Diabetic nephropathy (DN) is a major complication of diabetes. It is estimated that approximately 20–40% of type 2 diabetes (T2D) patients will develop renal disease [1
]. Moreover, DN is the leading cause of end-stage renal disease (ESRD) worldwide [3
]. Analysis of the international statistics collected in the US Renal Data System indicates that Taiwan has the highest incidence and the second highest prevalence of ESRD worldwide [4
]. To date, the most useful approach to identify renal disease in nonproteinuric diabetic patients is microalbuminuria (MA) testing. Patients with MA often progress to macroalbuminuria and ESRD. However, the prognostic value of MA as a biomarker of disease progression is questionable, as patients with MA may exhibit the preservation of renal function [6
]. In addition, some studies indicate that a subset of diabetic patients with normal albuminuria may develop nephropathy [7
], suggesting that patients may have declined renal function before the onset of MA. Moreover, MA is associated with high blood pressure [5
], and it represents a potential risk factor for cardiovascular events in individuals with or without diabetes [8
]. Therefore, the identification of reliable biomarkers for the early prediction of DN and the development of effective treatment to reduce the incidence of ESRD is necessary.
Proteins are functional executors of biological mechanisms, and they have been utilized as prognostic biomarkers for a number of diseases such as DN, cardiovascular disease, and cancer [9
]. With the development of protein separation techniques and mass spectrometry, liquid chromatography tandem mass spectrometry (LC-MS/MS)-based proteomics has been widely applied to the analysis of clinical samples for the elucidation of protein changes in complex mixtures. In the case of nephropathy, proteomic studies have identified several urinary biomarkers, such as alpha 1 antitrypsin [13
], extracellular glutathione peroxidase [14
], and apolipoprotein A-1 [15
], to be associated with renal functional decline.
In the present investigation, we used isobaric tags for relative and absolute quantification (iTRAQ) and label-free proteomics to identify potential urinary protein markers that are associated with nephropathy, which were verified by an enzyme-linked immunosorbent assay (ELISA) in a cross-sectional population study. We additionally performed a nested case-control study to evaluate the ability of one urinary protein, haptoglobin (HPT), on the discrimination of early renal functional decline (ERFD) among T2D patients in Taiwan.
In the present study, we identified urinary HPT as a candidate biomarker that is associated with nephropathy, and verified its discriminatory ability in a population of 208 individuals comprising diabetic and nondiabetic patients. The ability of HCR to predict ERFD was evaluated in 305 T2D patients, with an average follow-up of 4.2 years. Although the HCR and ACR were correlated (r2pearson
= 0.085, p
< 0.001, see Supplementary Figure S2 online
), HCR was identified as an independent predictor of ERFD among T2D patients who had not yet manifested significant kidney disease, as shown in data in logistic models. The HCR had higher sensitivity (28.8%), specificity (96.5%), and positive predictive value (68.0%) than the ACR (18.6%, 97.4%, and 64.7%, respectively) to predict ERFD in our cohort population. These data indicate that the HCR may be a more sensitive biomarker than the ACR, and it may be used as a confirmatory test for the early detection of decline in renal function. Moreover, when we limited our population to T2D patients with normal albumin levels in urine, patients with higher urinary HCR had a higher risk of developing ERFD, indicating that the HCR may be a valuable clinical tool for predicting nephropathy before the appearance of urinary albumin (when limited to T2D patients with ACR < 30 mg/g (n
= 226), the O.R. for developing ERFD was 4.15 (1.88, 9.16) when comparing individuals with higher HCR with those with lower HCR).
HPT has been identified as a biomarker for several clinical outcomes, including acute allograft rejection [11
], DN [21
], chronic renal insufficiency [23
], and glomerulonephritis. HPT is a heavily glycosylated alpha-2 sialoglycoprotein with a net negative charge [24
]. This protein was shown to bind hemoglobin and to prevent the loss of iron through the kidneys and oxidative damage to nephrons [25
], as the HPT–hemoglobin complex cannot be filtered at the glomerulus, owing to its large size and negative charge. Increased concentrations of HPT may be a response to renal tubular injury or oxidative stress. Moreover, HPT, similar to albumin, is a marker of increased glomerular permeability, as serum HPT can leak into urine when the glomerular permeability reaches a certain level [27
]. Recently, HPT has been identified as a biomarker of kidney damage in the Chinese population. Studies have shown that urinary HPT can predict ERFD among T2D patients with chronic kidney disease (CKD) of stages 1 and 2 [27
], or even in the CKD of stage 3 [28
]. In a 5.3-year cohort study, Yang et al. [23
] also identified HPT as a novel biomarker that complements urine albumin in predicting chronic renal insufficiency (< 60 mL/min per 1.73 m2
) in T2D patients with eGFR > 80 mL/min per 1.73 m2
. In the present study, we validated the ability of HPT to predict ERFD among T2D patients with CKD stages 1 and 2 among the Taiwanese population.
Our results showed that AMBP had the highest AUC value in the ELISA verification phase, including four cross-sectional groups. AMBP could pass through the glomerulus and subsequently be re-absorbed by the tubular tissues. AMBP is a stable urinary protein that has been used as a clinical parameter for the detection and differentiation of proteinuria. High-urine AMBP levels indicate proximal tubular dysfunction, and they are associated with interstitial fibrosis, tubular atrophy, and inflammation [11
]. A previous study using ProteinChip (H50 chip) analysis has shown that urinary AMBP is increased in DN [22
]. More, the increase of AMBP level in urinary exosomes from individuals with advanced stages of DN (CKD stages III–V, with high urinary albumin), compared with healthy subjects, was revealed by label-free comparative analysis [29
]. These results suggested that AMBP in total urine or urinary exosomes could be a potential biomarker for DN. However, this marker was not sufficient for the early prediction of ERFD, based on our preliminary results from a nested case-control study (see Supplementary Table S4 online
). The possible reasons for the lack of significant association between AMBP levels and ERFD may be the small differences of AMBP level between groups, and the small sample size. The predictive potential of AMBP should be determined in a larger patient population. Alternatively, AMBP may not increase at early stages of ERFD.
Our investigation had several strengths. To the best of our knowledge, the present work describes the largest cohort study, in which samples were collected nation-wide by the Taiwan Biobank. The HCR was validated as a biomarker that outperformed the ACR in predicting renal decline progression among T2D patients in Taiwan. Moreover, the result from our verification phase (cross-sectional study) indicates that the HCR may serve as a biomarker of nephropathy not only in diabetic patients but also in nondiabetic individuals. Finally, the HCR was found to be an independent predictor of ERFD in T2D patients, even before the appearance of urinary albumin.
However, there were also some limitations in this study. First, the study population was limited to patients with an ACR of less than 300 mg/g, and none of the patients had stage 3 CKD (eGFR level between 30 and 59 mL/min per 1.73 m2
); therefore, we did not test whether HPT could predict ERFD among T2D patients with stage 3 CKD, which has been demonstrated by Liu et al. [28
]. In addition, we did not collect information on the use of insulin or other drugs (such as renin-angiotensin system antagonists) that may alter renal function. However, according to our data, eGFR and SBP/diastolic blood pressure (DBP) values were similar between individuals with lower and higher HCRs at baseline. Third, the HPT biomarker was validated in a Taiwanese population only. Further validation steps should be performed in more diverse populations. Moreover, previous studies have shown that several protein and metabolic urinary markers such as CD59 and alpha‑1 antitrypsin were associated with albuminuria and with the capacity to predict the progression to high albuminuria [30
]. Therefore, this may be another direction for investigating the potential urinary protein markers in a cohort in which subjects are with normoalbuminuric at baseline, and will develop albuminuria in the future.