Protein carbamylation and chronic kidney disease progression in the Chronic Renal Insufficiency Cohort Study

Abstract

Background

Protein carbamylation is a post-translational protein modification caused, in part, by exposure to urea’s dissociation product cyanate. Carbamylation is linked to cardiovascular outcomes and mortality in dialysis-dependent end-stage kidney disease (ESKD), but its effects in earlier pre-dialysis stages of chronic kidney disease (CKD) are not established.

Methods

We conducted two nested case–control studies within the Chronic Renal Insufficiency Cohort Study. First, we matched 75 cases demonstrating CKD progression [50% estimated glomerular filtration rate (eGFR) reduction or reaching ESKD] to 75 controls (matched on baseline eGFR, 24-h proteinuria, age, sex and race). In the second study, we similarly matched 75 subjects who died during follow-up (cases) to 75 surviving controls. Baseline carbamylated albumin levels (C-Alb, a validated carbamylation assay) were compared between cases and controls in each study.

Results

At baseline, in the CKD progression study, other than blood urea nitrogen (BUN) and smoking status, there were no significant differences in any matched or other parameter. In the mortality group, the only baseline difference was smoking status. Adjusting for baseline differences, the top tertile of C-Alb was associated with an increased risk of CKD progression [odds ratio (OR) = 7.9; 95% confidence interval (CI) 1.9–32.8; P = 0.004] and mortality (OR = 3.4; 95% CI 1.0–11.4; P = 0.05) when compared with the bottom tertile. C-Alb correlated with eGFR but was more strongly correlated with BUN.

Conclusions

Our data suggest that protein carbamylation is a predictor of CKD progression, beyond traditional risks including eGFR and proteinuria. Carbamylation’s association with mortality was smaller in this limited sample size.

Graphical Abstract

carbamylation, CKD, mortality, progression, urea

KEY LEARNING POINTS

What is already known about this subject?

• protein carbamylation describes a post-translational protein modification that stems from a variety of causes including urea exposure in kidney disease;

• carbamylation results in molecular and cellular dysfunction and has been shown to associate with mortality in end-stage kidney disease;

• however, the clinical impact of carbamylation at earlier pre-dialysis stages of chronic kidney disease (CKD) is uncertain.

What this study adds?

• this study shows protein carbamylation independently associates with CKD progression and death in a diverse cohort of pre-dialysis CKD patients.

What impact this may have on practice or policy?

• carbamylation has been shown to be a modifiable process through interventions such a nutritional and dietary changes; and

• given the predialysis clinical risks observed, carbamylation could serve as a therapeutic target in patients with CKD.

INTRODUCTION

Protein carbamylation describes a non-enzymatic, post-translational protein modification caused by cyanate, which notably is the reactive dissociation product of urea (Figure 1). While generally found in low concentrations in vivo, as blood urea increases with impaired kidney function, so does cyanate and thus carbamylation burden [1]. Importantly, protein carbamylation is not solely related to urea; it has also been shown that free amino acids compete with proteins for reaction with cyanate, and amino acid deficiencies from a variety of causes can exacerbate carbamylation burden [2]. Additionally, the heme protein myeloperoxidase (MPO), which is secreted in high concentrations at inflammatory sites from stimulated neutrophils and monocytes, can react with environmentally derived thiocyanate (from diet, smoking and even air pollution) to generate cyanate and thus promote the carbamylation reaction to occur (Figure 1) [3, 4].

FIGURE 1

The carbamylation reaction.

Irreversible carbamylation reactions with amino groups can change the charge, structure and function of proteins, resulting in molecular and cellular dysfunction [5]. Such changes have been linked to several pathologic biochemical pathways relevant to patients with kidney disease such as accelerated atherosclerosis, kidney fibrosis and aging [5–9]. In several well-conducted studies with replication and using different carbamylation assays, carbamylation burden has been linked to mortality and cardiovascular outcomes in patients with both preserved kidney function as well as end-stage kidney disease (ESKD) [2, 3, 10]. To date, studies of the impact of protein carbamylation in nondialysis chronic kidney disease (CKD) patients are noticeably scant despite clear relevance and strong mechanistic plausibility that carbamylation could influence outcomes such as CKD progression and mortality in this population. The purpose of this investigation was to establish relative differences in carbamylation load between nondialysis CKD patients in the Chronic Renal Insufficiency Cohort (CRIC) Study who experienced significant CKD progression or mortality and those who did not experience such adverse outcomes.

MATERIALS AND METHODS

Study population

We created two independent nested case–control study populations within the CRIC Study, one for CKD progression and the other for all-cause mortality. The CRIC Study is a multicenter, prospective, observational cohort study of risk factors for cardiovascular disease (CVD), progression of CKD and mortality [11, 12]. From 1 January 2003 to 30 September 2008, a total of 3939 individuals aged 21–74 years with an estimated glomerular filtration rate (eGFR) of 20–70 mL/min/1.73 m² were enrolled across seven US clinical centers. Plasma samples used for this analysis were collected at the Year 1 (Y1) CRIC visit (1 year after initial subject enrollment) and included participants who were free of ESKD and had necessary plasma sample availability at the Y1 time. The study protocol was approved by institutional review boards at participating institutions, and the research was conducted in accordance with the ethical principles of the declaration of Helsinki.

Case–control selection

Figure 2 describes the case–control selection process for this study. After exclusions, 3046 participants were eligible at Y1. Of these, 786 experienced subsequent CKD progression (50% reduction in eGFR or reaching ESKD, see ‘Ascertainment of outcomes’ below) from the Y1 visit. From this group with CKD progression, 75 individuals were randomly selected and matched to 75 subjects who did not demonstrate CKD progression across the study period (from a potential pool of 1657 subjects). Frequency matching was based on Y1 eGFR, proteinuria, age, sex and race. Using similar methods, from a group of 784 distinct individuals who died during the subsequent study follow-up period, we randomly selected 75 and matched them to 75 survivors using the same matching criteria. Because no data existed on CKD progression and carbamylation, the sample size was selected to power the study at >80% to detect a mortality difference between cases and controls. This was based on assumptions using the most relevant existing data reporting the mortality association with carbamylated albumin (C-Alb) in ESKD dialysis patients [2, 13].

FIGURE 2

Flowchart of case–control assembly for the study. Note, prior to case selection, the groups of study participants are not mutually exclusive.

Exposure assay

Carbamylation exposure was measured using C-Alb levels from Y1 plasma samples (C-Alb, a validated measure of carbamylation burden), which were measured as the ratio of millimoles of C-Alb per mole of total albumin (analogous to the measurement of percent glycated hemoglobin) using high-performance liquid chromatography and tandem mass spectrometry. A complete description of the mass spectrometric assay for C-Alb and its analytical validation have been previously described [2]. All samples were analyzed over a 1-week period with blinding to case–control status of the samples. Repeat measurements of representative patient samples analyzed as an indicator of quality control demonstrated an intraassay coefficient of variance of 2.0% and an interassay coefficient of variance of 2.6%.

Covariates

Sociodemographic characteristics, medical history, lifestyle behaviors and other clinical data were obtained at the Y1 visit. Laboratory measurements using standard assays were also performed on samples from the Y1 visit. For this study, only MPO was not obtained at Y1 as it was only measured at CRIC participant enrollment. The eGFR was calculated using the CRIC-derived estimating equation [14]. Proteinuria was measured from the Y1 24-h urine collection.

Ascertainment of outcomes

CRIC Study participants were followed annually with clinic visits and interim telephone contact at 6 months. The primary outcomes, chosen a priori, were CKD progression defined as a composite of incident ESKD or a 50% decline in eGFR from baseline. ESKD was defined as receipt of long-term dialysis or a kidney transplant. Patient follow-up was censored at the first occurrence of voluntary study withdrawal, loss to follow-up, end of the follow-up period or death. The secondary outcome of interest was all-cause mortality, which was confirmed by death certificate review during the CRIC Study.

Statistical analysis

The balance between cases and controls before and after matching was assessed using standardized differences and can be seen in Table 1 and Supplementary data, Table S1. We expressed continuous variables as means [standard deviations (SDs)] or medians [interquartile ranges (IQRs)] and compared them with parametric or nonparametric tests, as appropriate. Because of an observed non-Gaussian frequency of C-Alb values and nonlinear relationships, we stratified participants into tertiles (lowest tertile as reference) based on their baseline C-Alb levels. The risk of CKD progression and all-cause mortality were examined for each tertile. The Cochran–Armitage test was performed to test a monotonic trend across C-Alb tertiles. We used conditional logistic regression models to estimate the association between baseline C-Alb and CKD progression and then all-cause mortality. Potential confounders were selected based on imbalances noted between cases and controls as well as biological plausibility. This included urea and smoking status for both the primary CKD progression analysis and the all-cause mortality analysis. To allay concerns for residual confounding, additional comorbidities and all matched parameters we added to a second multivariable model. Statistical significance was set at P < 0.05.

Table 1.

Baseline characteristics of the study populations by case–control status

Variable	CKD progression case (n = 75)	CKD progression control (n = 75)	P-value	Mortality case (n = 75)	Survivor control (n = 75)	P-value
Age, years	58.4 (9.5)	57.6 (9.8)	0.62	63.7 (7.4)	62.6 (8.1)	0.41
Female, n (%)	37 (49.3)	39 (52.0)	0.74	27 (36.0)	26 (34.7)	0.86
Race, n (%)
White	19 (25.3)	17 (22.7)		35 (46.7)	36 (48.0)
Black	47 (62.7)	49 (65.3)		35 (46.7)	34 (45.3)
Hispanic	7 (9.3)	8 (10.7)		4 (5.3)	4 (5.3)
Other	2 (2.7)	1 (1.3)	0.93	1 (1.3)	1 (1.3)	1.00
eGFR—CRIC equation, mL/min/1.73 m2	33.2 (22.3–38.1)	34.4 (28.0–38.8)	0.14	33.4 (23.3–42.6)	33.7 (25.5–41.7)	0.99
Proteinuria, g/24 h	1.2 (0.4–3.9)	1.0 (0.3–2.4)	0.23	0.3 (0.1–0.8)	0.3 (0.1–1.0)	0.85
Urea, mg/dL	32.0 (26.0–46.0)	30.0 (24.0–38.0)	0.05	36.0 (27.0–46.0)	33.0 (26.0–43.0)	0.61
Albumin, g/dL	3.9 (3.6–4.1)	3.9 (3.7–4.3)	0.24	4.0 (3.7–4.3)	4.1 (3.9–4.3)	0.20
HDL, mg/dL	45.0 (36.0–55.0)	44.0 (37.0–58.0)	0.90	43.0 (34.0–54.0)	46.0 (39.0–55.0)	0.23
MPOa, pmol/L	121.0 (90.7–195.1)	119.4 (86.8–176.3)	0.42	112.1 (85.3–144.6)	116.1 (83.6–138.7)	0.67
BMI, kg/m2	32.3 (7.8)	34.4 (7.8)	0.08	31.6 (6.4)	32.6 (9.5)	0.44
Systolic blood pressure, mmHg	135.7 (25.1)	129.5 (22.7)	0.11	129.8 (25.5)	125.8 (19.5)	0.29
Current smoking, n (%)	21 (28.0)	9 (12.0)	0.01	19 (25.3)	8 (10.7)	0.02
Diabetes, n (%)	46 (61.3)	43 (57.3)	0.62	47 (62.7)	40 (53.3)	0.25
CVD, n (%)	37 (49.3)	29 (38.7)	0.19	40 (53.3)	30 (40.0)	0.10
PVD, n (%)	8 (10.7)	5 (6.7)	0.38	9 (12.0)	6 (8.0)	0.41
Cause of CKD, n (%)
Diabetes	21 (28.0)	23 (30.7)		25 (33.3)	21 (28.0)
Hypertension	14 (18.7)	14 (18.7)	0.80	11 (14.7)	12 (16.0)	0.90
Other	8 (10.7)	11 (14.7)		10 (13.3)	12 (16.0)
Do not know	32 (42.7)	27 (36.0)		29 (38.7)	30 (40.0)
C-Alb, mmol/mol	6.9 (5.3–9.6)	5.0 (4.3–7.1)	0.0003	7.1 (5.4–10.3)	6.5 (5.2–8.5)	0.14

a

All values derived from CRIC Y1 study visit with exception of MPO derived at CRIC Study baseline. Categorical data are n (%). Continuous measures are means (SDs). Laboratory values are median (quartiles 1–3). Standardized differences are presented in Supplementary data, Table S1.

BMI, body mass index; PVD, peripheral vascular disease.

To explore the associations between C-Alb and eGFR, urea and albumin (selected a priori), correlation coefficients were generated by standardizing the variables and using linear regressions that controlled for the interdependence between cases and matched controls. This analysis was performed for all participants by pooling both the CKD progression and the mortality populations together, as well as for each of the two populations separately. To test the difference between the two correlation coefficients, we used Hittner’s modification of Dunn and Clark’s z test to account for the interdependence within the same participants [15, 16].

RESULTS

CKD progression

Baseline characteristics of the matched CKD progression cases (individuals experiencing a 50% decline in eGFR or ESKD, n = 75) and controls (matched on baseline eGFR, proteinuria, age, sex and race, n = 75) are shown in Table 1 (case–control selection flowchart shown in Figure 2). There were no statistical differences in the distributions of age, sex, race, baseline eGFR and 24-h proteinuria, though CKD progression cases trended toward worse renal indices compared with controls, with median baseline eGFR being 33.2 versus 34.4 mL/min/1.73 m² (P = 0.14), and 24-h proteinuria at 1.2 g/24 h versus 1.0 g/24 h (P = 0.23), respectively. Notably, blood urea nitrogen (BUN; 32 versus 30 mg/dL) was marginally higher in cases (P = 0.05) and there were more current smokers among the cases (28% versus 12%; P = 0.01). The median (IQR) time to reaching the CKD progression endpoint in our sample was 2.9 years (1.7–5.1) among the cases, compared with 8.2 years (4.7–9.2) of follow-up among the controls.

The median baseline C-Alb value was significantly higher in the CKD progression cases compared with controls (6.9 versus 5.0; P < 0.001; Table 1). Participants with higher baseline carbamylation were more likely to have CKD progression, with a CKD progression rate of 66, 54 and 30% for each of the three tertiles, respectively (monotonic trend test P < 0.001; Figure 3). In an unadjusted logistic regression model comparing tertiles of carbamylation between CKD progressors and nonprogressors, the odds ratio (OR) of the highest carbamylation tertile (C-Alb >7.33 mmol/mol) having CKD progression compared with the bottom tertile (C-Alb <5.01 mmol/mol) was 8.5 [95% confidence interval (CI) 2.6–27.7; P < 0.001]. Adjusting for baseline differences (smoking status and BUN) between the groups resulted in an OR of 7.9 (95% CI 1.9–32.8; P = 0.004; Table 2). Additional models adjusting for comorbidities and all matched parameters did not alter the direction or significance of these results (Supplementary data, Table S2).

FIGURE 3

Percentage of participants reaching clinical outcome by carbamylation tertile. (A) Percentage of individuals reaching the composite outcome of CKD progression (ESKD or 50% eGFR reduction). (B) Percentage of individuals reaching the outcome of mortality.

Table 2.

Carbamylation level and risk for clinical outcomes

Predictor	Unadjusted model OR (95% CI)	P-value	Multivariablemodel OR (95% CI)	P-value
CKD progression cohort carbamylation tertile
3 versus 1	8.5 (2.6–27.7)	<0.001	7.9 (1.9–32.8)	0.004
2 versus 1	2.9 (1.2–7.0)	0.01	2.7 (1.1–6.8)	0.04
Mortality cohort carbamylation tertile
3 versus 1	2.2 (0.8–5.9)	0.11	3.4 (1.0–11.4)	0.05
2 versus 1	1.4 (0.6–3.4)	0.48	1.6 (0.6–4.0)	0.36

OR = odds ratio demonstrating the effect estimates of the association between tertile of carbamylation and CKD progression and mortality, respectively (95% CI). The multivariable models adjust for the only variables that proved different between groups at baseline—BUN and smoking status.

All-cause mortality

Baseline characteristics of the matched all-cause mortality cases (individuals who died during the study follow-up period, n = 75) and controls (survivors through the study follow-up, n = 75) are also shown in Table 1. There was no statistical difference in age, sex, race, eGFR or 24-h proteinuria. Other variables appeared balanced except for current smoking status, with cases demonstrating a higher percentage of smokers (25.3% versus 10.7%, P = 0.02). The median (IQR) time to reaching the mortality endpoint among the cases was 6.4 years (3.8–8.3) versus 9.7 years of follow-up time in the controls (8.7–10.4).

From the lowest to the highest tertiles of baseline C-Alb levels, the observed all-cause mortality rates were 44, 48 and 58%, respectively (monotonic trend test P = 0.10; Figure 3). In the bivariate analysis comparing tertiles of carbamylation between death cases and surviving controls, the OR of the highest C-Alb tertile (>8.17 mmol) versus lowest (<5.65) was 2.2 (0.8–5.9; P = 0.26). After adjusting for baseline differences between the groups, the highest carbamylation tertile was significantly associated with an increase by more than a factor of three in the risk of all-cause mortality (adjusted OR = 3.4; 95% CI 1.0–11.4; P = 0.05). The directionality and lack of robust statistical significance were again observed with additional adjustments as presented in Supplementary data, Table S2.

Correlates to C-Alb levels

By aggregating all the C-Alb measures obtained for the two case–control populations presented above, we were able to establish correlations between C-Alb measures and clinical data obtained (Figure 4). While protein carbamylation was correlated to eGFR (left panel; r = −0.43; 95% CI −0.56 to −0.31; P < 0.001), BUN levels were a stronger correlate (middle panel; r = 0.76; 95% CI 0.66–0.86; P < 0.001; see Supplementary data, Table S3 for complete data on eGFR and urea distributions in the study populations). An additional observation (right panel) is presented showing no significant correlation between C-Alb and total albumin (C-Alb versus total albumin; r = –0.08; 95% CI −0.17 to 0.02; P = 0.14). Lastly, we saw no correlation between MPO and C-Alb (r = 0.02; 95% CI −0.06 to 0.09; P = 0.70; not shown in Figure 4).

FIGURE 4

Correlation between C-Alb (mmol/mol) and other clinical and laboratory parameters: estimated GFR using the CRIC equation (left panel), serum urea nitrogen (middle panel) and serum albumin (right panel).

DISCUSSION

This report shows for the first time that protein carbamylation burden, as measured by C-Alb, strongly associates with CKD progression, and to a lesser degree, all-cause mortality, in a representative sample population from the CRIC Study. This finding was independent of baseline eGFR, proteinuria, as well as age, sex and race. While carbamylation burden clearly tracks inversely with eGFR, we found that urea was a stronger correlate with carbamylation. Importantly, although C-Alb was correlated with baseline urea concentrations, the risk for CKD progression and death associated with C-Alb was still present after adjusting for urea levels. Finally, because C-Alb is measured as the ratio of carbamylated-to-non-C-Alb, it may be speculated that patients with lower total albumin would have higher C-Alb values, confounding the relationship between C-Alb and outcomes. It was therefore significant to observe no significant correlation between C-Alb and total albumin.

Protein carbamylation can stem from such divergent exposures as kidney disease, smoking, pollution and diet, ultimately leading to common pathways of molecular and cellular dysfunction relevant to clinical outcomes [5, 17]. Dozens of studies have shown results implicating carbamylation in changes in protein charge, conformation and stability, with consequent alterations in enzyme and hormone activity, binding properties and cellular expression and responses [5, 18]. For example, carbamylated proteins at concentrations present in uremia have been shown to activate glomerular mesangial cells to a profibrogenic phenotype and stimulate collagen deposition [7]. Moreover, cyanate exposure strongly inhibits the collagenase activity of purified human and rat mesangial cell matrix metalloproteinase-2 (MMP-2) [19]. Intraperitoneal injection of C-Alb in the amphibian kidney results in tubular cell damage and peritubular fibrosis via stimulation of multiple inflammatory signals [19]. In general, it is suggested that carbamylation of collagen or enzymes involved in extracellular matrix remodeling can disrupt the balanced remodeling of extracellular proteins and enhance fibrosis [8, 20]. Given the high concentrations of urea accumulated in the renal interstitium for its urine concentrating activities, it follows that proteins in the renal parenchyma may be targets for carbamylation, and proteomic studies have indeed shown that many kidney proteins are carbamylated in patients with kidney disease [21, 22]. These studies highlight plausible mechanisms by which hypercarbamylation could lead to common final pathways implicated in CKD progression.

Studies have also suggested that carbamylation can contribute to atherosclerosis and cardiovascular risk via its effects on lipoproteins, collagen, fibrin, proteoglycans and fibronectin [6, 23]. For example, carbamylation of low-density lipoprotein (LDL) may enhance its atherogenic properties partly by decreasing its binding to the LDL-receptor [3] and by preventing its clearance from circulation (a phenomenon seen in both animal models of CKD [24, 25], as well as in uremic patients [26–29]). Moreover, elegant work has shown mechanistically how carbamylation can exacerbate vascular calcification [30]. In well-conducted studies with replication and using different assays, carbamylation has been linked to mortality in patients with preserved kidney function as well as ESKD [2, 3]. Our study showed a clear trend toward an increased mortality risk that was likely weakened by a limited sample size. This will need to be pursued more thoroughly in future studies.

In general, our work challenges the controversial claim that excess urea itself is not pathogenic in CKD [31, 32]. First, urea-driven carbamylation has been associated with a diverse array of adverse processes [26, 33–39]. Furthermore, carbamylation is not simply a urea equivalent. Analogous to measuring glycated hemoglobin A1C to determine long-term glucose control in diabetes mellitus, carbamylated proteins more clearly depict urea load than blood urea measurements, which can fluctuate significantly in relation to GFR, diet, aging, hydration, volume, circulatory status, drugs and catabolic state [40, 41]. Moreover, carbamylation can also occur through nonuremic processes (e.g. MPO catalyzed oxidation of thiocyanate derived from diet and smoking) [3] and can be exacerbated by amino acid deficiencies [2]. This likely explains the independence of our findings from BUN levels despite a correlation between C-Alb and urea.

We did not see baseline differences in MPO across the study groups and we saw no correlation between MPO and C-Alb. Smoking was significantly higher in both case groups (CKD progression and mortality) compared with the control groups, and we therefore adjusted for smoking along with urea in all outcome analyses. Moreover, there were no differences in C-Alb levels between smokers and non-smokers in either study group. Whether smoking cessation or urea reduction could ultimately serve as adequate means of reducing carbamylation, and thus improve clinical outcomes, requires further investigation. Our findings were also independent of comorbidities including diabetes (Supplementary data, Table S2) despite greater numbers of diabetics in the cases versus controls for each study. Interestingly, glycation, as seen in diabetes, may compete with carbamylation for protein modification. Animal studies suggest, however, that at more intensive carbamylation states, competitive glycation appears less significant possibly explaining our findings [42, 43].

This line of investigation is compelling because of growing evidence that carbamylation may be a modifiable factor in CKD. Di Iorio et al. achieved significant carbamylation reductions through dietary interventions employing either a very low-protein diet supplemented with ketoanalogs or a Mediterranean diet, compared with a free diet [44]. It is notable that the ketoanalog-supplemented diet showed the most dramatic urea and carbamylation reductions, but the relative impact of the supplements versus the urea reduction on carbamylation is not known. We acknowledge that interventions such as low-protein diet and ketoanalog supplementation have shown mixed results with clinical outcomes in CKD, but no study to date has targeted carbamylation [45, 46]. Effect sizes might be greater and more consistent if such studies included only subjects with accelerated protein carbamylation.

Our study aligns with a recent report from Tan et al. examining carbamylated lipoproteins and the progression of diabetic kidney disease, but several distinctions warrant mention [47]. Tan et al. noted that carbamylated high-density lipoprotein (HDL), but not carbamylated LDL, associated with CKD progression in a cohort of Type 2 diabetics. The generalizability of the study (Chinese Type 2 diabetics only) was limited in the Tan et al. study, whereas our CRIC-based study was composed of a more diverse patient sampling, enhancing generalizability. Also, the inability to account for lipid-lowering therapy throughout the study and the diverging results for LDL versus HDL generate several questions about the observed protein carbamylation and CKD progression links. In this sense, our approach examining a deeply phenotyped, diverse CKD cohort enhances and amplifies previously noted signals.

Our study has several noteworthy strengths and limitations. While protein carbamylation is garnering increased attention in the biomedical literature, as noted above, studies in nondialysis CKD patients are scant. Leveraging the potent CRIC Study to examine carbamylation and employing rigorous matching criteria across several critically relevant variables enhances the generalizability and validity of our results. For this early investigation, the small sample size and lack of longitudinal measures of carbamylation are limits that can be overcome in future work and there are plans to validate these findings in the broader CRIC Study. We have shown that in ESKD patients, longitudinal assessments of carbamylation can offer important prognostic information for core clinical outcomes such as mortality, thus this could be pursued in future studies of CKD patients [13].

In conclusion, we report here that protein carbamylation as measured by C-Alb is predictive of important clinical outcomes such as CKD progression and this is independent of established predictors such as eGFR and proteinuria. Urea appears a dominant correlate of carbamylation, more so than eGFR, but alone cannot explain the association between carbamylation and clinical outcomes. These unique findings warrant additional study as protein carbamylation may bring us one step closer to fully understanding the biology underlying CKD progression and its complications, and ultimately advance how we treat CKD patients.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.

ACKNOWLEDGEMENTS

The CRIC Study investigators include Lawrence J. Appel, MD, MPH; Harold I. Feldman, MD, MSCE; Alan S. Go, MD; Jiang He, MD, PhD; Robert G. Nelson, MD, PhD, MS; Mahboob Rahman, MD; Panduranga S. Rao, MD; Vallabh O. Shah, PhD, MS; Raymond R. Townsend, MD; Mark L. Unruh, MD, MS.

FUNDING

S.K. is supported by National Institue of Health (NIH) award K23 DK106479 and R01 DK124453. This project was funded by the NIH National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) CKD Biomarkers Consortium Pilot and Feasibility Studies Program (5U01DK103225). A.B. was supported by NIH awards R01 HL133399, R56 HL133399 and K08 HL121801. J.P.L. is supported by K24-DK092290. A.S. is supported by K23DK120811 and core resources from the George M. O’Brien Kidney Research Center at Northwestern University (NU-GoKIDNEY) P30DK114857. This work was initially presented in oral abstract form by S.K. at the American Society of Nephrology Kidney Week 2019. Funding for the CRIC Study was obtained under a cooperative agreement from NIDDK (U01DK060990, U01DK060984, U01DK061022, U01DK061021, U01DK061028, U01DK060980, U01DK060963, U01DK060902 and U24DK060990). In addition, this work was supported in part by: the Perelman School of Medicine at the University of Pennsylvania Clinical and Translational Science Award NIH/NCATS UL1TR000003, Johns Hopkins University UL1 TR-000424, University of Maryland GCRC M01 RR-16500, Clinical and Translational Science Collaborative of Cleveland, UL1TR000439 from the National Center for Advancing Translational Sciences (NCATS) component of the National Institutes of Health and NIH roadmap for Medical Research, Michigan Institute for Clinical and Health Research (MICHR) UL1TR000433, University of Illinois at Chicago CTSA UL1RR029879, Tulane COBRE for Clinical and Translational Research in Cardiometabolic Diseases P20 GM109036, Kaiser Permanente NIH/NCRR UCSF-CTSI UL1 RR-024131 and Department of Internal Medicine, University of New Mexico School of Medicine Albuquerque, NM R01DK119199.

AUTHORS’ CONTRIBUTIONS

S.K. and J.P.L. with input from all authors designed the study; A.S., D.R., R.D., A.F., J.C., J.S., T.S., M.W. and J.P.L. contributed to the parent CRIC Study; A.B. and S.A.K. carried out the laboratory assay; R.T., A.S.A., S.N. and S.Z. contributed to the statistical analysis; S.K. drafted the manuscript; all authors approved the final version of the manuscript.

CONFLICT OF INTEREST STATEMENT

The results presented in this article have not been published previously in whole or part, except in abstract format. The authors report nothing additional to disclose beyond the above acknowledgements.

The CRIC Study investigators are listed in the Acknowledgements section.

Author notes

Sahir Kalim and Anders H Berg Co-first authors.