Abstract
Rationale & Objective:
HbA1c is widely used to estimate glycemia, yet it is less reliable in patients with chronic kidney disease (CKD). There is growing interest in the complementary use of glycated albumin (GA) to improve glycemic monitoring and risk stratification. However, whether GA associates with clinical outcomes in a non-dialysis dependent CKD population remains unknown.
Study Design:
Prospective cohort study.
Setting & Participants:
3110 participants with CKD from the Chronic Renal Insufficiency Cohort study.
Exposure:
Baseline GA levels.
Outcomes:
Incident end-stage kidney disease (ESKD), cardiovascular disease (CVD) events, and all-cause mortality.
Analytical Approach:
Cox proportional hazards regression.
Results:
Participant characteristics included mean age 59.0 (SD 10.8) years; 1357 (43.6%) female; 1550 (49.8%) with diabetes. The median GA was 18.7 (interquartile range, 15.8–23.3)%. During an average 7.9-year follow-up, there were 980 ESKD events, 968 CVD events, and 1084 deaths. Higher GA levels were associated with greater risks of all outcomes, regardless of diabetes status: hazard ratios for ESKD, CVD, and death among participants with the highest quartile compared with quartile 2 (reference) were 1.42 (95%CI, 1.19–1.69), 1.67 (CI, 1.39–2.01), and 1.63 (CI, 1.37–1.94), respectively. The associations with CVD and death appeared J-shaped, with increased risk also seen at the lowest GA levels. Among patients with coexisting CKD and diabetes, the associations of GA with outcomes remained significant even after adjusting for HbA1c. For each outcome, we observed a significant increase in the fraction of new prognostic information when both GA and HbA1c were added to models.
Limitations:
Lack of longitudinal GA measurements; HbA1c measurements were largely unavailable in participants without diabetes.
Conclusions:
Among patients with CKD, GA levels were independently associated with risks of ESKD, CVD, and mortality, regardless of diabetes status. GA added prognostic value to HbA1c among patients with coexisting CKD and diabetes.
Keywords: glycated albumin, chronic kidney disease, diabetes mellitus, clinical epidemiology
Plain Language Summary
HbA1c is widely used to estimate glycemia, yet it is less reliable in patients with CKD. There is growing interest in the complementary use of glycated albumin (GA) to improve glycemic monitoring and risk stratification. However, whether GA associates with clinical outcomes in a non-dialysis dependent CKD population remains unknown. In this cohort study of 3,110 individuals with non-dialysis dependent CKD, GA levels were independently associated with risks of ESKD, CVD, and mortality. The associations with CVD and mortality appeared to be J-shaped. Among patients with coexisting CKD and diabetes, GA added prognostic value to HbA1c. Thus, GA may be a valuable complementary test to HbA1c in patients with CKD.
INTRODUCTION
Chronic kidney disease (CKD) and diabetes are two major global health problems that commonly coexist and lead to significant morbidity and mortality.1 Diabetes is the leading cause of CKD and progression to end-stage kidney disease (ESKD), and glycemia is one of the most important modifiable risk factors for cardiovascular disease (CVD) and death among patients with CKD.1–3 Glycated hemoglobin (HbA1c), reflecting mean glycemia over a 2–3 month period, is the most widely utilized clinical method of glycemic monitoring,4 and is employed to diagnose diabetes as well as evaluate the efficacy of glycemic treatment.4 Seminal studies showed that increased levels of HbA1c were associate with higher risks of adverse clinical outcomes including CVD and death, even in those without diabetes.5 However, its reliability is known to be compromised in patients with CKD, due to CKD-related factors such as anemia and its treatment (e.g., iron and erythropoiesis-stimulating agents) leading to changes in erythrocyte lifespans and altered glycation accumulation.6–8 To date, the optimal glycemic monitoring strategy in patients with CKD remains a matter of debate.6
Glycated albumin (GA), a non-traditional glycemic marker reflecting mean glycemia across a shorter timeframe (2–3 weeks), has been proposed as a complement to HbA1c for glycemic monitoring, especially in situations where HbA1c has limitations.9–13 Less impacted by erythrocyte lifespan and hemoglobin alteration, GA may overcome anemia-related limitations of HbA1c.14 Although GA has been found to be comparable or even superior to HbA1c in reliably measuring time averaged glycemia in patients with CKD,15,16 its prognostic value in this population is largely unknown. Studies have suggested strong associations between GA levels and microvascular complications, cardiovascular outcomes, and mortality in the general population,17–21 patients with type 1 diabetes,22 and patients on dialysis,16,23–27 yet it remains unclear whether GA is associated with clinical outcomes in the largest population where HbA1c is known to be fraught: non-dialysis CKD patients. Such a knowledge gap is critical to address when considering the adoption of GA testing into clinical practice in this large patient population.11
Therefore, in this multicenter US cohort study of CKD patients, we investigated the prognostic value of GA. Our primary objective was to characterize the associations between GA levels and clinical outcomes in patients with CKD, both overall and in subgroups of patients with and without an established diagnosis of diabetes. Our secondary objective was to evaluate whether GA added prognostic values to HbA1c for risk stratification in patients with coexisting CKD and diabetes.
METHODS
Study population
The Chronic Renal Insufficiency Cohort (CRIC) is a multicenter prospective observational cohort of patients with mild to severe CKD, followed longitudinally to study risk factors for ESKD, CVD, and mortality. A total of 3939 patients aged 21 to 74 with estimated glomerular filtration rate (eGFR) of 20 to 70 mL/min/1.73m2 were enrolled across 7 US clinical centers from June 2003 to September 2008. The main exclusion criteria included pregnancy, New York Heart Association Class III–IV heart failure, HIV, cirrhosis, multiple myeloma, renal cancer, recent chemotherapy or immunosuppression, polycystic kidney disease, organ transplantation, or previously receiving dialysis for over one month. Further details of the CRIC study have been previously published.28,29 The research was conducted in accordance with the ethical principles of the Declaration of Helsinki. All participants provided written informed consent at enrollment. The CRIC study protocol was approved by the institutional review boards from each participating clinical center.
The year 1 (Y1) CRIC visit (1 year after initial enrollment) was the visit for which stored serum samples were available for measurement of GA. Therefore, this was considered the “baseline” visit of this study and is referred to as such in this report. A total of 3110 participants with CKD has samples available for the primary analysis. Of those, 1550 (49.8%) participants had a coexisting diagnosis of diabetes, which was determined by the following criteria: fasting glucose level >= 126 mg/dL, random glucose level >= 200 mg/dL, or self-reported use of insulin or other glucose-lowering medications. Among those with diabetes, a total of 1516 participants had HbA1c measurement for secondary analysis. A flowchart illustrating the selection process is shown in Figure S1.
Measurements of glycemic markers
GA was measured using high-performance liquid chromatography and tandem mass spectrometry (LC-MS/MS) after a single thaw of frozen serum samples that were collected from the Y1 visit and stored at −80 °C. Analogous to HbA1c measures (reported as the percent of total hemoglobin that is glycated), GA is expressed as the percentage of total serum albumin that is glycated. The GA assay used in our study was calibrated using reference materials obtained from the commercial GA assay (Lucica GA-L, Asahi Kasei Pharma, Tokyo, Japan). A complete description of the mass spectrometric GA assay and its analytical validation has been previously described.25 Two quality control samples with GA mean values of 10.9% and 18.2% were assayed with each 96-well plate of samples; the interassay coefficients of variance were 6.1% and 7.6%, respectively. HbA1c was measured at the CRIC Central Lab at the University of Pennsylvania using high-performance liquid chromatography with assay calibration using reference materials obtained from the National Glycohemoglobin Standardization Program.30
Outcomes ascertainment
Co-primary outcomes, chosen a priori, include 1) incident ESKD (requiring chronic dialysis or kidney transplant); 2) CVD events (a composite of myocardial infarction, congestive heart failure or stroke); and 3) all-cause mortality. The participants were followed every 6 months with telephone visit and annually with in-person clinic visits. ESKD status was obtained through semiannual surveillance with ascertainment supplemented with cross-linkage with the US Renal Data System. All cardiovascular outcomes were adjudicated by an independent clinical events committee. All deaths were verified by reviewing death certificate. Further details of the outcome ascertainment have been previously published.31 The observation period was continued until withdrawal of consent, loss to follow-up, end of follow-up (2017), or the occurrence of death.
Covariates
All the demographics, lifestyle behaviors, medical history, current medications, anthropometrics and blood pressure measurement were collected at baseline, except for sex and race which were self-reported at the study enrollment. All laboratory data were obtained using standardized assays performed on samples from the same baseline visit. The eGFR was calculated according to the CRIC-derived equation.32 Proteinuria was measured from 24-hour urine collection. All covariates had <1% missing values, except for 24-hour proteinuria (8%), LDL (3%) and triglyceride (3%). Missing covariates were imputed by the mean or median of the existing values as appropriate.
Statistical Analysis
The baseline characteristics were summarized as mean (SD) or median (interquartile range) for continuous variables and counts (percentages) for categorical variables in the overall population and then compared across quartiles of GA levels and based on diabetes status. Relationship between GA and serum albumin levels and HbA1c were compared using Spearman correlation coefficients and scatterplot.
Cox proportional hazards models were used to evaluate the associations of GA with risks of each outcome. The proportional hazard assumptions were verified and found tenable using Schoenfeld residual plots. The linearity of continuous GA was assessed by likelihood ratio tests. Given the evidence for nonlinearity, restricted cubic splines were plotted to characterize the shape of GA’s association with outcomes, with 4 knots placed at the 5th, 35th, 65th, and 95th percentiles. Additional analysis was performed by categorizing GA into quartiles, with Quartile 2 (Q2, 15.8–18.6%) as the reference group considering the GA cut points recommended in the general population20 and observed risks associated with values above or below this range based on the restricted cubic spline plots. Interaction with diabetes status and subgroup analysis stratified by diabetes were performed. Potential confounding variables included in adjusted models were selected a priori based on the literature review and clinical relevance. The two adjusted models were: 1) model 1 including age, sex, and self-reported race/ethnicity; 2) model 2 including all model 1 covariates plus body mass index (BMI), current smoking status, history of CVD, use of angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers, systolic blood pressure, eGFR, log-transformed 24-hour proteinuria, low density lipoprotein (LDL), and log-transformed triglyceride levels.
To study whether GA added prognostic value to HbA1c, the two glycemic markers were compared as risk factors for clinical outcomes among patients with diabetes. Given that HbA1c measurements were largely unavailable for patients without diabetes, this comparison was confined to a subset of 1516 patients with diabetes and HbA1c measurement available (Figure S1). In all the following analyses, we used quartiles of GA and HbA1c calculated in this subset of patients. Cox proportional hazards models were used to evaluate the association between GA or HbA1c and risks of each outcome, adjusting for covariates in model 2 (base model). We also added both glycemic markers to covariates in model 2 (resulting in model 3). We used the likelihood ratio test to assess whether the inclusion of glycemic markers (GA, HbA1c, or both GA and HbA1c) added new information about the risks of outcomes beyond the covariates in model 2. We quantified the added prognostic values of the glycemic markers as the fraction of new information, a metric estimating the additional proportion of explainable variation that is explained by inclusion of the glycemic markers.33,34 Variance in the predicted risk was calculated for the base model and for each of the three models expanded to include the glycemic markers (GA, HbA1c, and both GA and HbA1c). The Relative Explained Variation (REV) was determined as the ratio of the variance in predicted risk for the base model to that for the respective expanded model. The formula for the fraction of new information for the glycemic markers was (1-REV)*100%.
All analyses were conducted using R Version 4.3.0. All reported P values were two-sided, and a cut off 0.05 was used to indicate statistical significance.
Sensitivity Analysis
Several sensitivity analyses were conducted for the main findings. First, we performed multiple imputations for missing covariates. Second, we further added total serum albumin level as a covariate in the models. Lastly, we also performed Fine and Gray competing risks regression to account for death as competing events for ESKD and CVD outcomes.
RESULTS
Baseline characteristics of the study participants (n=3110), both overall and according to the quartiles of GA, are presented in Table 1. Participant demographics included age 59.0 (10.8) years; 1357 (43.6%) female; 1334 (42.9%) white. The mean baseline eGFR was 41.9 (16.4) mL/min/1.73m2, HbA1c was 7.4 (1.6)%, and the median GA value was 18.7 (15.8–23.3)%. As expected, those with higher GA levels were more likely to have a diagnosis of diabetes. Baseline characteristics according to diabetes status are presented in Table S1.
Table 1.
Baseline Characteristics According to Glycated Albumin Quartiles Among Study Participants
| Characteristic | Total (N=3,110) | Q1 (N=778) | Q2 (N=777) | Q3 (N=777) | Q4 (N=778) |
|---|---|---|---|---|---|
| GA range, % | 1.4–70.8 | 1.4–15.7 | 15.8–18.6 | 18.7–23.3 | 23.4–70.8 |
| Age, years | 59.0 (10.8) | 57.3 (11.7) | 58.1 (10.9) | 60.3 (10.5) | 60.1 (9.6) |
| Female, No. (%) | 1357 (43.6) | 376 (48.3) | 348 (44.8) | 329 (42.3) | 304 (39.1) |
| Race, No. (%) | |||||
| White | 1334 (42.9) | 395 (50.8) | 363 (46.7) | 327 (42.1) | 249 (32.0) |
| Black | 1321 (42.5) | 278 (35.7) | 308 (39.6) | 355 (45.7) | 380 (48.8) |
| Hispanic | 337 (10.8) | 78 (10.0) | 75 (9.7) | 67 (8.6) | 117 (15.0) |
| Other | 118 (3.8) | 27 (3.5) | 31 (4.0) | 28 (3.6) | 32 (4.1) |
| Medical History, No (%) | |||||
| Diabetes | 1550 (49.8) | 211 (27.1) | 238 (30.6) | 396 (51.0) | 705 (90.6) |
| Hypertension | 2802 (90.2) | 692 (89.2) | 678 (87.4) | 699 (90.0) | 733 (94.3) |
| CVD | 1132 (36.4) | 237 (30.5) | 250 (32.2) | 287 (36.9) | 358 (46.0) |
| CHF | 326 (10.5) | 80 (10.3) | 53 (6.8) | 79 (10.2) | 114 (14.7) |
| Stroke | 342 (11.0) | 69 (8.9) | 73 (9.4) | 103 (13.3) | 97 (12.5) |
| Current smoking, No. (%) | 389 (12.5) | 105 (13.5) | 102 (13.1) | 104 (13.4) | 78 (10.0) |
| BMI, kg/m2 | 32.2 (7.6) | 32.9 (7.8) | 31.9 (7.5) | 31.7 (7.8) | 32.4 (7.4) |
| SBP, mmHg | 127.2 (21.4) | 126.4 (21.6) | 125.2 (20.3) | 126.9 (21.4) | 130.3 (22.0) |
| Medication Use, No (%) | |||||
| ACEI/ARB | 2192 (70.8) | 525 (67.6) | 522 (67.4) | 542 (70.1) | 603 (78.1) |
| Oral hypoglycemics | 861 (27.8) | 138 (17.8) | 147 (19.0) | 239 (30.9) | 337 (43.7) |
| Insulin | 803 (25.9) | 75 (9.7) | 82 (10.6) | 196 (25.4) | 450 (58.3) |
| Antiplatelet | 1573 (50.8) | 336 (43.2) | 363 (46.9) | 394 (51.0) | 480 (62.2) |
| Statins | 1847 (59.6) | 419 (53.9) | 407 (52.6) | 484 (62.6) | 537 (69.6) |
| Laboratory Data | |||||
| Creatinine, mg/dL | 2.0 (0.9) | 1.9 (0.9) | 1.9 (0.9) | 2.0 (0.9) | 2.2 (1.0) |
| eGFR(CRIC), mL/min/1.73m2 | 41.9 (16.4) | 44.5 (17.1) | 44.3 (16.5) | 41.1 (16.1) | 37.5 (15.0) |
| CKD stage | |||||
| G2 | 461 (14.8) | 148 (19.0) | 153 (19.7) | 98 (12.6) | 62 (8.0) |
| G3 | 1819 (58.5) | 465 (59.8) | 450 (57.9) | 476 (61.3) | 428 (55.0) |
| G4–5 | 817 (26.3) | 161 (20.7) | 171 (22.0) | 201 (25.9) | 284 (36.5) |
| Cystatin C, mg/L | 1.7 (0.7) | 1.6 (0.7) | 1.6 (0.6) | 1.7 (0.6) | 1.8 (0.7) |
| BUN, mg/dL | 31.4 (16.1) | 28.0 (13.6) | 28.5 (14.1) | 31.8 (16.2) | 37.5 (18.1) |
| Hemoglobin, g/dL | 12.8 (1.8) | 13.3 (1.9) | 13.1 (1.8) | 12.7 (1.7) | 12.1 (1.7) |
| HbA1c, % | 7.4 (1.6) | 6.2 (0.9) | 6.5 (0.9) | 7.0 (1.1) | 8.3 (1.7) |
| Random glucose, mg/dL | 116.0 (50.3) | 97.4 (20.8) | 101.1 (27.8) | 112.3 (41.1) | 153.7 (72.4) |
| Serum albumin, g/dL | 4.0 (0.4) | 4.0 (0.5) | 4.1 (0.4) | 4.1 (0.4) | 4.0 (0.4) |
| LDL, mg/dL | 99.3 (34.8) | 104.3 (35.7) | 102.0 (32.8) | 98.1 (34.4) | 92.8 (35.3) |
| TG, mg/dL | 126 (89, 182) | 137 (96, 204) | 126 (90, 178) | 119 (89, 168) | 126 (85, 184) |
| Urine protein, g/24 hours | 0.2 (0.1, 0.9) | 0.2 (0.1, 1.0) | 0.1 (0, 0.6) | 0.2 (0.1, 0.7) | 0.3 (0.1, 1.2) |
Categorical variables are presented as counts (percentages). Percentages may not total 100 because of rounding. Continuous variables are presented as range, mean (SD), or median (interquartile range).
Abbreviations: HbA1c, glycated hemoglobin; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BMI, body mass index; BUN, blood urea nitrogen; CHF, congestive heart failure; CKD, chronic kidney disease; CRIC, chronic renal insufficiency cohort; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; GA, glycated albumin; LDL, low-density lipoprotein; No, number; SBP, systolic blood pressure; TG, triglyceride.
Among 3110 participants with CKD, serum albumin was not correlated with GA (r= −0.03, p=0.06). Among 1516 patients with coexisting CKD and diabetes, the overall correlation between GA and HbA1c was significant (r=0.58, p<0.001) but gradually decreased with more advanced stages of CKD (Figure S2): r=0.75 in CKD stage 2, r=0.61 in stage 3, and r=0.50 in stages 4–5.
During an average of 7.9 (4.1) years of follow-up, 980 (31.5%) individuals developed ESKD, 968 (31.1%) had CVD events, and 1084 (34.9%) died. The association between GA levels and outcomes appeared J-shaped, especially for CVD and death (Figure 1). There was an increased risk of all 3 outcomes among CKD patients with higher levels of GA (Table 2). For example, compared to reference group (Q2), individuals in the highest GA quartile demonstrated a 1.42-fold (95% CI, 1.19–1.69) increased risk of developing ESKD, a 1.67-fold (95% CI, 1.39–2.01) increased risk of CVD, and a 1.63-fold (95% CI, 1.37–1.94) increased risk of death in the fully adjusted model 2. Moreover, there was also an increased risk of CVD and death in those with low levels of GA: individuals in the lowest GA quartile showed a 1.25-fold (95% CI, 1.03–1.53) increased risk of CVD and a 1.31-fold (95% CI, 1.08–1.58) increased risk of death. We performed multiple imputation for missing covariates and no substantial differences were noted (Table S2). Adding total serum albumin as a covariate did not substantially change the results (Table S3). Results appeared similar when death was treated as a competing risk for ESKD or CVD outcomes (Table S4).
Figure 1. Adjusted Hazards Ratios (HRs) and 95% Confidence Intervals (CIs) for GA Levels and Risks of ESKD, CVD and Death Among Patients with CKD in the CRIC Study.
Adjusted HRs are from Cox proportional hazards models with adjustment for age, sex, race/ethnicity, BMI, current smoking status, history of CVD, use of angiotensin-converting enzyme inhibitors (ACEIs) or angiotensin II receptor blockers (ARBs), systolic blood pressure, eGFR, 24-hour proteinuria (with natural log transformation), LDL, and triglyceride (with natural log transformation). GA was modelled using restricted cubic splines (solid blue lines) with knots at the 5th, 35th, 65th, and 95th percentiles. All models are centered at the 50th percentile (GA=18.7%). The shaded areas are the CIs for the restricted cubic spline models. Bars are the frequency histograms showing the distribution of GA among all study participants, with light blue bars representing those with diabetes and red bars representing those without diabetes. Dark blue bars indicate that participants with and without diabetes have overlapping GA values. Abbreviations: DM, diabetes mellitus; GA, glycated albumin.
Table 2.
Risks of ESKD, CVD, and Death According to Glycated Albumin Quartilesa Among 3110 Study Participants with CKD
| Outcomes | Rate Per 1000 Patient-Year | Model 1b | P | Model 2c | P |
|---|---|---|---|---|---|
| ESKD | |||||
| Q1 | 26.94 | 1.03 (0.85, 1.26) | 0.7 | 0.97 (0.79, 1.18) | 0.8 |
| Q2 | 26.93 | 1.00 (reference) | NA | 1.00 (reference) | NA |
| Q3 | 33.80 | 1.26 (1.04, 1.52) | 0.02 | 0.93 (0.77, 1.13) | 0.5 |
| Q4 | 56.92 | 2.03 (1.70, 2.42) | <0.001 | 1.42 (1.19, 1.69) | <0.001 |
| CVD | |||||
| Q1 | 29.61 | 1.22 (1.00, 1.49) | 0.05 | 1.25 (1.03, 1.53) | 0.03 |
| Q2 | 25.47 | 1.00 (reference) | NA | 1.00 (reference) | NA |
| Q3 | 34.81 | 1.25 (1.03, 1.52) | 0.02 | 1.18 (0.97, 1.43) | 0.09 |
| Q4 | 58.88 | 2.09 (1.74, 2.50) | <0.001 | 1.67 (1.39, 2.01) | <0.001 |
| Death | |||||
| Q1 | 27.39 | 1.23 (1.02, 1.49) | 0.03 | 1.31 (1.08, 1.58) | 0.006 |
| Q2 | 23.99 | 1.00 (reference) | NA | 1.00 (reference) | NA |
| Q3 | 33.71 | 1.26 (1.05, 1.51) | 0.01 | 1.22 (1.02, 1.47) | 0.03 |
| Q4 | 49.82 | 1.91 (1.61, 2.27) | <0.001 | 1.63 (1.37, 1.94) | <0.001 |
Glycated Albumin (GA) quartiles range in the 3110 CRIC study participants with CKD: Q1: 1.4–15.7%; Q2: 15.8–18.6%; Q3: 18.7–23.3%; Q4: 23.4–70.8%.
Model 1: adjusted for age, sex, race/ethnicity.
Model 2: adjusted for age, sex, race/ethnicity, BMI, current smoking status (yes or no), history of cardiovascular disease (yes or no), systolic blood pressure, use of ACEIs or ARBs medications (yes or no), eGFR, 24-hour proteinuria (with log transformation), LDL, triglyceride (with log transformation).
There was no evidence of interaction between GA and diabetes status on the risks of any clinical outcomes (all P values for interaction >0.05). When we modeled the associations between GA and outcomes separately for participants with (Figure S3) and without a diagnosis of diabetes (Figure S4 and Table S5), we observed largely similar results as seen in the overall population.
Among 1516 patients with coexisting CKD and diabetes, the association between the highest quartile of GA and clinical outcomes were of similar magnitude to those for HbA1c, but GA demonstrated more robust associations to outcomes than HbA1c in the lower quartiles (Table 3). GA’s association with outcomes was attenuated but remained statistically significant when including HbA1c in the models. For each outcome, we observed the highest fraction of new prognostic information when both GA and HbA1c were added to the base model 2 (Table 4).
Table 3.
Risks of ESKD, CVD, and Death According to Quartiles of Glycated Albumina vs HbA1cb Among 1516 Study Participants with Coexisting CKD and Diabetes
| GA | HbA1c | |||||
|---|---|---|---|---|---|---|
| Rate Per 1000 Patient-Year | Model 2c | Model 3d | Rate Per 1000 Patient-Year | Model 2 | Model 3 | |
| ESKD | ||||||
| Q1 | 46.90 | 1.38 (1.08, 1.76) | 1.39 (1.08, 1.78) | 42.87 | 1.10 (0.87, 1.39) | 1.09 (0.86, 1.39) |
| Q2 | 43.49 | 1.00 (reference) | 1.00 (reference) | 45.99 | 1.00 (reference) | 1.00 (reference) |
| Q3 | 55.13 | 1.41 (1.12, 1.78) | 1.40 (1.11, 1.76) | 56.36 | 1.00 (0.79, 1.26) | 0.97 (0.77, 1.22) |
| Q4 | 61.32 | 1.59 (1.26, 2.02) | 1.49 (1.15, 1.92) | 62.50 | 1.34 (1.06, 1.68) | 1.22 (0.95, 1.58) |
| CVD | ||||||
| Q1 | 54.36 | 1.51 (1.19, 1.91) | 1.54 (1.21, 1.96) | 47.85 | 1.01 (0.81, 1.26) | 1.01 (0.81, 1.28) |
| Q2 | 41.16 | 1.00 (reference) | 1.00 (reference) | 51.10 | 1.00 (reference) | 1.00 (reference) |
| Q3 | 58.12 | 1.47 (1.16, 1.85) | 1.43 (1.13, 1.81) | 50.87 | 0.97 (0.77, 1.22) | 0.95 (0.75, 1.20) |
| Q4 | 64.40 | 1.70 (1.35, 2.15) | 1.52 (1.17, 1.98) | 68.26 | 1.37 (1.10, 1.70) | 1.27 (0.99, 1.64) |
| Death | ||||||
| Q1 | 40.28 | 1.16 (0.93, 1.46) | 1.22 (0.96, 1.54) | 40.47 | 0.89 (0.72, 1.11) | 0.91 (0.73, 1.13) |
| Q2 | 28.85 | 1.00 (reference) | 1.00 (reference) | 43.65 | 1.00 (reference) | 1.00 (reference) |
| Q3 | 48.91 | 1.35 (1.09, 1.68) | 1.31 (1.05, 1.63) | 40.24 | 0.89 (0.71, 1.12) | 0.87 (0.69, 1.09) |
| Q4 | 52.07 | 1.48 (1.18, 1.85) | 1.28 (1.01, 1.64) | 55.67 | 1.40 (1.13, 1.72) | 1.32 (1.04, 1.67) |
Glycated albumin (GA) quartiles ranges among patients with coexisting CKD and diabetes: Q1: 3.4–18.0%; Q2: 18.1–22.5%; Q3: 22.6–28.6%; Q4: 28.7–70.8%. Note that GA quartiles among patients with coexisting CKD and diabetes is different from the overall population.
HbA1c quartiles ranges among patients with coexisting CKD and diabetes: Q1: 4.6–6.3%; Q2: 6.4–7.2%; Q3: 7.3–8.2%; Q4: 8.3–14.4%.
Model 2: adjusted for age, sex, race/ethnicity, BMI, current smoking status (yes or no), history of cardiovascular disease (yes or no), systolic blood pressure, use of ACEIs or ARBs medications (yes or no), eGFR, 24-hour proteinuria (with log transformation), LDL, triglyceride (with log transformation).
Model 3: all covariates in Model 2 plus both glycemic markers (GA and HbA1c).
Table 4.
Fraction of New Prognostic Information from Including Glycated Albumin and/or HbA1c to Model 2a Among 1516 Study Participants with Coexisting CKD and Diabetes
| Outcomes | Model 2 +GA (vs Model 2) | Model 2 +HbA1c (vs Model 2) | Model 2 +GA+HbA1cb (vs Model 2) | |||
|---|---|---|---|---|---|---|
| Fraction of new information (%) | P value (likelihood ratio test) | Fraction of new information (%) | P value (likelihood ratio test) | Fraction of new information (%) | P value (likelihood ratio test) | |
| ESKD | 1.5 | <0.001 | 0.8 | 0.05 | 2.3 | 0.003 |
| CVD | 8.0 | <0.001 | 4.0 | 0.006 | 8.0 | <0.001 |
| Death | 4.2 | 0.003 | 6.3 | <0.001 | 8.3 | <0.001 |
Model 2: age, sex, race/ethnicity, BMI, current smoking status (yes or no), history of cardiovascular disease (yes or no), systolic blood pressure, use of ACEIs or ARBs medications (yes or no), eGFR, 24-hour proteinuria (with log transformation), LDL, triglyceride (with log transformation).
Model 2 +GA+HbA1c: this is the same as Model 3.
DISCUSSION
In a large representative US cohort of 3110 patients with non-dialysis dependent CKD, we demonstrated that GA levels were independently associated with the risks of ESKD, CVD and mortality, with significant results across subgroups with and without diabetes. The association of GA with CVD and mortality appeared to be J shaped—patients with either low or high GA levels had increased risks. Among patients with coexisting CKD and diabetes, the associations between GA and clinical outcomes persisted even after adjustment for HbA1c. Furthermore, GA added prognostic value to HbA1c, and modelling both glycemic markers together provided the greatest prognostic information. Taken together, these findings suggest that GA could potentially serve as a complementary test to HbA1c among patients with CKD.
HbA1c has been widely used for glycemic monitoring in the general population,35 yet the optimal approach in patients with CKD remains controversial.6 We and others have shown that CKD-related factors, such as anemia, can alter HbA1c in a glycemia-independent manner, therefore limiting HbA1c’s diagnostic and prognostic value among patients with CKD. Recent Kidney Disease Improving Global Outcomes (KDIGO) guidelines highlighted validation of alternative glycemic markers in CKD as a critical knowledge gap.6,30,36 GA, a promising hemoglobin-independent glycemic marker,11,12,14,15 has been widely used in clinical practice in Japan and was recently approved by the Food and Drug Administration for clinical use in the US.14,20,37 Accumulating evidence supports GA’s reliability in glycemic measurement in CKD, with a meta-analysis of 24 studies involving 3928 patients with advanced CKD demonstrating GA’s superiority over HbA1c in accurately assessing glycemia.15
However, the acceptance of a new glycemic marker in clinical practice is also dependent on establishing its prognostic value.18,38 For example, the epidemiological link between HbA1c and microvascular outcomes was central in its adoption into clinical guidelines and practice.35,39 Therefore, beyond glycemic monitoring, whether GA is associated with the micro- and macro-vascular complications of hyperglycemia is critical to establish.9,38,40 Until now, there was a paucity of data on the performance of GA as a prognostic marker in CKD. Previous studies conducted in the general population with largely preserved kidney function have linked GA to long-term outcomes such as incident diabetes, microvascular complications, cardiovascular outcomes, and mortality.17–21 GA was shown to be associated with prevalent CKD and diabetic nephropathy in two cross-sectional studies.41,42 Furthermore, a few studies in the dialysis population showed GA levels were associated with mortality and other important clinical outcomes,16,23–27 and suggested GA might be a superior risk marker than HbA1c in the dialysis setting.23,25 In the present study, we have addressed an important knowledge gap by establishing the prognostic value of GA among a CKD population, where the status quo HbA1c is known to be less reliable and alternatives are clearly needed.6,12 While GA demonstrated more robust associations to outcomes than HbA1c in patients with diabetes, GA’s associations with outcomes were also significant in those without diabetes, suggesting that GA has important prognostic value in CKD beyond glycemic monitoring alone.
In addition to providing a read out of time averaged glycemia, the strong association of GA with adverse clinical outcomes has additional pathophysiological underpinning. Previous mechanistic studies demonstrated the deleterious effect of GA as an early Amadori-type glycation protein and a precursor to harmful advanced glycation end-products.10,43 An extensive body of literature has documented the role GA plays in enhancing oxidative stress via generation of reactive oxygen species,44,45 activating various inflammatory signaling pathways (such as interleukin 6 and TGF-ß),46,47 inducing interstitial fibrosis,48 and modifying vascular clearance.49 This evidence suggests that beyond being a prognostic glycemic marker, GA may have a direct toxic or pathogenic effector role. While our study is observational and cannot establish causation, it contributes to the growing understanding that modified albumin, such as GA, has significant implications in CKD. Future trials are needed to evaluate if outcomes are improved when targeting a specific GA level in patients with CKD.
Our observation of a J-shaped association between GA and CVD as well as mortality is intriguing. This is consistent with a prior report from the Atherosclerosis Risk in Communities (ARIC) Study demonstrating a similar J-shaped association of GA with the risks of heart failure and mortality in the general population.18 Similarly, J-shaped associations of HbA1c with risk of mortality have previously been reported among different populations.5,50,51 There are several potential shared mechanisms. One might speculate that low GA or HbA1c is a generalized marker of ill health or a surrogate of malnutrition. There have been previous reports suggesting hypoalbuminemia could lead to GA changes independent of glycemia.52,53 However, we did not find any correlation between GA and serum albumin levels in our study. Furthermore, adding serum albumin as a covariate did not change our results, suggesting that GA’s association with outcomes was independent of total serum albumin levels. Another possibility is that those patients with CKD and low GA (or HbA1c) levels might have suffered from the deleterious impact of hypoglycemic events. Additional work is warranted to further explore the causes and implications of very low levels of glycemic markers.
All glycemic monitoring strategies have their own limitations. Self-monitoring of blood glucose with fingerstick is cumbersome. Continuous glucose monitoring (CGM) is an appealing option, but its cost has limited widespread adoption of this novel technology.54 There is also concern of variability within and across sensors, a lack of standardization of interstitial glucose measurement methods, and nonglycemic factors influencing sensor readings (e.g., sensor placement, blood flow).55,56 An additional challenge in CKD is the limited CGM studies to guide clinical targets or evaluate outcomes in this population.57 To date, HbA1c remains the standard of care with advantages including assay standardization and an extensive body of evidence supporting its use,35 but a major limitation is its suboptimal performance in CKD. GA may overcome at least anemia-related limitations of HbA1c use in CKD and integrate more intermediate glycemic information. The commercially available GA assay is rapid, technically easy and relatively inexpensive.14,58 However, there is concern that GA assays could be biased by hypoalbuminemia, a common condition among CKD patients, particularly with proteinuria or malnutrition.11 Some experts advocate for complementary use of HbA1c and GA,39 analogous to concomitant creatinine and cystatin C measurements for eGFR equations.59 In our study, we demonstrated that GA added prognostic values to HbA1c and enhanced risk stratification in CKD, where neither glycemic marker would be perfect to use alone. Previous studies have suggested using a glycation gap (calculated as the difference assessed by both HbA1c and glycated serum protein)60 or the GA/HbA1c ratio61 to reconcile information from different glycemic markers. However, how to best utilize both glycemic markers among patients with CKD will require additional studies.
Our results should be considered in the context of several limitations. First, time variations of GA levels were not accounted for as GA was only measured at baseline. Future studies incorporating longitudinal assessments of GA levels may better characterize risks. Second, HbA1c measurements were largely unavailable in participants without diabetes, therefore limiting our ability to compare GA with HbA1c to only those with diabetes. Despite the reduced power in the secondary analysis, we found that the direction and magnitude of the associations of GA and clinical outcomes were similar in subgroups and the total cohort. Third, we did not have information on diabetes type or duration. Lastly, CRIC study was primarily conducted prior to the wide adoption of sodium-glucose cotransporter-2 (SGLT2) inhibitors in CKD patients. Future investigation of the interplay between SGLT2 inhibitors, GA levels, and adverse clinical outcomes in CKD could provide valuable insights. Nevertheless, our findings were strengthened by long-term prospective follow-up with high retention rate and rigorous outcome ascertainment. Furthermore, the CRIC study contains extensive clinical data, which allowed us to use multivariable regression analysis to minimize the risk of residual confounding. Our study’s generalizability was enhanced by a large, representative US CKD population.
In conclusion, our results suggest that GA levels were independently associated with risks of ESKD, CVD, and mortality in patients with CKD, regardless of diabetes status. GA added prognostic value to HbA1c among patients with coexisting CKD and diabetes. These data support that GA may be used as a complementary test to HbA1c among patients with CKD.
Supplementary Material
Figure S1. Flowchart of the Selection of Study Participants.
Figure S2. The Scatterplot with Lowess Lines Showed the Relationship Between GA and HbA1c Among 1516 Patients with Coexisting CKD and Diabetes.
Figure S3. Adjusted HRs and 95% CIs for GA Levels and Risks of ESKD, CVD and Death Among Patients with Coexisting CKD and Diabetes in the CRIC Study.
Figure S4. Adjusted HRs and 95% CIs for GA Levels and Risks of ESKD, CVD and Death Among CKD Patients without Diabetes in the CRIC Study.
Table S1. Baseline Characteristics According to Diabetes Status Among CRIC Study Participants.
Table S2. Sensitivity analysis 1: Risks of ESKD, CVD, and Death according to glycated albumin quartiles among 3110 study participants with CKD, using multiple imputation for missing covariates (20 iterations and 40 datasets).
Table S3. Sensitivity analysis 2: Risks of ESKD, CVD, and Death according to glycated albumin quartiles among 3110 study participants with CKD, after adding total serum albumin level as a covariate in main model (Model 2).
Table S4. Sensitivity analysis 3: Risks of ESKD and CVD according to glycated albumin quartiles among 3110 study participants with CKD, using Fine and Gray competing risks regression to account for death as a competing event.
Table S5. Adjusted Hazard Ratios of ESKD, CVD, and Death According to GA Quartiles Among CKD Patients without Diabetes.
Acknowledgments:
The authors thank the CRIC investigators and participants for their important contributions. The CRIC study was conducted by the CRIC study Investigators and supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The data and biospecimens from the CRIC study reported here were supplied by the NIDDK Central Repository. This manuscript was not prepared in collaboration with Investigators of the CRIC study and does not necessarily reflect the opinions or views of the CRIC study, the NIDDK Central Repository, or the NIDDK.
Support:
Research reported in this publication was supported by grant R01DK124453 (Dr. Kalim) from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). Dr. Tang is supported by grant AHA23POST1010825 from the American Heart Association. Dr. Berg is supported by National Heart, Lung, and Blood Institute grant R01HL133399. Dr. Allegretti is supported by NIDDK grant K23DK128567. Dr. Lash is supported by NIDDK grants U01DK060980, R01DK072231-91, and K24DK092290. The funders did not have any role in study design, data collection, analysis, reporting, or the decision to submit for publication.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Financial Disclosure: The authors declare that they have no relevant financial interests.
Prior Presentation: Presented in part at the American Society of Nephrology (ASN) Kidney Week; November 2, 2023; Philadelphia, PA.
Data Sharing:
Anonymized data and original data reported in this paper of type observational data have been deposited to CRIC and will be uploaded to NIDDK Biorepository on publication.
References
- 1.Tuttle K, Wong L, St.Peter W, et al. Moving from Evidence to Implementation of Breakthrough Therapies for Diabetic Kidney Disease. CLIN J AM SOC NEPHROL. 2022;17(7):1092–1103. doi: 10.2215/CJN.02980322 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Navaneethan SD, Zoungas S, Caramori ML, et al. Diabetes Management in Chronic Kidney Disease: Synopsis of the KDIGO 2022 Clinical Practice Guideline Update. Ann Intern Med. 2023;176(3):381–387. doi: 10.7326/M22-2904 [DOI] [PubMed] [Google Scholar]
- 3.Xie Y, Bowe B, Mokdad AH, et al. Analysis of the Global Burden of Disease study highlights the global, regional, and national trends of chronic kidney disease epidemiology from 1990 to 2016. Kidney International. 2018;94(3):567–581. doi: 10.1016/j.kint.2018.04.011 [DOI] [PubMed] [Google Scholar]
- 4.ElSayed NA, Aleppo G, Aroda VR, et al. 6. Glycemic Targets: Standards of Care in Diabetes—2023. Diabetes Care. 2022;46(Supplement 1):S97–S110. doi: 10.2337/dc23-S006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Selvin E, Steffes MW, Zhu H, et al. Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults. N Engl J Med. 2010;362(9):800–811. doi: 10.1056/NEJMoa0908359 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kidney Disease: Improving Global Outcomes (KDIGO) Diabetes Work Group. KDIGO 2022 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease. Kidney Int. 2022;102(5S):S1–S127. doi: 10.1016/j.kint.2022.06.008 [DOI] [PubMed] [Google Scholar]
- 7.Sharif A, Baboolal K. Diagnostic Application of the A1c Assay in Renal Disease. Journal of the American Society of Nephrology. 2010;21(3):383–385. doi: 10.1681/ASN.2010010031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ng JM, Cooke M, Bhandari S, Atkin SL, Kilpatrick ES. The Effect of Iron and Erythropoietin Treatment on the A1C of Patients With Diabetes and Chronic Kidney Disease. Diabetes Care. 2010;33(11):2310–2313. doi: 10.2337/dc10-0917 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Welsh KJ, Kirkman MS, Sacks DB. Role of Glycated Proteins in the Diagnosis and Management of Diabetes: Research Gaps and Future Directions. Diabetes Care. 2016;39(8):1299–1306. doi: 10.2337/dc15-2727 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gan T, Liao B, Xu G. The clinical usefulness of glycated albumin in patients with diabetes and chronic kidney disease: Progress and challenges. Journal of Diabetes and its Complications. 2018;32(9):876–884. doi: 10.1016/j.jdiacomp.2018.07.004 [DOI] [PubMed] [Google Scholar]
- 11.Kirkman MS, Sacks DB. Glycated Albumin: Added Value or Redundancy in Diabetes Care? Clinical Chemistry. 2022;68(3):379–381. doi: 10.1093/clinchem/hvab261 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Fonseca V, Kohzuma T, Galindo RJ, DeSouza C. KDIGO recommendations for the evaluation of glycemic control in advanced chronic kidney disease. Kidney International. 2022;101(2):420. doi: 10.1016/j.kint.2021.11.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Sacks DB, Arnold M, Bakris GL, et al. Executive Summary: Guidelines and Recommendations for Laboratory Analysis in the Diagnosis and Management of Diabetes Mellitus. Diabetes Care. 2023;46(10):1740–1746. doi: 10.2337/dci23-0048 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kohzuma T, Tao X, Koga M. Glycated albumin as biomarker: Evidence and its outcomes. Journal of Diabetes and its Complications. 2021;35(11):108040. doi: 10.1016/j.jdiacomp.2021.108040 [DOI] [PubMed] [Google Scholar]
- 15.Gan T, Liu X, Xu G. Glycated Albumin Versus HbA1c in the Evaluation of Glycemic Control in Patients With Diabetes and CKD. Kidney International Reports. 2018;3(3):542–554. doi: 10.1016/j.ekir.2017.11.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chume FC, Schiavenin LG, Freitas PAC, Pimentel AL, Camargo JL. The usefulness of glycated albumin in patients with diabetes and renal disease: a scoping review. J Lab Precis Med. 2022;7:12–12. doi: 10.21037/jlpm-22-2 [DOI] [Google Scholar]
- 17.Selvin E, Rawlings AM, Grams M, et al. Fructosamine and glycated albumin for risk stratification and prediction of incident diabetes and microvascular complications: a prospective cohort analysis of the Atherosclerosis Risk in Communities (ARIC) study. The Lancet Diabetes & Endocrinology. 2014;2(4):279–288. doi: 10.1016/S2213-8587(13)70199-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Selvin E, Rawlings AM, Lutsey PL, et al. Fructosamine and Glycated Albumin and the Risk of Cardiovascular Outcomes and Death. Circulation. 2015;132(4):269–277. doi: 10.1161/CIRCULATIONAHA.115.015415 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mihara A, Ohara T, Hata J, et al. Association between serum glycated albumin and risk of cardiovascular disease in a Japanese community: The Hisayama Study. Atherosclerosis. 2020;311:52–59. doi: 10.1016/j.atherosclerosis.2020.08.016 [DOI] [PubMed] [Google Scholar]
- 20.Fang M, Daya N, Coresh J, Christenson RH, Selvin E. Glycated Albumin for the Diagnosis of Diabetes in US Adults. Clinical Chemistry. 2022;68(3):413–421. doi: 10.1093/clinchem/hvab231 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Rooney MR, Daya N, Tang O, et al. Glycated Albumin and Risk of Mortality in the US Adult Population. Clinical Chemistry. 2022;68(3):422–430. doi: 10.1093/clinchem/hvab232 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nathan DM, McGee P, Steffes MW, Lachin JM, the DCCT/EDIC Research Group. Relationship of Glycated Albumin to Blood Glucose and HbA1c Values and to Retinopathy, Nephropathy, and Cardiovascular Outcomes in the DCCT/EDIC Study. Diabetes. 2014;63(1):282–290. doi: 10.2337/db13-0782 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Freedman BI, Andries L, Shihabi ZK, et al. Glycated Albumin and Risk of Death and Hospitalizations in Diabetic Dialysis Patients. CJASN. 2011;6(7):1635–1643. doi: 10.2215/CJN.11491210 [DOI] [PubMed] [Google Scholar]
- 24.Shafi T, Sozio SM, Plantinga LC, et al. Serum Fructosamine and Glycated Albumin and Risk of Mortality and Clinical Outcomes in Hemodialysis Patients. Diabetes Care. 2013;36(6):1522–1533. doi: 10.2337/dc12-1896 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chen CW, Drechsler C, Suntharalingam P, Karumanchi SA, Wanner C, Berg AH. High Glycated Albumin and Mortality in Persons with Diabetes Mellitus on Hemodialysis. Clinical Chemistry. 2017;63(2):477–485. doi: 10.1373/clinchem.2016.258319 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hoshino J, Abe M, Hamano T, et al. Glycated albumin and hemoglobin A1c levels and cause-specific mortality by patients’ conditions among hemodialysis patients with diabetes: a 3-year nationwide cohort study. BMJ Open Diab Res Care. 2020;8(1):e001642. doi: 10.1136/bmjdrc-2020-001642 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hoshino J, Hamano T, Abe M, et al. Glycated albumin versus hemoglobin A1c and mortality in diabetic hemodialysis patients: a cohort study. Nephrology Dialysis Transplantation. 2018;33(7):1150–1158. doi: 10.1093/ndt/gfy014 [DOI] [PubMed] [Google Scholar]
- 28.Feldman HI, Appel LJ, Chertow GM, et al. The Chronic Renal Insufficiency Cohort (CRIC) Study: Design and Methods. J Am Soc Nephrol. 2003;14(suppl 2):S148. doi: 10.1097/01.ASN.0000070149.78399.CE [DOI] [PubMed] [Google Scholar]
- 29.Lash JP, Go AS, Appel LJ, et al. Chronic Renal Insufficiency Cohort (CRIC) Study: Baseline Characteristics and Associations with Kidney Function. CLIN J AM SOC NEPHROL. 2009;4(8):1302. doi: 10.2215/CJN.00070109 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tang M, Berg A, Rhee EP, et al. The Impact of Carbamylation and Anemia on HbA1c’s Association With Renal Outcomes in Patients With Diabetes and Chronic Kidney Disease. Diabetes Care. 2023;46(1):130–137. doi: 10.2337/dc22-1399. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Chen J, Budoff MJ, Reilly MP, et al. Coronary Artery Calcification and Risk of Cardiovascular Disease and Death Among Patients With Chronic Kidney Disease. JAMA Cardiol. 2017;2(6):635. doi: 10.1001/jamacardio.2017.0363 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Anderson AH, Yang W, Hsu C yuan, et al. Estimating GFR Among Participants in the Chronic Renal Insufficiency Cohort (CRIC) Study. American Journal of Kidney Diseases. 2012;60(2):250–261. doi: 10.1053/j.ajkd.2012.04.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Harrell FE Jr. Statistically efficient ways to quantify added predictive value of new measurements. Statistical Thinking. Published October 7, 2018. Accessed January 20, 2024. https://www.fharrell.com/post/addvalue/ [Google Scholar]
- 34.Fronczek J, Polok K, de Lange DW, et al. Relationship between the Clinical Frailty Scale and short-term mortality in patients ≥ 80 years old acutely admitted to the ICU: a prospective cohort study. Crit Care. 2021;25(1):231. doi: 10.1186/s13054-021-03632-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Selvin E Hemoglobin A1c—Using Epidemiology to Guide Medical Practice: Kelly West Award Lecture 2020. Diabetes Care. 2021;44(10):2197–2204. doi: 10.2337/dci21-0035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Bomholt T, Adrian T, Nørgaard K, et al. The Use of HbA1c, Glycated Albumin and Continuous Glucose Monitoring to Assess Glucose Control in the Chronic Kidney Disease Population Including Dialysis. Nephron. 2021;145(1):14–19. doi: 10.1159/000511614 [DOI] [PubMed] [Google Scholar]
- 37.Selvin E, Sacks DB. Monitoring Glycemic Control in End-Stage Renal Disease: What Should Be Measured? Clinical Chemistry. 2017;63(2):447–449. doi: 10.1373/clinchem.2016.265744 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Mehrotra R, Kalantar-Zadeh K, Adler S. Assessment of Glycemic Control in Dialysis Patients with Diabetes: Glycosylated Hemoglobin or Glycated Albumin? Clinical Journal of the American Society of Nephrology. 2011;6(7):1520. doi: 10.2215/CJN.04210511 [DOI] [PubMed] [Google Scholar]
- 39.Cohen RM, Herman WH. Are glycated serum proteins ready for prime time? The Lancet Diabetes & Endocrinology. 2014;2(4):264–265. doi: 10.1016/S2213-8587(14)70003-8 [DOI] [PubMed] [Google Scholar]
- 40.Caramori ML, Tuttle KR, Tandon N, et al. The authors reply: Kidney International. 2022;101(2):420–421. doi: 10.1016/j.kint.2021.11.019 [DOI] [PubMed] [Google Scholar]
- 41.Selvin E, Francis LMA, Ballantyne CM, et al. Nontraditional Markers of Glycemia. Diabetes Care. 2011;34(4):960–967. doi: 10.2337/dc10-1945 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Kim H, Tang O, Rebholz CM, et al. Associations of Glycated Albumin and HbA1c with Chronic Kidney Disease in US Adults. The Journal of Applied Laboratory Medicine. 2022;7(4):842–853. doi: 10.1093/jalm/jfac006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Molitoris BA, Wagner MC. Is Albumin Toxic to the Kidney?: It Depends. Clinical Journal of the American Society of Nephrology. 2023;18(9):1222–1224. doi: 10.2215/CJN.0000000000000153 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Rodiño-Janeiro BK, González-Peteiro M, Ucieda-Somoza R, González-Juanatey JR, Álvarez E. Glycated albumin, a precursor of advanced glycation end-products, up-regulates NADPH oxidase and enhances oxidative stress in human endothelial cells: molecular correlate of diabetic vasculopathy. Diabetes/Metabolism Research and Reviews. 2010;26(7):550–558. doi: 10.1002/dmrr.1117 [DOI] [PubMed] [Google Scholar]
- 45.Qi W, Niu J, Qin Q, Qiao Z, Gu Y. Astragaloside IV attenuates glycated albumin-induced epithelial-to-mesenchymal transition by inhibiting oxidative stress in renal proximal tubular cells. Cell Stress Chaperones. 2014;19(1):105–114. doi: 10.1007/s12192-013-0438-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Wang HJ, Lo WY, Lin LJ. Angiotensin-(1–7) decreases glycated albumin-induced endothelial interleukin-6 expression via modulation of miR-146a. Biochem Biophys Res Commun. 2013;430(3):1157–1163. doi: 10.1016/j.bbrc.2012.12.018 [DOI] [PubMed] [Google Scholar]
- 47.Chen S, Cohen MP, Lautenslager GT, Shearman CW, Ziyadeh FN. Glycated albumin stimulates TGF-βbgr;1 production and protein kinase C activity in glomerular endothelial cells1. Kidney International. 2001;59(2):673–681. doi: 10.1046/j.1523-1755.2001.059002673.x [DOI] [PubMed] [Google Scholar]
- 48.Gross ML, Piecha G, Bierhaus A, et al. Glycated and carbamylated albumin are more “nephrotoxic” than unmodified albumin in the amphibian kidney. American Journal of Physiology-Renal Physiology. 2011;301(3):F476–F485. doi: 10.1152/ajprenal.00342.2010 [DOI] [PubMed] [Google Scholar]
- 49.Molitoris BA, Sandoval RM, Yadav SPS, Wagner MC. Albumin uptake and processing by the proximal tubule: physiological, pathological, and therapeutic implications. Physiol Rev. 2022;102(4):1625–1667. doi: 10.1152/physrev.00014.2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Carson AP, Fox CS, McGuire DK, et al. Low Hemoglobin A1c and Risk of All-Cause Mortality Among US Adults Without Diabetes. Circulation: Cardiovascular Quality and Outcomes. 2010;3(6):661–667. doi: 10.1161/CIRCOUTCOMES.110.957936 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Navaneethan SD, Schold JD, Jolly SE, Arrigain S, Winkelmayer WC, Nally JV. Diabetes Control and the Risks of ESRD and Mortality in Patients With CKD. American Journal of Kidney Diseases. 2017;70(2):191–198. doi: 10.1053/j.ajkd.2016.11.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Inaba M, Okuno S, Kumeda Y, et al. Glycated Albumin Is a Better Glycemic Indicator than Glycated Hemoglobin Values in Hemodialysis Patients with Diabetes: Effect of Anemia and Erythropoietin Injection. JASN. 2007;18(3):896–903. doi: 10.1681/ASN.2006070772 [DOI] [PubMed] [Google Scholar]
- 53.Fukami K, Shibata R, Nakayama H, Yamada K, Okuda S, Koga M. Serum albumin-adjusted glycated albumin reflects glycemic excursion in diabetic patients with severe chronic kidney disease not treated with dialysis. Journal of Diabetes and its Complications. 2015;29(7):913–917. doi: 10.1016/j.jdiacomp.2015.06.003 [DOI] [PubMed] [Google Scholar]
- 54.Hassanein M, Shafi T. Assessment of glycemia in chronic kidney disease. BMC Med. 2022;20(1):117. doi: 10.1186/s12916-022-02316-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Selvin E, Wang D, Rooney MR, et al. Within-Person and Between-Sensor Variability in Continuous Glucose Monitoring Metrics. Clinical Chemistry. 2023;69(2):180–188. doi: 10.1093/clinchem/hvac192 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Siegmund T, Heinemann L, Kolassa R, Thomas A. Discrepancies Between Blood Glucose and Interstitial Glucose-Technological Artifacts or Physiology: Implications for Selection of the Appropriate Therapeutic Target. J Diabetes Sci Technol. 2017;11(4):766–772. doi: 10.1177/1932296817699637 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Galindo R, de Boer I, Neumiller J, Tuttle K. Continuous Glucose Monitoring to Optimize Management of Diabetes in Patients with Advanced CKD. CLIN J AM SOC NEPHROL. Published online August 31, 2022:CJN.04510422. doi: 10.2215/CJN.04510422 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Cohen RM, Sacks DB. Comparing Multiple Measures of Glycemia: How to Transition from Biomarker to Diagnostic Test? Clinical Chemistry. 2012;58(12):1615–1617. doi: 10.1373/clinchem.2012.196139 [DOI] [PubMed] [Google Scholar]
- 59.Inker LA, Eneanya ND, Coresh J, et al. New Creatinine- and Cystatin C–Based Equations to Estimate GFR without Race. New England Journal of Medicine. 2021;385(19):1737–1749. doi: 10.1056/NEJMoa2102953 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Cohen RM, Holmes YR, Chenier TC, Joiner CH. Discordance Between HbA1c and Fructosamine: Evidence for a glycosylation gap and its relation to diabetic nephropathy. Diabetes Care. 2003;26(1):163–167. doi: 10.2337/diacare.26.1.163 [DOI] [PubMed] [Google Scholar]
- 61.Hoshino J, Abe M, Hamano T, et al. Glycated albumin to glycated hemoglobin ratio and mortality in diabetic patients on dialysis: a new association. Nephrology Dialysis Transplantation. Published online October 29, 2022:gfac297. doi: 10.1093/ndt/gfac297 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Flowchart of the Selection of Study Participants.
Figure S2. The Scatterplot with Lowess Lines Showed the Relationship Between GA and HbA1c Among 1516 Patients with Coexisting CKD and Diabetes.
Figure S3. Adjusted HRs and 95% CIs for GA Levels and Risks of ESKD, CVD and Death Among Patients with Coexisting CKD and Diabetes in the CRIC Study.
Figure S4. Adjusted HRs and 95% CIs for GA Levels and Risks of ESKD, CVD and Death Among CKD Patients without Diabetes in the CRIC Study.
Table S1. Baseline Characteristics According to Diabetes Status Among CRIC Study Participants.
Table S2. Sensitivity analysis 1: Risks of ESKD, CVD, and Death according to glycated albumin quartiles among 3110 study participants with CKD, using multiple imputation for missing covariates (20 iterations and 40 datasets).
Table S3. Sensitivity analysis 2: Risks of ESKD, CVD, and Death according to glycated albumin quartiles among 3110 study participants with CKD, after adding total serum albumin level as a covariate in main model (Model 2).
Table S4. Sensitivity analysis 3: Risks of ESKD and CVD according to glycated albumin quartiles among 3110 study participants with CKD, using Fine and Gray competing risks regression to account for death as a competing event.
Table S5. Adjusted Hazard Ratios of ESKD, CVD, and Death According to GA Quartiles Among CKD Patients without Diabetes.
Data Availability Statement
Anonymized data and original data reported in this paper of type observational data have been deposited to CRIC and will be uploaded to NIDDK Biorepository on publication.