Article Carbamylated Lipoproteins and Progression of Diabetic Kidney

CJASN ePress. Published on February 19, 2020 as doi: 10.2215/CJN.11710919 Article

Carbamylated Lipoproteins and Progression of Diabetic Kidney Disease

Kathryn C.B. Tan ,1 Ching-Lung Cheung,2 Alan C.H. Lee ,1 Joanne K.Y. Lam ,1 Ying Wong,1 and Sammy W.M. Shiu1

Abstract Background and objectives Protein carbamylation is a consequence of uremia and carbamylated lipoproteins Departments of contribute to atherogenesis in CKD. Proteins can also be carbamylated by a urea-independent mechanism, and 1Medicine and whether carbamylated lipoproteins contribute to the progression of CKD has not been investigated. 2Pharmacology and Pharmacy, University Design, setting, participants, & measurements A case-control study was performed to determine whether there of Hong Kong, Hong were changesinplasma levels of carbamylatedlipoproteinsin individualswithtype 2diabetes witheGFR .60 ml/ Kong, China min per 1.73 m2 compared with a group of age- and sex-matched healthy controls. A cohort of 1320 patients with $ 2 Correspondence: Prof. type 2 diabetes with baseline eGFR 30 ml/min per 1.73 m was longitudinally followed up to evaluate the Kathryn C.B. Tan, association between carbamylated lipoproteins and progression of CKD. The primary kidney outcome was Department of defined as doubling of serum creatinine and/or initiation of KRT during follow-up. Plasma carbamylated LDLs Medicine, University and HDLs was measured by ELISA. of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong . 2 ResultsIn individuals with diabetes with eGFR 60 ml/min per 1.73 m , both plasma carbamylated LDL and HDL Kong, China. Email: levels were higher compared with healthy controls (P,0.001). After a mean follow-up of 9 years of the diabetic [email protected] cohort, individuals in the top quartile of carbamylated LDL (hazard ratio, 2.21; 95% confidence interval, 1.42 to 3.46; P,0.001) and carbamylated HDL (hazard ratio, 4.53; 95% confidence interval, 2.87 to 7.13; P,0.001) had higher risk of deterioration of kidney function compared with those in the lowest quartile. On multivariable Cox regression analysis, plasma carbamylated LDL was no longer associated with kidney outcome after adjusting for baseline eGFR and potential confounding factors. However, the association between plasma carbamylated HDL and kidney outcome remained significant and was independent of HDL cholesterol.

Conclusions Plasma carbamylated HDL but not carbamylated LDL was independently associated with progression of CKD in patients with type 2 diabetes. CJASN 15: ccc–ccc, 2020. doi: https://doi.org/10.2215/CJN.11710919

Introduction during the metabolism of thiocyanate by myeloper- Carbamylation is a post-translational modification oxidase. In the presence of hydrogen peroxide, of proteins owing to the spontaneous nonenzymatic myeloperoxidase catalyzes thiocyanate oxidation reaction between isocyanic acid and specificfree to form cyanate, which is then rapidly converted to functional groups of proteins. During the process, its reactive form, isocyanic acid. Myeloperoxidase is “ ” 2 the carbamoyl moiety ( CONH2) is added to the released from activated leukocytes during inflam- amino terminus residues of protein like lysine, leading mation, and Wang et al. (4) showed that myeloper- to the formation of «-carbamyl-lysine (homocitrulline). oxidase played a significant role in protein carbamylation This results in alterations in the structural and func- at sites of inflammation and atherosclerotic pla- tional properties of proteins. Carbamylation of proteins ques. It has been suggested that carbamylated can cause partial or complete loss of function and/or proteins are involved in the progression of various trigger specific inappropriate and potentially pathologic chronic diseases, in particular kidney and cardio- cellular responses (1). Because isocyanic acid is pre- vascular disorders (5). dominantly formed by the decomposition of urea, Lipoproteins are subjected to carbamylation in the proteins are subjected to increased carbamylation in circulation and, in patients with ESKD, plasma levels patients with CKD and plasma levels of isocyanic acid of carbamylated LDLs and carbamylated HDLs are can be increased up to three-fold in patients with elevated and may cause detrimental cardiovascular uremia (2,3). Wang et al. (4) have recently shown that effects (6,7). Carbamylation of lipoproteins is a major proteins can also be carbamylated by a urea-independent contributory factor of atherosclerosis in patients with mechanism. Decomposition of urea is not the only source CKD, and carbamylated LDL is a nontraditional risk of isocyanic acid formation and cyanate can be formed factor for cardiovascular events in these patients (8). www.cjasn.org Vol 15 March, 2020 Copyright © 2020 by the American Society of Nephrology 1 2 CJASN

Although carbamylation of lipoproteins is a consequence Primary kidney outcome was defined as a doubling of of kidney disease, whether carbamylated lipoproteins serum creatinine from baseline and/or initiation of KRT. may also contribute to the pathogenesis and progression Rapid kidney progression was a secondary outcome and of kidney disease has not been investigated. It has been was defined as a rate of eGFR decline .5 ml/min per year. suggested that changes in lipid metabolism may play a Length of follow-up was calculated as the time from role in the development of kidney disease (9). We have baseline examination to the date of doubling of creatinine, previously shown that in patients with type 2 diabetes date of dialysis or death, or last follow-up as per the without kidney impairment, carbamylation of LDL was censoring date of January 31, 2017, whichever was earliest. increased and carbamylation was mainly driven by Serum creatinine was measured annually during follow-up elevated myeloperoxidase activity (10). However, and more frequently as clinically indicated in those with whether plasma carbamylated HDL is higher in patients eGFR ,60 ml/min per 1.73 m2. with type 2 diabetes without kidney impairment is Plasma levels of lipids, carbamylated LDL and carba- unclear because patients with type 2 diabetes tend to mylated HDL, glucose, hemoglobin A1c (HbA1c), and have low plasma HDL and apoA concentrations. In this creatinine were measured in fasting blood samples taken study, we aim to investigate whether carbamylated at baseline. Serum creatinine was measured by the Jaffe lipoproteins are elevated in patients with type 2 diabetes method and eGFR was calculated by Modification of Diet with and without kidney impairment and to determine in Renal Disease Study equation. Albuminuria status was whether carbamylated lipoproteins are associated with determined by urine albumin-to-creatinine ratio from at the progression of CKD. least two random urine samples collected on two separate occasions within 6 months. Plasma total cholesterol and triglyceride was determined enzymatically, and HDL Materials and Methods cholesterol was measured using a homogenous method To evaluate the relationship between plasma carbamy- with polyethylene glycol–modified enzymes and lated lipoproteins and kidney function in patients with type a-cyclodextrin. LDL cholesterol was calculated by the 2 diabetes, plasma carbamylated LDL and carbamylated Friedewald equation or measured directly if plasma tri- HDL was measured in baseline samples from a cohort of glyceride was .399 mg/dl. individuals with type 2 diabetes who were being pro- Plasma carbamylated LDL concentration was measured spectively followed up to study the pathogenesis and using an in-house sandwich ELISA with polyclonal rabbit progression of complications in type 2 diabetes in Chinese anti-human carbamylated LDL antibody as the capture patients. Details of the cohort has previously been reported antibody, as previously described (10). Each sample was (11). In brief, the cohort was recruited from 1996 to 2014 assayed in triplicates and results were expressed as micro- and individuals with type 2 diabetes were invited to gram of carbamylated LDL per milliliter of plasma. The participate on their first clinic visit and major exclusion intra- and interassay coefficients of variation were 5.2% and criteria were non-Chinese descent, type 1 diabetes, 9.3%, respectively. To measure plasma level of carbamy- malignancy or major illness with limited life expectancy, lated HDL, polyclonal rabbit anti-human carbamylated any hospitalization in the preceding 3 months, or un- HDL antibody was raised with carbamylated HDL (Alfa willing to return for regular follow-up. Clinical data and Aesar, Lancashire, UK) as immunogen by a similar protocol fasting blood samples were collected at baseline. In this used to develop polyclonal rabbit anti-human carbamy- study, plasma carbamylated LDL and carbamylated HDL lated LDL antibody (10). Plasma carbamylated HDL was determined in all participants with available stored was assayed by sandwich ELISA using polyclonal rabbit blood samples and follow-up data. A subgroup of partic- anti-human carbamylated HDL antibody as the capture ipants with type 2 diabetes with eGFR .60 ml/min per antibody and polyclonal goat anti-human ApoA1-HRP 1.73 m2 were randomly chosen and compared with a antibody as the secondary detecting antibody. A 100-ml group of controls matched for age and sex, using a case- aliquot of 2.5 mg/ml of polyclonal rabbit anti-human control design to address the issue whether plasma levels carbamylated HDL antibody was coated overnight at 4°C of carbamylated LDL and carbamylated HDL were higher in pH 9.6 bicarbonate coating buffer. On the following day, in individuals with diabetes without kidney impairment. the plates were rinsed five times with washing buffer (0.1% Healthy individuals without medical illness and not re- Tween 20 and PBS) followed by blockade of nonspecific ceiving medications were recruited as control from the binding with 5% (w/v) BSA/PBS for 2 hours at room community by advertisement. Informed consent was temperature. A 100-ml aliquot at a 1:100 dilution of plasma obtained from all participants and the study was approved was added for a 2-hour incubation at room temperature. by the Ethics Committee of the University of Hong Kong After five washes, 100 ml of 1:1000 dilution of polyclonal goat (approval number UW14–040). anti-human ApoA1-horseradish peroxidase antibody (1:1000; Data collected from the longitudinal follow-up of the Biodesign) was added and incubated for 1 hour at room cohort were used to further investigate whether plasma temperature. At the end of the incubation, five washes were carbamylated lipoproteins were associated with progres- applied before 100 ml per well of 3,39,5,59-tetramethylbenzi- sion of kidney disease in patients with type 2 diabetes. dine substrate (TMB substrate kit; Pierce) was added for 2 m Participants with eGFR ,30 ml/min per 1.73 m and/or on 20 minutes followed with 50 l per well of 2 M H2SO4 as a KRT at baseline were excluded from the analysis because stop solution. The concentration of plasma carbamylated we intended to evaluate the association between carbamy- HDL was calculated using a 450 -nm absorbance standard lated lipoproteins and kidney disease progression in curve of purified human carbamylated HDL purchased from patients with diabetes with less advanced kidney disease. Alfa Aesar. Each sample was assayed in triplicate, and CJASN 15: ccc–ccc, March, 2020 Carbamylated HDL and Diabetic Kidney Disease, Tan et al. 3

results were expressed as microgram per milliliter car- carbamylated LDL (r50.13; P,0.001) and carbamylated bamylated HDL. The intra- and interassay coefficients of HDL (r50.18; P,0.001) also correlated with age but not variation were 6.8% and 9.6%, respectively. with body mass index (BMI), HbA1c, or triglyceride. There Results were expressed as mean and SD or as median were no significant correlations between plasma carbamy- and interquartile range if the data were not normally lated LDL and LDL cholesterol or between carbamylated distributed. Skewed data were logarithmically transformed HDL and HDL cholesterol. To determine whether plasma before analyses were made. Comparisons between two levels of carbamylated LDL and carbamylated HDL were groups and multiple groups were done using independent higher in patients with normal kidney function, plasma sample t test and ANOVA, respectively, and trend-test was carbamylated LDL and carbamylated HDL levels in a used to look for a trend effect of the variable on the random subgroup of patients with eGFR .60 ml/min per outcome. Pearson correlation coefficient was used to test 1.73 m2 and not receiving any lipid-lowering therapy were the relationships between variables. Multivariable Cox compared with levels in a group of nondiabetic controls regression analysis was used to estimate the hazard ratios matched for age and sex using a case-control design. (HRs) and 95% confidence intervals (95% CIs). The vari- Although patients with diabetes had similar eGFR and LDL ables included in Cox regression models were those that cholesterol and lower HDL cholesterol than controls were statistically significant in unadjusted analysis or were (P,0.001), both plasma carbamylated LDL and carbamy- biologically relevant. Participants with no available base- lated HDL was significantly higher (P,0.001) (Table 2). line serum samples or those with missing follow-up data These differences remained significant even after adjust- were excluded in the analysis. ing for BMI. To investigate whether plasma lipids and plasma carbamylated lipoproteins were associated with progression of Results kidney disease, the diabetic cohort was prospectively Out of the 1679 participants with diabetes who were followed up. Participants with eGFR ,30 ml/min per recruited, plasma carbamylated LDL and carbamylated 1.73 m2 and/or on KRT at baseline and those with missing HDL was measured in 1380 patients with available follow-up data were excluded (n5299), and data from 1320 blood samples and follow-up data (Figure 1). There was individuals were available for analysis (Figure 1). Doubling no sex difference in plasma carbamylated LDL but men of serum creatinine was observed in 270 individuals over a had higher plasma carbamylated HDL than women mean follow-up period of 965 years (71 patients required (38615 mg/ml versus 36614 mg/ml; P50.01). Participants dialysis and six received kidney transplant). The mortality were divided according to eGFR and the results are shown rate of the whole cohort was 8.3 per 1000 patient-years. in Table 1. As expected, there was a progressive increase in Patients with doubling of serum creatinine had higher plasma levels of carbamylated lipoproteins as kidney baseline plasma triglyceride (142 mg/dl [97–216], median P , – P, function worsened ( trend 0.001), and both plasma carba- [interquartile range], versus 115 mg/dl [80 168]; 0.001) mylated LDL (r520.20; P,0.001) and carbamylated HDL and LDL cholesterol (119638 mg/dl versus 109635 mg/dl; (r520.29; P,0.001) correlated with eGFR. Plasma P,0.001) but lower HDL cholesterol (45612 mg/dl versus

1679 subjects

299 subjects excluded - 124 no baseline serum samples - 175 with missing follow up data or already on kidney replacement therapy

1320 subjects 60 subjects with eGFR< with eGFR 30 30 ml/min/1.73m2 and ml/min/1.73m2 not on kidney replacement therapy

1050 subjects 270 subjects without doubling with doubling of creatinine on of creatinine follow up on follow up

Figure 1. | Flow chart describing the study participants. 4 CJASN

Table 1. Clinical characteristics and plasma carbamylated lipoproteins in individuals with diabetes according to baseline eGFR

eGFR, ml/min per 1.73 m2 Characteristic $60 $45 to ,60 $30 to ,45 ,30

N 1032 161 127 60 Age, yr 5369616863696268 Men/women, % 51/49 54/46 65/35 72/28 Duration of diabetes, yr 1267146816691568 BMI, kg/m2 2664266427642764 Smoker, % 10 7 8 13 Hypertension, % 59 83 99 98 Normo/micro/macroalbuminuria, % 60/31/9 32/39/29 10/30/60 0/13/87 Retinopathy, % 35 62 72 78 Cardiovascular disease, % 8 22 31 42 ACEI/ARB, % 41 54 70 53 Lipid-lowering therapy, % 21 38 58 68 Systolic BP, mm Hg 130619 139623 144628 147621 Diastolic BP, mm Hg 77610 766977611 75611 Fasting glucose, mg/dl 157648 148651 161672 148653 6 6 6 6 HbA1c,% 8.5 1.6 8.5 1.4 8.7 1.7 8.7 1.6 eGFR, ml/min per 1.73 m2 83616 546438642266 Total cholesterol, mg/dl 188639 187646 185642 180642 Triglyceride, mg/dl 115 (80–177) 133 (89–177) 151 (106–221) 142 (106–217) LDL cholesterol, mg/dl 112674 110638 105638 101636 HDL cholesterol, mg/dl 48613 47614 44612 44616 Carbamylated LDL, mg/ml 397 (269–409) 433 (249–525) 502 (306–612) 535 (304–650) Carbamylated HDL, mg/ml 34611 38612 48614 66628

Data are expressed as mean6SD or median (interquartile range). BMI, body mass index; ACEI/ARB, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker; HbA1c, hemoglobin A1c.

48613 mg/dl; P50.002) than those without doubling of cholesterol was associated with kidney outcome and creatinine. Both baseline plasma carbamylated LDL individuals with diabetes with HDL cholesterol in the (4556222 mg/ml versus 3986201 mg/ml; P,0.001) and top quartile had significantly lower risk of doubling of carbamylated HDL (46613 mg/ml versus 33611 mg/ml; serum creatinine than those in the first quartile (HR, 0.63; P,0.001) was significantly higher in the group with 95% CI, 0.45 to 0.90; P50.01). However, the association was doubling of serum creatinine even after adjusting for no longer significant after adjusting for baseline eGFR, age, sex, BMI, HbA1c, eGFR, smoking, duration of diabetes, albuminuria status, age, sex, BMI, duration of diabetes, albuminuria status, and lipid-lowering therapy. HbA1c,smoking,systolicBP,ACEI/ARB,andlipid- Neither baseline LDL cholesterol nor triglyceride was lowering therapy. Baseline clinical characteristics of the associated with doubling of serum creatinine. Serum HDL patients with diabetes, stratified by quartiles of

Table 2. Clinical characteristics and plasma carbamylated lipoproteins in individuals with diabetes with eGFR >60 ml/min per 1.73 m2 and the nondiabetic control group

Characteristic Control Diabetes

N 228 213 Age, yr 48674866 Men/women, % 50/50 49/51 BMI, kg/m2 25632664 Smoker, % 14 12 Duration of diabetes, yr — 9(6–14) Lipid-lowering therapy, % — 0 Fasting glucose, mg/dl 8769158655 6 6 HbA1c,% 5.6 0.5 8.5 1.6 eGFR, ml/min per 1.73 m2 81612 83616 Total cholesterol, mg/dl 189626 189631 Triglyceride, mg/dl 89 (62–124) 115 (80–177) LDL cholesterol, mg/dl 112624 113628 HDL cholesterol, mg/dl 55614 48612 Carbamylated LDL, mg/ml 267 (171–388) 340 (214–548) Carbamylated HDL, mg/ml 26610 31612

6 — Data are expressed as mean SD or median (interquartile range). BMI, body mass index; , not applicable; HbA1c, hemoglobin A1c. CJASN 15: ccc–ccc, March, 2020 Carbamylated HDL and Diabetic Kidney Disease, Tan et al. 5

carbamylated LDL and HDL, are shown in Supplemental in patients with good glycemic control, the data were . Tables 1 and 2, respectively. There was a graded associ- analyzed after excluding patients with HbA1c 7.5% and ation between increasing quartiles of carbamylated LDL the HR was 1.60 (95% CI, 1.26 to 2.03; P,0.001) in the fully and carbamylated HDL and kidney outcome, and individ- adjusted model. uals in the top quartile had the highest risk (Table 3). Per 1- SD change in carbamylated LDL and carbamylated HDL, the HR was 1.29 (95% CI, 1.17 to 1.43; P,0.001) and 1.94 Discussion (95% CI, 1.78 to 2.12; P,0.001), respectively, for the Several studies have shown that elevated levels of primary kidney outcome. On multivariable Cox regres- carbamylated proteins are associated with adverse prog- sion analysis, the association between plasma carbamy- nosis and higher mortality in patients with and without lated LDL and doubling of serum creatinine was no impaired kidney function and in patients with chronic longer significant after adjusting for baseline clinical heart failure (4,12–15). Because measures of systemic pro- characteristics, eGFR, and potential confounding factors tein carbamylation burden may be associated with cardio- (Supplemental Table 3). Nevertheless, the association vascular and mortality outcomes, it has been suggested between plasma carbamylated HDL and kidney outcome that select carbamylated proteins can serve as useful was still significant, and further adjusting for HDL biomarkers of disease and the presence of kidney impair- cholesterol and LDL cholesterol did not change the result ment is not necessarily a prerequisite (4,5). Thus far, all of (Table 4). This would suggest that the association be- the studies have evaluated the role of carbamylated tween carbamylated HDL and kidney outcome was proteins on cardiovascular end points and mortality, but independent of HDL cholesterol. Inclusion of carbamy- there are no available data on the relationship between lated LDL in the model also did not affect the result and carbamylated lipoproteins and the progression of diabetic there was no interaction between carbamylated HDL and kidney disease. Our study is the first prospective study to carbamylated LDL on kidney outcome. investigate the association between plasma carbamylated In addition, we evaluated whether carbamylated lipo- lipoproteins and CKD in patients with type 2 diabetes. First proteins were associated with rapid kidney progression of all, we have shown that both plasma carbamylated LDL (defined as a rate of eGFR decline .5 ml/min per year) that and carbamylated HDL was higher even in individuals was observed in 183 participants. Carbamylated HDL was with type 2 diabetes without CKD (eGFR .60 ml/min per significantly associated with rapid progression in the fully 1.73 m2). This is in keeping with our previous study adjusted model (HR, 1.43; 95% CI, 1.20 to 1.70; P,0.001) showing that increased carbamylation of LDL in patients (Table 4). Furthermore, 77 participants needed KRT and with type 2 diabetes with normal kidney function was carbamylated HDL was also associated with the risk of driven by myeloperoxidase activity owing to their elevated entering ESKD requiring KRT (HR, 1.68; 95% CI, 1.45 to levels of myeloperoxidase (10). As expected, the highest 2.20; P,0.001). There was no relationship between carba- plasma levels of carbamylated lipoproteins were seen in mylated LDL and rapid kidney progression or the risk of individuals with diabetes with advanced CKD, because entering ESKD. The mean HbA1c level of our diabetic protein carbamylation was predominantly driven by urea cohort was 8.5% and glycemic control was suboptimal. To in these individuals. determine whether the relationship between carbamylated We have further demonstrated that plasma carbamyla- HDL and the primary kidney outcome was still significant ted lipoproteins were associated with loss of kidney

Table 3. Associations of quartiles of plasma carbamylated lipoproteins with primary and secondary kidney outcomes

Doubling of Serum Creatinine Rapid Kidney Progression Quartile No. P No. P HR (95% CI) P HR (95% CI) P of Events Value trend of Events Value trend Carbamylated LDL ,274 mg/ml 43 Reference ,0.001 36 Reference 0.07 $274 and 57 1.25 (0.77 0.37 42 0.86 (0.58 0.43 ,397 mg/ml to 2.03) to 1.26) $397 and 78 1.96 (1.25 0.004 51 1.29 (0.71 0.10 ,459 mg/ml to 3.09) to 2.47) $459 mg/ml 92 2.21 (1.42 ,0.001 54 1.38 (0.93 0.11 to 3.46) to 2.05) Carbamylated HDL ,29 mg/ml 39 Reference ,0.001 35 Reference 0.005 $29 and ,33 mg/ml 43 1.04 (0.59 0.69 40 0.97 (0.60 0.65 to 1.84) to 1.42) $33 and ,42 mg/ml 70 1.65 (1.06 0.02 47 1.13 (0.66 0.35 to 2.76) to 1.94) $42 mg/ml 118 4.53 (2.87 ,0.001 61 1.98 (1.19 0.01 to 7.13) to 3.29)

HR, hazard ratio; 95% CI, 95% conﬁdence interval. CJASN 6

Table 4. Associations between plasma carbamylated HDL and primary and secondary kidney outcomes

Doubling of Serum Creatinine Rapid Kidney Progression

Parameter Model 1 Model 2 Model 1 Model 2

HR (95% CI) P Value HR (95% CI) P Value HR (95% CI) P Value HR (95% CI) P Value

Carbamylated HDL, mg/ml 1.55 (1.39 to 1.73) ,0.001 1.55 (1.38 to 1.73) ,0.001 1.43 (1.20 to 1.70) ,0.001 1.43 (1.20 to 1.70) ,0.001 Age, yr 1.34 (1.12 to 1.59) 0.001 1.34 (1.12 to 1.60) 0.001 1.22 (0.96 to 1.55) 0.10 1.23 (0.97 to 1.57) 0.09 Sex, men/women 0.68 (0.52 to 0.89) 0.005 0.69 (0.53 to 0.91) 0.009 1.09 (0.74 to 1.59) 0.67 1.16 (0.78 to 1.71) 0.46 BMI, kg/m2 1.27 (1.12 to 1.44) ,0.001 1.26 (1.10 to 1.43) 0.001 0.99 (0.82 to 1.19) 0.90 0.94 (0.78 to 1.14) 0.50 Duration of diabetes, yr 0.97 (0.86 to 1.10) 0.62 0.97 (0.86 to 1.10) 0.68 0.95 (0.79 to 1.15) 0.60 0.96 (0.79 to 1.16) 0.96 Smoker, no/yes 1.38 (0.89 to 2.14) 0.15 1.38 (0.89 to 2.15) 0.15 1.55 (0.86 to 2.86) 0.15 1.47 (0.81 to 2.66) 0.21 Systolic BP, mm Hg 1.18 (1.04 to 1.34) 0.01 1.18 (1.04 to 1.34) 0.01 1.52 (1.24 to 1.85) ,0.001 1.50 (1.22 to 1.83) ,0.001 , , , , HbA1c, % 1.29 (1.14 to 1.47) 0.001 1.29 (1.14 to 1.47) 0.001 1.41 (1.19 to 1.67) 0.001 1.43 (1.21 to 1.70) 0.001 Baseline eGFR, ml/min per 1.73 m2 0.52 (0.43 to 0.63) ,0.001 0.53 (0.44 to 0.63) ,0.001 0.84 (0.71 to 0.98) 0.03 0.85 (0.72 to 0.99) 0.04 Normoalbuminuria Reference ,0.001 Reference ,0.001 Reference ,0.001 Reference ,0.001 Microalbuminuria 2.64 (1.85 to 3.78) ,0.001 2.65 (1.84 to 3.80) ,0.001 3.20 (2.01 to 5.09) ,0.001 3.27 (2.04 to 5.24) ,0.001 Macroalbuminuria 8.13 (5.47 to 12.07) ,0.001 8.29 (5.55 to 12.35) ,0.001 15.12 (8.79 to 26.0) ,0.001 16.05 (9.25 to 27.85) ,0.001 ACEI/ARB, no/yes 1.11 (0.86 to 1.45) 0.42 1.12 (0.85 to 1.46) 0.43 1.11 (0.77 to 1.61) 0.58 1.11 (0.76 to 1.61) 0.61 Lipid-lowering therapy, no/yes 1.28 (0.96 to 1.52) 0.11 1.23 (0.97 to 1.56) 0.08 1.13 (0.81 to 1.58) 0.47 1.13 (0.80 to 1.61) 0.49 LDL cholesterol, mg/dl — 1.04 (0.92 to 1.19) 0.54 — 1.03 (0.85 to 1.24) 0.80 HDL cholesterol, mg/dl — 0.95 (0.83 to 1.09) 0.49 — 0.82 (0.63 to 1.01) 0.05

The given hazard ratio is for 1-SD change in plasma carbamylated HDL and continuous covariates at baseline. Model 1: adjusted for carbamylated HDL, age, sex, BMI, duration of diabetes, smoking, systolic BP, HbA1c , baseline eGFR, albuminuria status, ACEI/ARB therapy, and lipid-lowering therapy. Model 2: further adjusted for LDL cholesterol and HDL cholesterol. HR, hazard ﬁ — ratio; 95% CI, 95% con dence interval; BMI, body mass index; HbA1c , hemoglobin A1c ; ACEI/ARB, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker; , not applicable. CJASN 15: ccc–ccc, March, 2020 Carbamylated HDL and Diabetic Kidney Disease, Tan et al. 7

function in our cohort of patients with type 2 diabetes. The initiation of lipid-lowering therapy and progression of majority of previous studies have focused on the vascular disease were not taken into account. Furthermore, kidney effects of carbamylated lipoproteins and the process of biopsy was only performed in a small number of partic- carbamylation increases the atherogenic properties of LDL ipants to ascertain the definitive cause of diabetic kidney and impairs the protective function of HDL (8,16,17). Our disease. Another limitation is that our cohort of patients is data show that mainly carbamylated HDL level was likely to be more complex than those in primary care or in associated with the progression of CKD, as plasma carba- the general population because our unit is a secondary/ mylated HDL but not carbamylated LDL remained an tertiary referral center, and we cannot completely exclude independent determinant of kidney decline even after selection bias in the recruitment process. Hence, our results adjustment for baseline eGFR, albuminuria status, and may not be generalized to other patient populations. other potential confounding factors. This would suggest Missing baseline samples and follow-up data reduced that plasma carbamylated LDL only reflected kidney func- the sample size for analysis and may also bias our results, tion, whereas plasma carbamylated HDL might probably but there were no significant differences between those have a pathogenic role. This is in keeping with the results of with complete data and those with missing data. a recent Mendelian randomization study showing that only In conclusion, both plasma carbamylated LDL and HDL was associated with kidney function. Analysis of data carbamylated HDL were significantlyhigher in individuals from the largest lipid and CKD cohorts demonstrated that with type 2 diabetes with and without kidney impairment. genetically higher HDL cholesterol concentration was Only plasma carbamylated HDL but not carbamylated causally associated with better kidney function, but there LDL was independently associated with progression of was no association between genetically altered LDL kidney disease in type 2 diabetes. cholesterol or triglyceride concentration and kidney function (18). Acknowledgments There is substantial evidence suggesting that altered Prof. Tan designed and oversaw the study, data collection, HDL metabolism may contribute to the pathogenesis and analysis and manuscript writing. Dr. Cheung contributed to study progression of kidney disease. Recent studies have dem- design and analyzed the data. Dr. Shiu performed the laboratory fi onstrated a signi cant association between low HDL assays and analyzed the data. Dr. Lam, Dr. Lee, and Dr. Wong re- cholesterol levels and risks of incident CKD and CKD cruited the subjects and collected clinical data. All authors approved progression (19,20). In patients with type 2 diabetes from the final version of the manuscript. Prof. Tan is the guarantor the Action in Diabetes and Vascular Disease: preterAx of this work. and diamicroN-MR Controlled Evaluation (ADVANCE) study, low HDL cholesterol level was shown to be an Disclosures independent risk factor for the development and progres- Dr. Cheung, Dr. Lam, Dr. Lee, Dr. Shiu, Prof. Tan, and Dr. Wong sion of diabetic nephropathy (21). A global case-control have nothing to disclose. study also showed an association between HDL cholesterol with kidney disease in type 2 diabetes (22). We have found Funding that carbamylated HDL was associated with the progres- This study is supported by an Endowment Fund established for sion of CKD independent of HDL cholesterol in our study. the Sir David Todd Professorship in Medicine, University of Hong It has been shown that carbamylated HDL is a better metric Kong, awarded to Prof. Tan. of HDL function than HDL cholesterol. Carbamylation of HDL leads to HDL dysfunction and results in a loss of anti- Supplemental Material inflammatory and antioxidative properties of HDL (7,23). This article contains the following supplemental material online at Carbamylated HDL is also defective in promoting reverse http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN. cholesterol transport and facilitates cholesterol accumula- 11710919/-/DCSupplemental. tion in cells and tissues (24,25). Although the pathophys- Supplemental Table 1. Clinical characteristics of individuals with iologic process linking dyslipidemia to diabetic kidney diabetes according to quartiles of carbamylated LDL. disease is still unclear, one hypothesis is that dysfunctional Supplemental Table 2. Clinical characteristics of individuals with HDL particles impair reverse cholesterol transport in the diabetes according to quartiles of carbamylated HDL. kidney and contribute to intrarenal lipid accumulation, Supplemental Table 3. Association between plasma carbamylated leading to glomerulosclerosis and tubulointerstitial dam- LDL and primary kidney outcome. age (9,26). Hence, the production of nonfunctioning HDL particles by carbamylation may therefore also be a cause References and a consequence of CKD. Whether carbamylated lipo- 1. Jaisson S, Pietrement C, Gillery P: Protein carbamylation: proteins (especially carbamylated HDL) are involved in the Chemistry, pathophysiological involvement, and biomarkers. pathogenesis and progression of diabetic kidney disease or Adv Clin Chem 84: 1–38, 2018 are just markers of CKD warrants further investigation. 2. Nilsson L, Lundquist P, Ka˚gedal B, Larsson R: Plasma cyanate concentrations in chronic renal failure. Clin Chem 42: 482–483, Our study has several limitations. The main limitation is 1996 that we have not performed any functional measurement of 3. Kalim S, Karumanchi SA, Thadhani RI, Berg AH: Protein carba- HDL particles, like cholesterol efflux capacity. Experimen- mylation in kidney disease: Pathogenesis and clinical implica- tal studies have shown that carbamylation of HDL affects tions. Am J Kidney Dis 64: 793–803, 2014 multiple functions of HDL (7,17,23–25). We do not have 4. Wang Z, Nicholls SJ, Rodriguez ER, Kummu O, Ho¨rkko¨ S, Barnard J, Reynolds WF, Topol EJ, DiDonato JA, Hazen SL: Protein car- serial measurement of carbamylated lipoproteins over bamylation links inflammation, smoking, uremia and athero- time, and changes in treatment like the subsequent genesis. Nat Med 13: 1176–1184, 2007 8 CJASN

5. Delanghe S, Delanghe JR, Speeckaert R, Van Biesen W, 19. Baragetti A, Norata GD, Sarcina C, Rastelli F, Grigore L, Speeckaert MM: Mechanisms and consequences of carbamoy- Garlaschelli K, Uboldi P, Baragetti I, Pozzi C, Catapano AL: High lation. Nat Rev Nephrol 13: 580–593, 2017 density lipoprotein cholesterol levels are an independent pre- 6. Apostolov EO, Ray D, Savenka AV, Shah SV, Basnakian AG: dictor of the progression of chronic kidney disease. J Intern Med Chronic uremia stimulates LDL carbamylation and atheroscle- 274: 252–262, 2013 rosis. J Am Soc Nephrol 21: 1852–1857, 2010 20. Bowe B, Xie Y,Xian H, Balasubramanian S, Al-Aly Z: Low levels of 7. Sun JT,Yang K, Lu L, Zhu ZB,Zhu JZ, Ni JW, Han H, Chen N, Zhang high-density lipoprotein cholesterol increase the risk of incident RY: Increased carbamylation level of HDL in end-stage renal kidney disease and its progression. Kidney Int 89: 886–896, 2016 disease:Carbamylated-HDL attenuated endothelial cell function. 21. Morton J, Zoungas S, Li Q, Patel AA, Chalmers J, Woodward M, Am J Physiol Renal Physiol 310: F511–F517, 2016 Celermajer DS, Beulens JW, Stolk RP, Glasziou P, Ng MK; AD- 8. Apostolov EO, Basnakian AG, Ok E, Shah SV: Carbamylated low- VANCE CollaborativeGroup: Low HDL cholesterol and the risk of density lipoprotein: Nontraditional risk factor for cardiovascular diabetic nephropathy and retinopathy: Results of the ADVANCE events in patients with chronic kidney disease. J Ren Nutr 22: study. Diabetes Care 35: 2201–2206, 2012 134–138, 2012 22. Sacks FM, Hermans MP, Fioretto P, Valensi P, Davis T, Horton E, 9. Wahl P, Ducasa GM, Fornoni A: Systemic and renal lipids in Wanner C, Al-Rubeaan K, Aronson R, Barzon I, Bishop L, Bonora kidney disease development and progression. Am J Physiol Renal E, Bunnag P, Chuang LM, Deerochanawong C, Goldenberg R, Physiol 310: F433–F445, 2016 Harshﬁeld B, Herna´ndez C, Herzlinger-Botein S, Itoh H, Jia W, 10. Shiu SW, Xiao SM, Wong Y, Chow WS, Lam KS, Tan KC: Carba- Jiang YD, Kadowaki T, Laranjo N, Leiter L, Miwa T, Odawara M, mylation of LDL and its relationship with myeloperoxidase in type Ohashi K, Ohno A, Pan C, Pan J, Pedro-Botet J, Reiner Z, Rotella 2 diabetes mellitus. Clin Sci (Lond) 126: 175–181, 2014 CM, Simo R, Tanaka M, Tedeschi-Reiner E, Twum-Barima D, 11. Tan KCB, Cheung CL, Lee ACH, Lam JKY, Wong Y, Shiu SWM: Zoppini G, Carey VJ: Association between plasma triglycerides Galectin-3 is independently associated with progression of ne- and high-density lipoprotein cholesterol and microvascular phropathy in type 2 diabetes mellitus. Diabetologia 61: kidney disease and retinopathy in type 2 diabetes mellitus: A 1212–1219, 2018 global case-control study in 13 countries. Circulation 129: 12. Koeth RA, Kalantar-Zadeh K, Wang Z, Fu X, Tang WH, Hazen SL: 999–1008, 2014 Protein carbamylation predicts mortality in ESRD. JAmSoc 23. Holzer M, Gauster M, Pfeifer T, Wadsack C, Fauler G, Stiegler P, Nephrol 24: 853–861, 2013 Koefeler H, Beubler E, Schuligoi R, Heinemann A, Marsche G: 13. Tang WH, Shrestha K, Wang Z, Borowski AG, Troughton RW, Protein carbamylation renders high-density lipoprotein dys- Klein AL, Hazen SL: Protein carbamylation in chronic systolic functional. Antioxid Redox Signal 14: 2337–2346, 2011 heart failure: Relationship with renal impairment and adverse 24. Holzer M, Zangger K, El-Gamal D, Binder V, Curcic S, Konya V, long-term outcomes. J Card Fail 19: 219–224, 2013 Schuligoi R, Heinemann A, Marsche G: Myeloperoxidase- 14. Drechsler C, Kalim S, Wenger JB, Suntharalingam P, Hod T, derived chlorinating species induce protein carbamylation Thadhani RI, Karumanchi SA, Wanner C, Berg AH: Protein car- through decomposition of thiocyanate and urea: Novel pathways bamylation is associated with heart failure and mortality in di- generating dysfunctional high-density lipoprotein. Antioxid Re- abetic patients with end-stage renal disease. Kidney Int 87: dox Signal 17: 1043–1052, 2012 1201–1208, 2015 25. Anderson JL, Gautier T,Nijstad N, To¨lle M, Schuchardt M, van der 15. Jaisson S, Kerkeni M, Santos-Weiss IC, Addad F, Hammami M, Giet M, Tietge UJ: High density lipoprotein (HDL) particles from Gillery P: Increased serum homocitrulline concentrations are end-stage renal disease patients are defective in promoting re- associated with the severity of coronary artery disease. Clin Chem verse cholesterol transport. Sci Rep 7: 41481, 2017 Lab Med 53: 103–110, 2015 26. Vaziri ND: Lipotoxicity and impaired high density lipoprotein- 16. Sirpal S: Myeloperoxidase-mediated lipoprotein carbamylation mediated reverse cholesterol transport in chronic kidney disease. as a mechanistic pathway for atherosclerotic vascular disease. J Ren Nutr 20[Suppl]: S35–S43, 2010 Clin Sci (Lond) 116: 681–695, 2009 17. Santana JM, Brown CD: High-density lipoprotein carbamylation and Received: Accepted: dysfunction in vascular disease. Front Biosci 23: 2227–2234, 2018 September 27, 2019 January 7, 2020 18. Lanktree MB, The´riault S, Walsh M, Pare´ G: HDL cholesterol, LDL cholesterol, and triglycerides as risk factors for CKD: A Mendelian Published online ahead of print. Publication date available at randomization study. Am J Kidney Dis 71: 166–172, 2018 www.cjasn.org. Supplemental material is neither peer-reviewed nor thoroughly edited by CJASN. The authors alone are responsible for the accuracy and presentation of the material.

Supplemental Material Table of Contents:

Supplemental Table 1: Clinical characteristics of diabetic individuals according to quartiles of carbamylated LDL

Supplemental Table 2: Clinical characteristics of diabetic individuals according to quartiles of carbamylated HDL

Supplemental Table 3: Association between plasma carbamylated LDL and primary kidney outcome

Supplemental material is neither peer-reviewed nor thoroughly edited by CJASN. The authors alone are responsible for the accuracy and presentation of the material.

Supplemental Table 1: Clinical characteristics of diabetic individuals according to quartiles of carbamylated LDL

Quartiles of carbamylated LDL Q1 Q2 Q3 Q4 Age (years) 54 ± 9 53 ± 10 57 ± 10 58 ± 10 M/F (%) 54/46 53/47 54/46 53/47 Duration of diabetes (years) 13 ± 7 11 ± 7 13 ± 8 15 ± 8 BMI (kg/m2) 26 ± 5 26 ± 4 27 ± 4 26 ± 4 Smoker (%) 10 11 8 10 Hypertension (%) 66 64 75 82 Normo/micro/macroalbuminuria (%) 56/28/16 53/37/10 46/34/20 45/30/25 Retinopathy (%) 43 41 48 50 Cardiovascular disease (%) 14 12 13 16 ACEI/ARB (%) 41 50 48 53 Lipid lowering therapy (%) 28 31 28 32 Systolic BP (mmHg) 130 ± 21 131 ± 19 137 ± 20 133 ± 24 Diastolic BP (mmHg) 77 ± 9 76 ± 10 77 ± 9 77 ± 9 Fasting glucose (mg/dl) 160 ± 50 153 ± 49 146 ± 45 162 ± 58 HbA1c (%) 8.7 ± 1.7 8.3 ± 1.6 8.3 ± 1.4 8.7 ± 1.6 eGFR (ml/min/1.73 m2) 78 ± 21 80 ± 18 68 ± 22 68 ± 23 Total cholesterol (mg/dl) 174 ± 31 193 ± 43 197 ± 46 189 ± 39 Triglyceride (mg/dl) 106 (80 - 168) 115 (80 - 177) 142 (96 - 186) 124 (89 - 186) LDL-cholesterol (mg/dl) 97 ± 27 116 ± 39 116 ± 39 116 ± 35 HDL-cholesterol (mg/dl) 50 ± 12 46 ± 12 46 ± 15 46 ± 12

Data are expressed as mean ± SD or median (interquartile range).

Supplemental material is neither peer-reviewed nor thoroughly edited by CJASN. The authors alone are responsible for the accuracy and presentation of the material.

Supplemental Table 2: Clinical characteristics of diabetic individuals according to quartiles of carbamylated HDL

Quartiles of carbamylated HDL Q1 Q2 Q3 Q4 Age (years) 55 ± 10 53 ± 10 55 ± 10 58 ± 9 M/F (%) 48/52 52/48 52/48 59/41 Duration of diabetes (years) 13 ± 7 12 ± 8 13 ± 8 14 ± 8 BMI (kg/m2) 27 ± 5 26 ± 4 26 ± 4 26 ± 4 Smoker (%) 11 10 8 9 Hypertension (%) 66 57 70 77 Normo/micro/macroalbuminuria (%) 58/31/11 62/28/10 49/36/15 32/38/30 Retinopathy (%) 41 45 43 52 Cardiovascular disease (%) 13 10 13 19 ACEI/ARB (%) 40 49 52 50 Lipid lowering therapy (%) 26 27 30 35 Systolic BP (mmHg) 130 ± 22 129 ± 18 133 ± 22 137 ± 23 Diastolic BP (mmHg) 77 ± 9 76 ± 10 78 ± 9 76 ± 10 Fasting glucose (mg/dl) 157 ± 49 155 ± 49 157 ± 45 157 ± 59 HbA1c (%) 8.6 ± 1.6 8.5 ± 1.6 8.5 ± 1.5 8.5 ± 1.6 eGFR (ml/min/1.73 m2) 78 ± 18 80 ± 18 74 ± 22 64 ± 23 Total cholesterol (mg/dl) 182 ± 35 189 ± 39 193 ± 43 189 ± 43 Triglyceride (mg/dl) 115 (80 - 177) 124 (80 - 177) 115 (80 - 177) 124 (80 – 186) LDL-cholesterol (mg/dl) 104 ± 35 112 ± 35 116 ± 39 112 ± 35 HDL-cholesterol (mg/dl) 46 ± 12 46 ± 15 46 ± 13 46 ± 12

Data are expressed as mean ± SD or median (interquartile range).

Supplemental material is neither peer-reviewed nor thoroughly edited by CJASN. The authors alone are responsible for the accuracy and presentation of the material.

Supplemental Table 3: Association between plasma carbamylated LDL and primary kidney outcome (doubling of serum creatinine)

Model 1 Model 2 Hazard ratio (95% CI) p value Hazard ratio (95% CI) p value Carbamylated LDL 1.05 (0.93, 1.19) 0.43 1.04 (0.91, 1.19) 0.55 (µg/ml) Age (years) 1.35 (1.13, 1.60) 0.001 1.34 (1.12, 1.60) 0.001 Sex (M/F) 0.65 (0.50, 0.85) 0.001 0.66 (0.50, 0.87) 0.003 BMI (kg/m2) 1.29 (1.13, 1.46) <0.001 1.27 (1.11, 1.45) <0.001 Duration of diabetes 0.95 (0.84, 1.08) 0.46 0.96 (0.85, 1.09) 0.56 (years) Smoker (N/Y) 1.11 (0.72, 1.72) 0.63 1.11 (0.71, 1.72) 0.65 Systolic BP (mmHg) 1.15 (1.01, 1.30) 0.03 1.15 (1.02, 1.31) 0.03

HbA1c (%) 1.27 (1.12, 1.43) <0.001 1.27 (1.11, 1.44) <0.001 Baseline eGFR 0.42 (0.35, 0.51) <0.001 0.42 (0.35, 0.51) <0.001 (ml/min/1.73 m2) Normoalbuminuria Reference <0.001 Reference <0.001 Microalbuminuira 2.89 (2.03, 4.11) <0.001 2.91 (2.04, 4.16) <0.001 Macroalbuminuira 8.14 (5.53, 11.98) <0.001 8.36 (5.65, 12.37) <0.001 ACEI/ARB(N/Y) 1.06 (0.82, 1.38) 0.66 1.07 (0.82, 1.40) 0.61 Lipid lowering 1.28 (0.92, 1.61) 0.10 1.31 (0.92, 1.62) 0.08 therapy(N/Y) LDL-cholesterol - 1.08 (0.95, 1.23) 0.26 (mg/dl) HDL-cholesterol - 0.93 (0.81, 1.07) 0.31 (mg/dl)

The given hazard ratio is for 1-SD change in plasma carbamylated LDL and continuous covariates at baseline. Model 1: adjusted for carbamylated LDL, age, sex, BMI, duration of diabetes, smoking, systolic BP, HbA1c, baseline eGFR, albuminuria status, ACEI/ARB therapy and lipid lowering therapy. Model 2: further adjusted for LDL-cholesterol and HDL-cholesterol.