a normal UACR (<30 mg/g•Cr). Thus, the alterations will be detected before the increase in urine microalbumin. Further, we consider that the possibilities of increased biomarkers in T2DM and DKD patients with a normal UACR are as follows: the biomarkers are specific for hyperglycemia without kidney injury; and/or the biomarkers are novel biomarkers for renal impairment and increase earlier than UACR. In the present study, however, the data shown in Table
4 indicates that the levels of biomarkers did not increase with the increase in HbA
1c. Higher uNGAL and uCysC levels may not directly resulted from hyperglycemia, but from chronic tubular impairment, which may be more sensitive in the diagnosis of DKD than microalbuminuria, which is the current biomarker in use. Because biomarkers mirroring tubular impairment can be determined before microalbuminuria, biomarker levels predict the appearance of microalbumin in the urine. In previous studies, the AUC for uNGAL identifying microalbuminuria in T2DM patients was 0.759 with a sensitivity of 66.7% and a specificity of 88.9% [
33]. Vikas Garg et al. [
24] reported that once the values were corrected by urinary creatinine, the AUCs were promoted and achieved 0.956 (urinary NGAL-to-creatinine ratio) and 0.867 (urinary CysC-to-creatinine ratio). Our study showed that the AUCs reached 0.841 (uNGAL) and 0.805 (uCysC), which are high for clinical use and may be sensitive markers for early renal damage in T2DM; however, uCLU may not have the advantage of earlier elevation. The level of uCLU in the normal UACR group was similar to control subjects (Table
3), although the AUC (0.783) was significant for predicting microalbuminuria.
Previous investigations have demonstrated that NGAL is highly expressed in the kidney proximal tubules of diabetic rats [
13,
26] and uNGAL is markedly elevated in microalbuminuric patients with T1DM [
22,
34] or T2DM [
35,
36], uNGAL is positively correlated with proteinuria, and can reflect the severity of diabetic nephropathy [
36,
37]. Urinary CysC has also been suggested as an indicator of tubular dysfunction for the early diagnosis of DKD [
21,
38,
39]. Previous results and conclusions include an early rise of uCysC in T2DM patients [
24], an association with a declining GFR [
21], and a rise in urine ACR [
24], which are consistent with an alteration in the degree of albuminuria [
21]. Compared with the preceding investigations, our study directly highlighted the clinical performance in the diagnosis of DKD and microalbuminuria and elucidated the clinical basis for use of early biomarkers.
As a biomarker for nephrotoxic lesions [
16,
40], urinary CLU has not been investigated to identify the renal damage in clinical T2DM patients. We selected urinary CLU as a potential renal injury biomarker in T2DM patients for the following reasons. First, it has been confirmed that uCLU can reflect proximal tubular damage [
41,
42], an important mechanism underlying diabetic renal insults. Second, CLU functions as a defense factor to prevent renal fibrosis [
41,
43], the common nephropathogenesis of CKD. Third, the findings of animal experiments [
25,
44] have demonstrated that the uCLU level is significantly elevated and closely associated with the severity of histopathologic nephropathy resulting from obesity. The present results showed that the level of uCLU markedly increased in renal-impaired patients, was positively correlated with albuminuria, and mirrored the severity of renal damage in the T2DM cohort. Moreover, uCLU has a suitable AUC with a reasonable sensitivity and specificity for the diagnosis of DKD, macroalbuminuria, and microalbuminuria, even though the elevation may not be earlier than microalbumin. We believe that the potential nephropathogenesis related to CLU is worthy of further study and the TGF-β-related hyperglycemia pathway may be the putative mechanism. High glucose stimulates TGF-β expression and bioactivity in the proximal tubule [
45] via extracellular regulated kinase 1/2 and protein kinase C activation [
30,
46] and promotes inflammatory cells, particularly T cells and macrophages, to release TGF-β, which are early events that set up fibrogenesis [
47]. Highly-expressed TGF-β acts as the most potent cytokine for renal fibrogenesis and stimulus for epithelial-myofibroblast transdifferentiation of proximal tubular epithelial cells [
30,
48]. Further, TGF-β induces the up-regulation of CLU in renal tubular epithelium via the activation of AP-1 transcriptor protein and protein kinase C [
49]. By post-translation processing, CLU is immediately converted to two subtypes. The α subtype is excreted into the urine and the β subtype is accumulated in the cytoplasm of the renal tubular epithelial cells [
50]. As a component of glomerular immune deposits, over-expressed CLU may be anti-apoptotic and pro-survival and attenuate the development of renal fibrosis [
41].
The main limitation to our study was that it involved a relatively small research in a single center, thus needs further validation at the multi-center level as a consecutive cohort study.