Identification of Novel Urinary Biomarkers for Predicting Renal Prognosis in Patients With Type 2 Diabetes by Glycan Profiling in a Multicenter Prospective Cohort Study: U-CARE Study 1

Several biomarkers have been demonstrated to predict renal prognosis in patients with diabetes. Blood levels of tumor necrosis factor receptors 1 and 2 are well-known prognostic indicators for diabetic kidney disease (DKD) (1,2). Similarly, markers of tubulointerstitial injury, inflammation, and filtration have been reported to predict the progression of DKD (3–6). These biomarkers partly allow us to predict renal prognosis at an early stage of DKD independently of established clinical factors, such as the estimated glomerular filtration rate (eGFR) and albuminuria. However, despite early detection of high-risk patients and recent new protective treatments for DKD (7–9), the management of patients with DKD and rapid deterioration of renal function remains difficult. Thus, new biomarkers are needed to identify the pathogenesis of DKD and new therapeutic targets.

The role of glycans and their enzymatic modification (glycosylation) lately have attracted much attention in relation to research on cancer and metabolic diseases, including diabetes and DKD. Ohtsubo et al. (10) reported that human pancreatic β-cell–specific GnT-4a glycosyltransferase, which generates the core β1–4 GlcNAc linkage in N-glycans, has a protective effect against diabetes by increasing GLUT2 expression and maintaining insulin secretion. Differences of glycosylation, including O-GlcNAcylation, were reported to contribute to the progression of DKD in rats (11,12). There have been few reports about differences of glycosylation in patients with DKD because of technical obstacles to glycan analysis as a result of the complicated structures of these molecules and the time-consuming processes required for mass spectrometry (MS). However, the evanescent-field fluorescence-assisted lectin microarray that we reported previously enables high-throughput quantification of glycan binding to 45 specific lectins (13,14). In the preliminary analysis of our previous study (14), the urinary glycosylation pattern varied considerably, whereas the changes in serum glycosylation were barely detectable among patients with various kidney diseases, including DKD, associated with almost equivalent proteinuria and eGFR. Moreover, studies of the serum N-glycan profile have revealed that differences of N-glycosylation are associated with diabetic complications and glycemic control in patients with type 1 and 2 diabetes (15,16), but the association between the urinary glycosylation profile and renal prognosis has not been investigated in patients with diabetes to the best of our knowledge.

Accordingly, we investigated the relationship between urinary excretion of O- and N-glycans binding to 45 lectins and the renal prognosis in patients with type 2 diabetes. In addition, we assessed the incremental predictive value of adding promising glycans to a model that contained established clinical variables, including albuminuria and eGFR.

This prospective cohort study was initiated in 2012. Among 688 patients with type 2 diabetes admitted to eight hospitals in Japan from June 2012 to March 2013, 675 patients were eligible for enrollment. Exclusion criteria were a diagnosis of slowly progressive type 1 diabetes during follow-up or a baseline eGFR <15 mL/min/1.73 m² (Supplementary Fig. 1). The diagnosis of diabetes was based on Japanese Diabetes Society criteria (17). In addition, 134 volunteers who underwent medical checkups at Okayama Health Foundation in March 2016 and were confirmed to have neither diabetes nor chronic kidney disease (CKD) (17,18) were enrolled as control subjects to compare differences in glycosylation. The protocol for this study was approved by the ethics committee of Okayama University Hospital in June 2012 (identification number: H24-003). This study was registered with the University Hospital Medical Information Network in June 2012. Written informed consent was obtained from all patients with diabetes, whereas comprehensive anonymous consent was obtained from control subjects.

Urine samples collected in the early morning and stored at baseline (patients in 2012–2013, control subjects in 2016) were used to measure urinary glycans and urinary albumin levels in 2015–2016. All specimens were aliquoted, stored at −80°C until measurement, and thawed for the first time to perform this study. The average storage period until measurement was 2.1 ± 0.1 years for patients and 0.9 ± 0.0 years for control subjects. As shown in Supplementary Table 1, the influence of the difference in frozen storage duration on measurement of glycosylation is negligible. Urinary glycans were measured by the evanescent-field fluorescence-assisted lectin microarray, which is a new method of glycan profiling (13,19–21). In brief, we measured urinary levels of Cy3-labeled glycoproteins that bound to 45 lectins with different specificities, and urinary glycan intensity can be measured in 300 samples in 3 working days (Supplementary Fig. 2). Cy3 binds to primary amine in principle. Urinary albumin was labeled by Cy3, although it lacks glycan modification, resulting in minimal background reactivity. Urinary creatinine also was labeled by Cy3, causing background reactivity like albumin. In preliminary experiments, both urinary albumin and creatinine were significantly associated with the intensity of background reactivity (r between log[urinary albumin concentration] and log[background intensity (BG-I)] = 0.697, r between log[urinary creatinine concentration (UCr)] and log[BG-I] = 0.166) (Supplementary Fig. 3). We also observed that the intensity of glycan reactivity (glycan intensity) in 24-h urine and spot urine samples was more closely correlated with the net glycan intensity (Net-I = raw glycan intensity [Raw-I] – BG-I) than with the Net-I/UCr ratio or the Raw-I/UCr ratio (spot urine/24-h urine Net-I ratio 1.20 ± 0.74 [mean ± SD], Net-I/UCr ratio 1.77 ± 2.69, Raw-I/UCr ratio 1.68 ± 2.00) (Supplementary Fig. 4). Furthermore, comparison of glycan intensity between original urine samples and 10-fold diluted urine samples showed that all the Raw-I ratios (ratio = original urine intensity/10-fold diluted urine intensity) and most of the Net-I ratios were <10, suggesting that the urinary glycan index did not increase linearly, but quadratically (Supplementary Table 2). On the basis of these results, we performed all analyses by using the glycan index appropriately transformed from the Net-I according to its distribution.

GFR was estimated by using the Japanese coefficient–modified Chronic Kidney Disease Epidemiology Collaboration equation (22). The baseline urinary albumin/creatinine ratio (UACR) in milligrams per gram creatinine was measured in a spot urine specimen, and normoalbuminuria, microalbuminuria, and macroalbuminuria were defined as UACR <30, ≥30 and <300, and ≥300 mg/gCr, respectively (15). HbA_1c data are presented as National Glycohemoglobin Standardization Program values according to the recommendations of the Japanese Diabetes Society and the International Federation of Clinical Chemistry (23). BMI was calculated as weight in kilograms divided by the square of height in meters. Mean arterial pressure (MAP) was calculated as two-thirds of diastolic pressure plus one-third of systolic pressure (mmHg). Hypertension was defined as a baseline blood pressure (BP) ≥140/90 mmHg or use of antihypertensive drugs. The average annual values of clinical parameters, including systolic BP (SBP), diastolic BP (DBP), MAP, and HbA_1c plus the use of an ACE inhibitor (ACE-I) or angiotensin receptor blocker (ARB) during follow-up were compared between all patients with and without outcome and those stratified according to baseline eGFR categories. The grade of diabetic retinopathy was determined by an ophthalmologist at baseline (24). In this study, cardiovascular disease (CVD), stroke, and peripheral arterial disease (PAD) were defined as events requiring admission for treatment, cerebral bleeding or infarction requiring admission for treatment, and PAD requiring admission for intervention or surgery, respectively. Cardiovascular events were defined as any CVD, stroke, or PAD event.

The primary study end point was defined as a decrease of eGFR by at least 30% from baseline or commencement of dialysis for end-stage renal disease (ESRD). None of the patients received a kidney transplant during follow-up.

Data are summarized as percentages or as the mean ± SD, as appropriate. All skewed variables were subjected to log-transformation to improve normality before analysis. Correlations among glycan indexes were evaluated by Pearson correlation analysis. Univariate and multivariate linear regression analyses were used to explore the association of urinary glycan index with baseline HbA_1c and age. In both regression models, the dependent variable was glycan index. In each multivariate model, the β-coefficient for HbA_1c was adjusted for baseline age, sex, and duration of diabetes, whereas the β-coefficient for age was adjusted for sex and duration of diabetes. The cumulative incidence rate of the primary outcome was estimated by Kaplan-Meier method for urinary glycan quartiles in all patients, and incidence rates were compared with the log-rank test. In addition, we compared Kaplan-Meier curves between groups of patients with normoalbuminuria or microalbuminuria stratified according to urinary glycan quartiles (Q1–3 vs. Q4 or Q1 vs. Q2–4) and compared groups of patients with macroalbuminuria and higher or lower glycan indexes than the median value. The Cox proportional hazards model was used to calculate the hazard ratio (HR) and 95% CI for the death-censored end point. In the multivariate model, HRs were adjusted for age, sex, MAP, HbA_1c, eGFR, and log-transformed UACR at baseline. These covariates were selected as potential confounders on the basis of biological plausibility and metabolic memory (25,26). We also examined another multivariate model without baseline HbA_1c to evaluate the effect of baseline HbA_1c levels on the outcome. Improvement in discriminating the risk of the study outcome at the median follow-up time (4.0 years) was assessed by analysis of category-free net reclassification improvement (NRI) and absolute/relative integrated discrimination improvement (IDI) as reported elsewhere (27–29). The median follow-up time was selected as the cutoff point for analysis because it was previously used in similar biomarker studies (3,30). We also used the Akaike information criterion (AIC) to compare the fit of various models. Moreover, we compared Harrell’s concordance index (C-index) between multivariate Cox proportional hazards models with or without glycan biomarkers. The 95% CIs for category-free NRI and IDI and the differences of the C-index were computed from 5,000 bootstrap samples to adjust for optimism bias. Two-tailed P < 0.05 was considered to indicate statistical significance. Analyses and creation of graphs were performed with Stata/SE version 14.0 (StataCorp) and Origin version 2017 (OriginLab) software.

The median follow-up period was 4.0 years (interquartile range [IQR] 3.9–4.0 years). During follow-up, the primary end point occurred in 63 (9%) patients, and 12 (2%) patients died as a result of causes other than ESRD after refusing dialysis or transplantation.

The baseline characteristics of the subjects are shown in Table 1. Their mean ± SD age was 63 ± 11 years, 61% were men, and the median known duration of diabetes was 11.0 years (IQR 6.2–17.7 years). The baseline SBP, DBP, and MAP was 131.0 ± 17.1, 74.8 ± 10.9, and 93.5 ± 11.7 mmHg, respectively. One-third of the patients had diabetic retinopathy (any type), and their mean baseline HbA_1c was 7.1 ± 1.1% (54.3 ± 12.0 mmol/mol). In addition, the mean baseline eGFR was 71.4 ± 17.1 mL/min/1.73 m², and median UACR was 17.3 mg/gCr (IQR 7.8–71.1 mg/gCr); 594 patients had normoalbuminuria (64%) or microalbuminuria (24%). With respect to BP and glycemic control during follow-up, the average SBP, DBP, MAP, and HbA_1c were not significantly different between patients with and without outcome. As well, the use of an ACE-I or ARB during follow-up was not significantly different between the patients with and without outcome in groups stratified according to baseline eGFR categories (Supplementary Table 3).

Univariate and multivariate regression analysis in all patients and patients with baseline eGFR ≥60 mL/min/1.73 m² revealed that lower HbA_1c was significantly associated with glycan complex structures that have higher GlcNAc [recognized by PHA(L), PHA(E), DSA, LEL, STL, and WGA], higher Siaα3Galβ4GlcNAc (recognized by MAL_I and ACG), higher GalNAcβ4GlcNAc(recognized by WFA), and higher GalNAcα3GalNAc (recognized by DBA), whereas higher HbA_1c was significantly associated with higher Manα3Man (recognized by GNA). Most of these significant associations were attenuated in the patients with baseline eGFR <60 mL/min/1.73 m² (Supplementary Table 4). Similar regression analysis for age showed that older age was associated with higher Fucα2GalβGlcNAc (recognized by UEA_I), higher (Galβ4GlcNAc)n (polylactosamine, recognized by LEL), higher (GlcNAcβ4MurNAc)n (peptidoglycan backbone, recognized by STL), and higher Siaα3Galβ3(Siaα6)GalNAc (recognized by MAH) (Supplementary Table 5).

Unadjusted and adjusted HRs for glycan binding to the panel of 45 lectins with different specificities and the reported structure of the glycan binding to each lectin are shown in Fig. 1. A number of urinary glycans were significantly correlated with the renal outcome in the univariate Cox regression model, whereas the urinary glycans binding to six specific lectins (Sambucus nigra [SNA], Ricinus communis [RCA120], Dolichos biflorus [DBA], Agaricus bisporus [ABA], Artocarpus integrifolia [Jacalin], and Amaranthus caudatus [ACA]) were significantly correlated with renal outcome in both the univariate and multivariate models. In the multivariate model adjusted for known indicators of DKD progression, including baseline eGFR and UACR, the HR for positive glycan binding to SNA (+1 SD for the glycan index) was 1.42 (95% CI 1.14–1.76), whereas the HR for glycan binding to RCA120 was 1.28 (1.01–1.64); DBA, 0.80 !(0.64–0.997); ABA, 1.29 (1.02–1.64); Jacalin, 1.30 (1.02–1.67); and ACA, 1.32 (1.04–1.67) (Fig. 1A). These associations remained largely unchanged when average MAP and/or HbA_1c during follow-up were incorporated into the multivariate model and when baseline HbA_1c was eliminated from the multivariate model (Supplementary Tables 6 and 7). As shown in Fig. 1B, the glycans Siaα2-6Gal/GalNAc, Galβ4GlcNAc, and GalNAcα3GalNAc were reported to bind with SNA, RCA120, and DBA, respectively, whereas Galβ3GalNAc was reported to bind with ABA, Jacalin, and ACA.

We stratified the patients with diabetes into four CKD heat map groups and 11 categories by using the baseline UACR and eGFR (31). Comparisons of glycan binding to SNA, RCA120, DBA, ABA, Jacalin, and ACA among the control group and patients with diabetes stratified according to the CKD heat map groups and 11 categories are shown in Supplementary Fig. 5. Overall, the glycan index values were higher in the severe categories of the CKD heat map, albuminuria, and eGFR, except for the glycan index of DBA. Of note, the glycan index of SNA was significantly higher in the green heat map group than in the control group, even though both groups were defined by normoalbuminuria and eGFR >60 mL/min/1.73 m². Correlations among positive binding to ABA, Jacalin, and ACA were extremely strong (r = 0.952 between ABA and Jacalin, r = 0.931 between ABA and ACA, and r = 0.942 between ACA and Jacalin) as was the correlation between SNA and RCA120 (r = 0.921) (Supplementary Table 8).

Kaplan-Meier curves stratified according to quartiles for baseline urinary glycan binding to SNA, RCA120, DBA, ABA, Jacalin, and ACA are shown in Fig. 2. The cumulative incidence rate of the renal outcome was significantly higher in the highest quartile for urinary glycan binding to SNA, RCA120, ABA, Jacalin, or ACA than in the other quartiles, whereas it was significantly higher in the lowest quartile for glycan binding to DBA than in the other quartiles (Fig. 2A). Similar results for glycans binding to SNA, ABA, Jacalin, and ACA were obtained in patients with normoalbuminuria/microalbuminuria (Fig. 2B). Among patients with macroalbuminuria, the cumulative incidence rate was significantly higher in those with higher glycan index values than in those with lower glycan index values, except in the case of the glycan index for DBA (Fig. 2C).

The category-free NRI, absolute/relative IDI, and AIC for predicting the primary outcome at the median follow-up time (4.0 years) obtained by adding the glycan indexes as well as the difference of the C-index between Cox regression models with or without the urinary glycan indexes are summarized in Table 2. The addition of any single glycan index to the multivariate model did not improve prediction, whereas adding all six glycan indexes significantly improved risk classification, integrated discrimination, and AIC (category-free NRI 0.51 [95% CI 0.22–0.80], relative IDI 0.18 [0.01–0.35], AIC decrease from 296 to 287). Similarly, when four glycan indexes (on the basis of the reported glycan specificities) were added, risk classification, integrated discrimination, and AIC also were significantly improved (Table 2). However, the C-index did not increase significantly when these glycan indexes were added to the multivariate Cox regression model (Table 2).

Glycans are involved in various biological processes, including development, immunity, infection, hormone actions, cell adhesion, and oncogenesis (19,32,33). One of the most prominent features of glycans is structural heterogeneity, and this distinguishes glycans from other major biopolymers, such as nucleic acids and proteins. Such heterogeneity is largely attributable to the glycan fabrication process and depends on structural modification by glycosyltransferases, which is known as glycosylation (19).

Podocalyxin is the molecule in the kidneys for which the functional effects of changes in glycosylation have been most thoroughly investigated (34). It is a glycoprotein expressed by podocytes that plays an essential role in the maintenance of podocyte slit pore integrity and effective glomerular filtration (35). Podocalyxin is decorated by O-glycans, including an abundance of sialic acid, and removal of sialic acid leads to effacement of the podocyte foot processes and proteinuria in mice secondary to loss of the negative charge on the glomerular basement membrane (36). Similarly, doxycycline-induced global deficiency of glycosyltransferase core 1 synthase and glycoprotein-N-acetylgalactosamine 3-β-galactosyltransferase 1 (C1galt1) leads to marked reduction of SNA-binding Siaα2-6Gal/GalNAc on podocalyxin in mice followed by albuminuria, rapid progression of glomerulosclerosis, effacement of podocyte foot processes, and thickening of the glomerular basement membrane (37). These results might suggest that expression of mature O-glycans modified by a key glycosyltransferase in the kidneys, especially on podocytes, is crucial to the maintenance of the normal renal architecture. If abnormalities of renal O-glycosylation occur, parts of mature O-glycans could undergo reduction and be excreted in the urine. Although C1galt1 is not involved in the modification of Siaα2-6Gal/GalNAc (Supplementary Fig. 6A), deletion of C1galt1 induces a decrease of Siaα2-6Gal/GalNAc and an increase of Tn antigen in mouse podocalyxin (37), suggesting that impairment of one O-glycosyltransferase also affects other O-glycosylation pathways and leads to the creation of abnormal O-glycans. The current study findings might support these suggestions as follows.

We found that urinary levels of glycans binding to SNA, ABA, Jacalin, and ACA were significantly higher in the severe CKD heat map group and that the cumulative incidence rate of the renal outcome was significantly higher in the highest glycan quartiles than in the other glycan quartiles not only in all patients but also in patients with normoalbuminuria/microalbuminuria (Supplementary Fig. 5 and Fig. 2A and B). Of note, urinary excretion of Siaα2-6Gal/GalNAc binding to SNA was significantly higher in non-CKD patients with diabetes than in non-CKD control subjects without diabetes, suggesting that a change in the glycosylation of Siaα2-6Gal/GalNAc could occur in the early stage of DKD before the onset of albuminuria or detectable deterioration of renal function. On the basis of these results and the above-mentioned hypothesis that changes of glycosylation may lead to renal structural damage, it might be reasonable that indexes for these glycans may be significantly associated with the risk of the renal outcome independently of baseline albuminuria and eGFR (Fig. 1).

Galβ4GlcNAc binds to RCA120 and is involved in N- and O-glycosylation. During O-glycosylation, Galβ4GlcNAc plays a role in the extension of cores 2 and 4, which are finally modified by sialic acid (32) (Supplementary Fig. 6A). Therefore, if impairment of the enzyme adding sialic acid to cores 2 and 4 exists, urinary excretion of Galβ4GlcNAc could increase by the same mechanism as that which increases urinary levels of Siaα2-6Gal/GalNAc and Galβ3GalNAc. In this study, the glycan index for RCA120 was strongly correlated with that for SNA (r = 0.921) (Supplementary Table 8), which suggests that the common underlying mechanism is impairment of sialylation. In addition, Ravidà et al. (12) demonstrated that glycans binding to RCA120 gradually show a significant decrease in the renal cortex of diabetic rats with progression of DKD, whereas these glycans increase in nondiabetic rats. Therefore, increased urinary levels of Galβ4GlcNAc might reflect abnormal O-glycosylation in kidney tissues.

With regard to GalNAcα3GalNAc, we could not reasonably explain why urinary excretion of this glycan binding to DBA was negatively correlated with the renal outcome. DBA is well known for recognizing the epitope of glycans on the surface of type A red blood cells (33). It has also been reported to bind to glomerular lesions in human DKD and to the distal tubules in human kidney tissue, but further investigations have not been performed (38,39). DBA recognizes core 5 (GalNAcα3GalNAc), which is a rare component of O-glycans, and the relevant glycosyltransferase has not yet been reported (32). Therefore, additional investigation of the role of this glycan in the kidneys and DKD is needed.

Taken together, changes of glycosylation in DKD seem to be strongly associated with renal prognosis because abnormal glycosylation might be involved in the progression of DKD. Supplementary Fig. 6 shows a scheme of our hypothesis regarding the mechanism by which abnormalities of glycosylation may be associated with progression of DKD. We also speculate that the urinary glycosylation difference well reflects the local glycosylation changes in kidney tissues rather than alterations of systemic glycosylation on the basis of the discrepancies of glycan profiles between urine and serum/plasma samples (15,16,40). Serum protein N-glycan profiling has been shown to lower levels of fucosylated biantennary glycans, and higher levels of complexes that have more GlcNAc and heavily galactosylated and sialylated glycans have been associated with higher HbA_1c levels in patients with diabetes (15,16), whereas those associations were not observed in the current urinary glycan profiling. Similarly, plasma IgG N-glycan profiling revealed that higher levels of bisecting GlcNAc and lower levels of sialylation were correlated with older age, especially >60 years (40), whereas these results were not seen in the current study with urine samples. In the preliminary data of our previous study (14), we found completely different glycosylation patterns between serum and urine samples as well as different urinary glycosylation among patients with biopsy-proven kidney diseases, including DKD, and similar degrees of proteinuria and eGFR, which might support our speculation. These hypotheses are based on in vivo and clinical studies of human DKD, so additional investigation of glycosylation changes in human kidney tissue is required.

Potential limitations of this study were that the current lectin microarray system does not allow complete determination of glycan structures as can MS, and unknown preferred glycan structure to each lectin might have biased the results. However, this technique is useful for differentiating urinary glycan profiling in individuals, and it enabled us to measure a wide range of urinary glycan intensity in high throughput without a liberation process, although conventional methods, including MS, require both prior liberation and subsequent labeling (19). Another limitation was that this study included patients with various stages of CKD, and renal biopsy was not performed in all subjects. Therefore, we cannot exclude the possibility that the decline of eGFR in some patients resulted from kidney diseases other than DKD. However, in previous well-known cohort studies of type 2 diabetes, DKD was not always confirmed by renal biopsy (1,3). Moreover, in the patients with diabetes, renal biopsy is considered when hematuria, granular casts, sudden onset of nephrotic syndrome, and rapidly progressive glomerulonephritis are indicated, whereas such conditions were not observed during follow-up in the current study. We hope that an ongoing research biopsy study of DKD at our institution will solve this issue in the future.

In conclusion, we demonstrate that urinary excretion of glycans binding to several lectins, including SNA (recognizing Siaα2-6Gal/GalNAc), RCA120 (Galβ4GlcNAc), DBA (GalNAcα3GalNAc), ABA, Jacalin, and ACA (Galβ3GalNAc), are significantly associated with renal outcome in patients with type 2 diabetes. The addition of the combined glycan index to a model with standard risk factors significantly improves prediction of renal outcome, suggesting that these urinary glycans may be novel predictors of renal prognosis in patients with type 2 diabetes. The findings could provide new insights into changes of glycosylation related to DKD. The mechanisms that underlie differences of urinary glycan excretion and changes of glycosylation in DKD should be investigated further.

Acknowledgments. The authors thank Dr. Kazuyuki Hida (National Hospital Organization Okayama Medical Center), Dr. Tatsuaki Nakato (Okayama Saiseikai General Hospital), Dr. Takashi Matsuoka (Kurashiki Central Hospital), Dr. Ikki Shimizu (The Sakakibara Heart Institute of Okayama), Dr. Tomokazu Nunoue (Tsuyama Chuo Hospital), Dr. Katsuhiro Miyashita (Japanese Red Cross Okayama Hospital), Dr. Shinichiro Ando (Okayama City General Medical Center), and Dr. Akiho Seki (Okayama Health Foundation) for providing data and for patient treatments. The authors also thank Drs. Ryo Kodera and Satoshi Miyamoto (Okayama University Hospital), Drs. Akihiro Katayama, Chigusa Higuchi, Hidemi Takeuchi, Yusuke Shibata, Ichiro Nojima, Yuzuki Kano, Yuriko Yamamura, Nozomu Otaka, Chieko Kawakita, Keigo Hayashi, and Yasuhiro Onishi (Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama) for collecting data. The authors greatly appreciate Dr. Hirofumi Makino (Okayama University) for critical suggestions in the design of the current investigation.

Funding. This work was partly supported by the Ministry of Health, Labour and Welfare in Japan (Health Labor Sciences Research Grant 201413003), the Japan Agency for Medical Research and Development (AMED) (grant 17ek0210095h0001), Novo Nordisk Pharma Ltd. (Junior Scientist Development Grant [2016–2017]) from the Okinaka Memorial Institute for Medical Research (a grant in 2017), and The Yukiko Ishibashi Foundation (a grant in 2016).

Duality of Interest. M.Ya. was a former employee of GP BioSciences Co., Ltd., and is currently an employee of GlycoTechnica, Ltd. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. K.M. contributed to designing the research, analyzing and interpreting data, measuring urinary glycan levels, collecting and summarizing clinical data, and writing the manuscript. M.I. contributed to collecting, summarizing, and assessing clinical data. S.Y. contributed to measuring urinary glycan levels and collecting and summarizing clinical data. S.T., A.T., H.A.U., J.E., A.N., and D.O., contributed to managing patients and assessing data. M.Yo. contributed to interpreting data and performing statistical analyses. M.Ya. contributed to measuring urinary glycan levels; interpreting data, especially urinary glycan data; and writing the manuscript. K.S. contributed to managing patients and assessing data. J.W. was responsible for the study design, supervised data collection and data analysis, and contributed to drafting and editing the manuscript. J.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Part of this work was presented at the American Society of Nephrology Scientific Oral Sessions, New Orleans, LA, 31 October–5 November 2017, and the International Society of Nephrology Frontiers Meetings, Tokyo, Japan, 22–25 February 2018.