Comparison of cardiovascular event predictability between the 2009 and 2021 Chronic Kidney Disease Epidemiology Collaboration equations in a Korean chronic kidney disease cohort: the KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease study

Article information

Kidney Res Clin Pract. 2023;42(6):700-711
Publication date (electronic) : 2023 March 22
doi : https://doi.org/10.23876/j.krcp.22.206
1Department of Internal Medicine, Seoul National University Hospital, Seoul, Republic of Korea
2Department of Internal Medicine, Ewha Womans University College of Medicine, Ewha Womans University Seoul Hospital, Seoul, Republic of Korea
3Department of Biostatistics, Korea University College of Medicine, Seoul, Republic of Korea
4Medical Research Collaborating Center, Seoul National University Hospital, Seoul, Republic of Korea
5Department of Preventive Medicine, Seoul National University College of Medicine, Seoul, Republic of Korea
6Department of Internal Medicine, Seoul National University Bundang Hospital, Seongnam, Republic of Korea
7Department of Internal Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea
8Department of Internal Medicine, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea
Correspondence: Kook-Hwan Oh Department of Internal Medicine, Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 03080, Republic of Korea. E-mail: khoh@snu.ac.kr
Received 2022 September 12; Revised 2022 October 10; Accepted 2022 November 9.

Abstract

Background

The 2009 Chronic Kidney Disease Epidemiology Collaboration creatinine-based estimated glomerular filtration rate (eGFRcr) equation contains a race component that is not based on biology and may cause a bias in results. Therefore, the 2021 eGFRcr and creatinine-cystatin C–based eGFR (eGFRcr-cysC) equations were developed with no consideration of race. This study compared the cardiovascular event (CVE) and all-cause mortality and CVE combined predictability among the three eGFR equations in Korean chronic kidney disease (CKD) patients.

Methods

This study included 2,207 patients from the KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease. Receiver operating characteristic (ROC) and net reclassification improvement (NRI) index were used to compare the predictability of the study outcomes according to the 2009 eGFRcr, 2021 eGFRcr, and 2021 eGFRcr-cysC equations.

Results

The overall prevalence of CVE and all-cause mortality were 9% and 7%, respectively. There was no difference in area under the curve of ROC for CVE and mortality and CVE combined among all three equations. Compared to the 2009 eGFRcr, both the 2021 eGFRcr (NRI, 0.013; 95% confidence interval [CI], –0.002 to 0.028) and the eGFRcr-cysC (NRI, –0.001; 95% CI, –0.031 to 0.029) equations did not show improved CVE predictability. Similar findings were observed for mortality and CVE combined predictability with both the 2021 eGFRcr (NRI, –0.019; 95% CI, –0.039–0.000) and the eGFRcr-cysC (NRI, –0.002; 95% CI, –0.023 to 0.018).

Conclusion

The 2009 eGFRcr equation was not inferior to either the 2021 eGFRcr or eGFRcr-cysC equation in predicting CVE and the composite of mortality and CVE in Korean CKD patients.

Graphical abstract

Introduction

Estimated glomerular filtration rate (eGFR) is a surrogate marker of kidney function calculated using endogenous filtration byproducts of creatinine and cystatin C (cysC). In 1999, the Modification of Diet in Renal Disease (MDRD) eGFR equation was developed from a cohort of Caucasian and African-American chronic kidney disease (CKD) patients with glomerular filtration rate (GFR) less than 60 mL/min/1.73 m2 [1]. However, eGFR calculated from this equation was often underestimated in patients with GFR greater than 60 mL/min/1.73 m2. To overcome this limitation, the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) developed a new eGFR equation using creatinine in 2009. The new equation was more accurate than that of the MDRD, especially in patients with GFR less than 60 mL/min/1.73 m2 [2]. Since the introduction of the 2009 CKD-EPI eGFR creatinine (eGFRcr) equation, it has been employed widely in global medical practice.

Factors including ethnicity, age, and sex, which are associated with amount of muscle mass and muscle metabolism, affect the creatinine level and, consequently, eGFR calculations [3]. Young to middle-aged African-Americans, especially males, often have higher creatinine levels compared to Caucasians [4]. Therefore, both MDRD and CKD-EPI eGFRcr equations had correction factors for age, sex, and African-American race [1,2]. However, this led to disproportionate diagnosis of CKD in different ethnicities due to varying levels of creatinine and cysC and use of a ‘race’ coefficient developed to correct eGFR difference between African-American and white individuals [5]. Also, lack of consideration of individual diversity in African-Americans often led to inappropriate early or delayed referral to nephrologists [6,7]. Therefore, in 2021, the CKD-EPI group developed two new eGFR equations using creatinine (eGFRcr) and creatinine-cysC (eGFRcr-cysC) and omitting the race factor [8].

In 2022, the National Kidney Foundation and American Society of Nephrology published a joint statement that recommends use of the 2021 eGFRcr or eGFRcr-cysC equation over the current 2009 eGFRcr equation [9]. Other studies have also reported that the 2021 eGFR equations were more accurate than the 2009 eGFRcr equation, especially in African-Americans with lower kidney function [10,11]. Therefore, the new 2021 equations have the greatest benefits in CKD screening, detection, and risk prediction in African-Americans adults [9]. However, its benefits in an Asian population are unclear as the percentage of Asian CKD patients included in previous landmark studies was low, and those included were classified as ‘non-black,’ which mostly included Caucasians [8,10,11]. Also, previous studies that proposed an Asian coefficient for MDRD and 2009 CKD-EPI eGFR equations reported variable values depending on Asian ethnicity and study methods [12].

It is widely known that cardiovascular morbidity and mortality risk are elevated in both early and advanced stages of CKD [13,14]. Therefore, it is important to accurately predict cardiovascular event (CVE) risk in CKD patients. As Korea is a monoethnic Asian country, this study was conducted to evaluate the efficiency of the new 2021 eGFRcr and eGFRcr-cysC equations, which do not include a race factor, compared to the 2009 eGFRcr equation. The predictability of CVE and composite of all-cause mortality and CVE among eGFR equations was compared in a Korean nondialysis CKD cohort.

Methods

Study design

The KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease (KNOW-CKD) is a national, prospective, multicenter study of Korean nondialysis CKD patients from nine major nephrology centers of university hospitals in Korea. Study exclusion criteria are 1) inability or unwillingness to provide written consent; 2) previous maintenance dialysis or organ transplantation; 3) heart failure (New York Heart Association functional class 3 or 4) or cirrhosis (Child-Pugh class 2 or 3); 4) history of or current malignancy; 5) pregnancy; or 6) single kidney due to trauma or donation.

From September 2011 to January 2016, a total of 2,238 adult nondialysis CKD patients between the ages of 20 and 75 years were enrolled. Patients with missing information on CVE and all-cause mortality due to follow-up loss within 6 months of study entry were excluded.

The KNOW-CKD study was conducted in accordance with the principles of the Declaration of Helsinki and was supervised by the Korea Centers for Disease Control and Prevention. The study was approved by the Institutional Review Boards of all nine university hospitals including Seoul National University Hospital in 2011 (No. 1104-089-359). A detailed study protocol of the KNOW-CKD has been previously published [15].

Laboratory and clinical variables

All laboratory and clinical variables were collected from patients on their initial visit to the enrolled hospital.

Blood samples were collected after at least 8 hours of fasting. Baseline laboratory measurements were hemoglobin (Hb), blood urea nitrogen (BUN), creatinine, cysC, sodium, potassium, calcium, phosphorus, uric acid, parathyroid hormone (PTH), high-sensitivity C-reactive protein (hs-CRP), troponin T, fasting glucose, HbA1C, total cholesterol, high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C).

Serum creatinine, cysC, PTH, and urine protein and creatinine values were measured at one central laboratory (LabGenomics). Serum creatinine was measured using the isotope dilution mass spectrometry-traceable method, and serum cysC was measured using the immunonephelometry method for consistency [16]. Other laboratory measurements were conducted in the appropriate hospital laboratory.

Baseline clinical information on age, sex, underlying comorbidity, medication, and lifestyle patterns including cigarette smoking status (never, former, current) were collected using self-reported questionnaires with the assistance of trained staff. Information regarding CVE was collected through medical interview and review of the patient’s electronic medical records during the hospital visits initially and at 6 months after enrollment and annually after that. Review of the CVE by the nephrologist in charge of the KNOW-CKD study in each hospital was repeated, and CVE was classified into 10 categories: acute myocardial infarction, hospitalization for unstable angina or heart failure, percutaneous coronary artery intervention or coronary bypass graft surgery, ischemic stroke, hemorrhagic stroke, carotid artery disease, peripheral artery disease, symptomatic arrhythmia, and any other CVE that required hospitalization or intervention. Finally, CVE was cross-checked by another nephrologist among the participating hospitals of the KNOW-CKD study to ensure accuracy and objectivity of the outcome data. For fatal CVE, information regarding the time and causes of mortality was obtained from the patient’s electronic medical records or the Korean Statistical Information Service.

Blood pressure was measured by a trained nurse using an electronic sphygmomanometer after 5 minutes of rest in a sitting position. Hypertension was defined as (a) systolic blood pressure (SBP) of >140 mmHg or diastolic blood pressure (DBP) of >90 mmHg or (b) previous diagnosis of hypertension. Mean arterial pressure (MAP) was calculated using the following equation: MAP = DBP + 1/3 (SBP – DBP).

Diabetes mellitus (DM) was defined as (a) fasting serum glucose of >126 mg/dL or (b) previous diagnosis of DM.

Estimated glomerular filtration rate calculation

Each eGFR was calculated using three CKD-EPI equations: 2009 CKD-EPI eGFRcr, 2021 CKD-EPI eGFRcr, and 2021 eGFRcr-cysC equations [2,8]. These equations are listed below.

2009 CKD-EPI eGFRcr equation

Female

- Serum creatinine (Scr) ≤ 0.7 mg/dL: 144 × (Scr/0.7)–0.329 × (0.993)Age

- Scr > 0.7 mg/dL: 144 × (Scr/0.7)–1.209 × (0.993)Age

Male

- Scr ≤ 0.9 mg/dL: 141 × (Scr/0.9)–0.411 × (0.993)Age

- Scr > 0.9 mg/dL: 141 × (Scr/0.9)–1.209 × (0.993)Age

2021 CKD-EPI eGFRcr equation

Female

- Scr ≤ 0.7 mg/dL: 142 × (Scr/0.7)–0.241 × 0.9938Age × 1.012

- Scr > 0.7 mg/dL: 142 × (Scr/0.7)–1.200 × 0.9938Age × 1.012

Male

- Scr ≤ 0.9 mg/dL: 142 × (Scr/0.9)–0.302 × 0.9938Age

- Scr > 0.9 mg/dL: 142 × (Scr/0.9)–1.200 × 0.9938Age

2021 CKD-EPI eGFRcr-cysC equation

Female

- Scr ≤ 0.7 mg/dL

1) Serum cysC (ScysC) ≤ 0.8 mg/dL: 135 × (Scr/0.7)–0.219 × (ScysC/0.8)–0.323 × 0.9961Age × 0.963

2) ScysC > 0.8 mg/dL: 135 × (Scr/0.7)–0.219 × (ScysC/0.8)–0.778 × 0.9961Age × 0.963

- Scr > 0.7 mg/dL

1) ScysC ≤ 0.8 mg/dL: 135 × (Scr/0.7)–0.544 × (ScysC/0.8)–0.323 × 0.9961Age × 0.963

2) ScysC > 0.8 mg/dL: 135 × (Scr/0.7)–0.544 × (ScysC/0.8)–0.778 × 0.9961Age × 0.963

Male

- Scr ≤ 0.9 mg/dL

1) ScysC ≤ 0.8 mg/dL: 135 × (Scr/0.9)–0.144 × (ScysC/0.8)–0.323 × 0.9961Age

2) ScysC > 0.8 mg/dL: 135 × (Scr/0.9)–0.144 × (ScysC/0.8)–0.778 × 0.9961Age

- Scr > 0.9 mg/dL

1) ScysC ≤ 0.8 mg/dL: 135 × (Scr/0.9)–0.544 × (ScysC/0.8)–0.323 × 0.9961Age

2) ScysC > 0.8 mg/dL: 135 × (Scr/0.9)–0.544 × (ScysC/0.8)–0.778 × 0.9961Age

Study outcomes

As cardiovascular disease is one of the most critical complications of CKD, primary outcome was defined as the first occurrence of either non-fatal or fatal CVE. CVE included acute myocardial infarction, hospitalization for unstable angina or heart failure, percutaneous coronary artery intervention or coronary bypass graft surgery, ischemic stroke, hemorrhagic stroke, carotid artery disease, peripheral artery disease, symptomatic arrhythmia, or any other CVE that required hospitalization or intervention. The secondary outcome was the composite event of all-cause mortality and CVE.

Statistical analysis

Baseline characteristics were analyzed according to CVE. Continuous variables were expressed as mean ± standard deviation or median and interquartile range and analyzed using the Kruskal-Wallis test or one-way analysis of variance. Categorical variables were expressed as percentages. Comparison of categorical variables was conducted using chi-square test. Cox proportional hazard analysis was used to evaluate the predictive risk of CVE and all-cause mortality and CVE combined according to eGFR equation. Evaluated risks were expressed as hazard ratio (HR) and 95% confidence interval (CI). For Cox proportional hazard analysis, the multivariable model was adjusted for age, sex, DM, smoking, body mass index (BMI), MAP, LDL-C, ejection fraction (EF), and proteinuria (urine protein/creatinine ratio, >0.2 g/day). Receiver operating characteristic (ROC) curve analysis was conducted to compare the predictability of the three equations on study outcome. Also, the net reclassification improvement (NRI) index was calculated to compare improvement in predictability of one equation over another. The multivariable model for ROC and NRI analysis was adjusted for age, sex, DM, smoking, BMI, MAP, LDL-C, and EF. Two-sided p-values of <0.05 were considered statistically significant. All statistical analyses were performed using R version 4.0.4 (R Foundation for Statistical Computing).

Results

Baseline characteristics of the study population

From September 2011 to January 2016, a total of 2,238 adult nondialysis CKD patients between the ages of 20 and 75 years were enrolled. Among them, 31 patients were excluded due to missing information on CVE and all-cause mortality or to follow-up loss within 6 months of study entry. Finally, 2,207 patients were enrolled in this study. The median follow-up duration of the above patients was 8.6 years (Fig. 1).

Figure 1.

Flow diagram of patient enrollment.

KNOW-CKD, KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease.

Among the total of 2,207 patients, the overall prevalence of CVE was 9.1% (n = 200). The types of CVE experienced by the 200 patients included acute myocardial infarction (n = 24), hospitalization for unstable angina (n = 23), hospitalization for heart failure (n = 14), percutaneous coronary artery intervention or coronary bypass graft surgery (n = 25), ischemic stroke (n = 34), hemorrhagic stroke (n = 17), carotid artery disease (n = 2), peripheral artery disease (n = 7), symptomatic arrhythmia (n = 13), and any other CVE that required hospitalization or intervention (n = 41).

Baseline characteristics of the study population were compared according to the presence of CVE. CVE patients were older with a larger percentage of males. In CVE patients, serum BUN, creatinine, and cysC levels were slightly higher and the eGFR level calculated using the 2009 eGFRcr equation was slightly lower than in non-CVE patients. Also, fasting glucose and HbA1C levels were higher with a higher percentage of underlying DM. Regarding the cardiovascular aspect, the percentage of patients with underlying coronary artery disease was approximately six-fold higher (23.5% vs. 4.3%, p < 0.001) in patients with CVE. In addition, CVE patients had slightly higher levels of hs-CRP and troponin T and lower left ventricular EF percentage as measured with echocardiogram. In comparison of lipid profiles, CVE patients had lower HDL-C, LDL-C, and total cholesterol levels. Regarding lifestyle, the percentages of current and former smokers were higher among CVE patients (Table 1).

Baseline characteristics of study population according to CVE

Comparison of estimated glomerular filtration rate level according to estimated glomerular filtration rate equation

In comparison of the three eGFR equations, eGFR level was approximately 3 mL/min/1.73 m2 higher when calculated using the 2021 eGFRcr equation and 1.5 mL/min/1.73 m2 higher when calculated using the 2021 eGFRcr-cysC equation compared to that calculated using the 2009 eGFRcr equation. The distribution of difference in calculated eGFR level between eGFR equations is shown using the Bland-Altman plot (Fig. 2).

Figure 2.

Bland-Altman plot of differences in eGFR among the three equations.

(A) Difference between 2009 eGFRcr and 2021 eGFRcr equation. (B) Difference between 2009 eGFRcr and 2021 eGFRcr-cysC equation. (C) Difference in eGFR between 2021 eGFRcr and 2021 eGFRcr-cysC equation.

eGFR, estimated glomerular filtration rate; eGFRcr, creatinine-based eGFR; eGFRcr-cysC, creatinine-cystatin C–based eGFR; SD, standard deviation.

In comparison of 2009 and 2021 eGFRcr equations, CKD stage classification according to the 2021 eGFRcr equation allocated a higher number of patients to CKD stages 1 and 2 (38.9% vs. 35.1%) and a lower number of patients to CKD stages 3 to 5 (61.1% vs. 64.9%). A similar trend was observed when CKD stages were classified according to the 2021 eGFRcr-cysC equation. Prevalence of CVE and composite of all-cause mortality and CVE were highest in patients with CKD stages 3 and 4 for all three equations (Table 2).

Comparison of average eGFR, CVE and composite of all-cause mortality and CVE prevalence according to CKD stages in each eGFR equation

Predictive value of estimated glomerular filtration rate for cardiovascular event and composite of all-cause mortality and cardiovascular event

The predictive value of eGFR for CVE, calculated using each of the three eGFR equations, was statistically significant only in the unadjusted model. In that model, every 10 mL/min/1.73 m2 increase in eGFR was associated with lower predicted risk of CVE in all three equations (2009 eGFRcr equation: HR, 0.90; 95% CI, 0.86–0.95; 2021 eGFRcr equation: HR, 0.91; 95% CI, 0.87–0.96; and 2021 eGFRcr-cysC equation: HR, 0.90; 95% CI, 0.85–0.95). However, in prediction of the composite of all-cause mortality and CVE, significantly lower predictive risks were observed in all univariate and multivariate models across all three equations (2009 eGFRcr equation: HR, 0.94; 95% CI, 0.89–0.99; 2021 eGFRcr equation: HR, 0.94, 95% CI, 0.90–0.99; and 2021 eGFRcr-cysC equation: HR, 0.92, 95% CI, 0.87–0.97) (Table 3).

Predictive value of each eGFR to CVE and composite of all-cause mortality and CVE

Comparison of cardiovascular event and the composite of all-cause mortality and cardiovascular event predictability

The area under the ROC curve (AUC) for CVE predictability was similar among the three equations (2009 eGFRcr equation: 0.715; 95% CI, 0.679–0.752; 2021 eGFRcr equation: 0.715; 95% CI, 0.679–0.752; and 2021 eGFRcr-cysC equation: 0.716; 95% CI, 0.679–0.753) (Fig. 3A). Similar findings were observed in comparison of predictability for the composite of all-cause mortality and CVE with slightly increased AUC value for the 2021 eGFRcr-cysC equation compared to that of 2009 and 2021 eGFRcr equations (2009 eGFRcr equation: 0.747; 95% CI, 0.719–0.776; 2021 eGFRcr equation: 0.747; 95% CI, 0.719–0.776; and 2021 eGFRcr-cysC equation: 0.751; 95% CI, 0.723–0.779) (Fig. 3B).

Figure 3.

Receiver operating characteristic curve analysis for CVE (A) and the composite of all-cause mortality and CVE predictability (B).

AUC, area under the curve; CVE, cardiovascular event.

Additionally, NRI was used to compare the predictability of CVE and the composite of all-cause mortality and CVE among eGFR equations. Neither the 2021 eGFRcr nor 2021 eGFRcr-cysC equation had improved predictive power for CVE and all-cause mortality outcomes compared to the 2009 eGFRcr equation, and none of the NRI values were significant. Also, in comparison of the 2021 eGFRcr and eGFRcr-cysC equations, there was no difference in predictability of the study outcomes (Table 4).

NRI for comparison of predictability according to eGFR equations

Subgroup analysis of comparison of cardiovascular event and the composite of all-cause mortality and cardiovascular event predictability

As the prevalence of CVE and all-cause mortality was higher in CKD stages 3 to 5, subgroup analysis of outcome predictability was conducted in CKD stages 1 and 2 (eGFR of ≥60 mL/min/1.73 m2) and 3 to 5 (eGFR of <60 mL/min/1.73 m2) groups. In comparison of CVE prediction, AUC values were similar among all three equations in both subgroups (Supplementary Fig. 1A, B; available online). According to the NRI analysis, there was no significant improvement in predictability among the three equations in CKD stages 3 to 5. However, in CKD stages 1 and 2, the 2009 eGFRcr equation had slightly improved predictability compared to the 2021 eGFRcr equation (NRI [%], –0.048; 95% CI, –0.064 to –0.032; p < 0.001). However, no significant improvement was observed in comparison to other equations (Supplementary Table 1, available online).

In comparison of predictability of combined all-cause mortality and CVE, the AUC values were similar among the three equations in both subgroups (Supplementary Fig. 1C, D, available online). In NRI analysis, there was no significant improvement of predictability in one equation over the other in CKD stages 1 and 2. However, in CKD stages 3 to 5, the 2009 eGFRcr equation had slightly improved predictability compared to the 2021 eGFRcr equation (NRI [%], –0.026; 95% CI, –0.050 to –0.002; p = 0.04). Also, the 2021 eGFRcr-cysC equation had slightly improved predictability compared to the 2021 eGFRcr equation (NRI [%], 0.039; 95% CI, 0.008–0.070; p = 0.01) (Supplementary Table 1, available online).

Discussion

The results of this study show that the average eGFR level calculated using the 2021 eGFRcr equation was approximately 3 mL/min/1.73 m2 higher and 1.5 mL/min/1.73 m2 higher than the results using the 2021 eGFRcr-cysC equation and the 2009 eGFRcr equation, respectively. There was an overall reduction in prevalence of CKD stages 3 to 5 when the 2021 eGFR equations were used for CKD stage classification (2009 eGFRcr, 64.9% vs. 2021 eGFRcr, 61.1% vs. 2021 eGFRcr-cysC, 63.4%). The AUC values for predictability of CVE and the composite of all-cause mortality and CVE were similar across equations, and the NRI values were not statistically significant. This suggests that the 2009 eGFRcr equation is not inferior in predicting CVE and the composite of all-cause mortality and CVE compared to the 2021 eGFRcr and eGFRcr-cysC equations.

The overall prevalence of CVE in this study was 9.1% (n = 200). CVE risk is increased in CKD patients and is the cause of death in approximately 40% to 50% of cases, which is approximately two-fold higher than in the general population with normal kidney function [17]. The CVE and mortality risks are increased in CKD patients through both traditional and non-traditional risk factors [13]. Traditional risk factors including DM, hypertension, dyslipidemia, and smoking aggravate atherosclerosis, which is associated not only with cardiovascular disease but also CKD progression [18]. Nontraditional risk factors include accelerated vascular calcification in vessels and cardiac valves and chronic systemic inflammation [19,20]. The overall CVE prevalence of this study was relatively low compared to other western CKD cohort studies, where the CVE prevalences were 33.4% (CRIC, United States), 47.2% (CRISIS, United Kingdom), and 39.1% (MERENA, Spain) [2124]. Also, another study that compared longitudinal outcomes across multiple international CKD cohorts, including the current KNOW-CKD cohort, showed that the CVE risk was lower in Korean and Japanese CKD cohorts compared to western CKD cohorts [25]. This finding may be due to differences in genetics, lifestyle patterns including diet, and lower incidence of traditional cardiovascular risk factors in Asian CKD patients compared to western patients [25,26]. Due to the relatively small number of CVEs in this study, the predictive value of eGFR for CVE did not show a statistically significant association across the three eGFR equations. However, the association between eGFR and the composite of all-cause mortality and CVE was significant in both univariate and multivariate analyses, where the analysis was conducted with a larger number of clinical outcome events. This finding shows that the power of event prediction was valid and significant in all three eGFR equations. Even though the overall CVE prevalence of this study was relatively low, the CVE group had distinctive traditional and nontraditional characteristics associated with CVE. The group was older with a higher percentage of metabolic comorbidities including hypertension, DM, and coronary artery disease. Also, inflammatory markers including hs-CRP were elevated, and the percentage of former and current smokers was higher in the CVE group.

As CVE and all-cause mortality events were concentrated in CKD stage 3 to 5 patients, subgroup analysis was conducted to compare the outcome predictability in early (stages 1 and 2) and advanced (stages 3 to 5) CKD stages. For CVE, the 2009 eGFRcr equation had slightly improved predictability compared to the 2021 eGFRcr equation in CKD stages 1 and 2. For the composite of all-cause mortality and CVE, the 2009 eGFRcr equation had slightly improved predictability compared to the 2021 eGFRcr equation. Also, the 2021 eGFRcr-cysC equation had slightly improved predictability compared to the 2021 eGFRcr equation in CKD stages 3 to 5. These findings are in line with the core results of our study that the current 2009 eGFRcr equation was not inferior to the 2021 eGFRcr and eGFRcr-cysC equations in CVE prediction. The 2021 eGFRcr-cysC equation had slightly improved predictability compared to the 2021 eGFRcr equation in advanced CKD patients, in accordance with the findings of Inker et al. [8], which showed that the 2021 eGFRcr-cysC equation was more accurate than the 2021 eGFRcr equation, with smaller differences in eGFR between the race groups. However, direct comparison of outcome predictability power and eGFR accuracy may not be appropriate.

The average eGFR level calculated using the 2021 eGFRcr equation was approximately 3 mL/min/1.73 m2 higher and 1.5 mL/min/1.73 m2 higher than that calculated using the 2021 eGFRcr-cysC equation and the 2009 eGFRcr equation. Therefore, approximately 10% of patients (224 of 2,207) were reclassified to a lower CKD category when using the 2021 eGFRcr equation compared to the 2009 eGFRcr equation. For the 2021 eGFRcr-cysC equation, approximately 8% of patients (181 of 2,145) were reclassified into a lower CKD category, while 7% of patients (141 of 2,147) were reclassified into a higher CKD category (Supplementary Table 2, available online). These findings are in agreement with the study by Inker et al. [8] that the calculated eGFR of the ‘non-black’ subpopulation was overestimated when using both 2021 eGFRcr and eGFRcr-cysC equations compared to the 2009 eGFRcr equation [8]. These findings are attributed to the changes in the variable constants of the new 2021 eGFR equations.

The prevalence of CKD 3 to 5 was reduced by 3.8% when using the 2021 eGFRcr equation and by 1.5% when using the 2021 eGFRcr-cysC equation. The Kidney Disease Improving Global Outcomes guidelines recommend thorough work-up, treatment, and regular follow-up for management of CKD and its complications, especially when the eGFR is less than 60 mL/min/1.73 m2. CVE risk is increased dramatically in CKD stages 3 to 5, and careful examinations and risk stratification are needed [13]. Inappropriate diagnosis of advanced CKD can be problematic in medication dose adjustment and unnecessary limitations of medication prescriptions including renin-angiotensin-system blockade agents. Even though the prevalence of CKD stages 3 to 5 was reduced using the 2021 eGFR equations, neither CVE nor the composite of all-cause mortality and CVE predictability differed with CKD 3 to 5 prevalence. This is in line with a previous study showing that the clinical significance of the new 2021 eGFR equations is minimal, especially in non-black patients [11].

The main limitation of this study is that the overall CVE prevalence was relatively low. Therefore, subtle differences in calculated eGFR among the three equations may not have sufficient statistical power to result in changes in CVE predictability. Also, there was no information on measured GFR using exogenous filtration markers. Therefore, no definite validation of accuracy of one equation over another was achieved as a direct comparison of calculated to measured eGFR values. However, to our knowledge, this is the first study to evaluate the efficacy of the 2021 eGFRcr and eGFRcr-cysC equations in a large-scale, all-Asian CKD cohort. As the KNOW-CKD study is an ongoing prospective cohort, CVE and all-cause mortality events were clearly defined and accurately documented. Also, serum creatinine and cysC were measured in a central laboratory, ensuring the accuracy of eGFR calculations.

In conclusion, the 2009 eGFRcr equation was not inferior to either the 2021 eGFRcr or eGFRcr-cysC equation in predicting risks of CVE and the composite of all-cause mortality and CVE in Korean CKD patients. Further longitudinal studies with higher CVE prevalence and availability of measured GFR are needed to validate the efficacy of the new 2021 eGFRcr and eGFRcr-cysC equations in Asian populations.

Supplementary Materials

Notes

Conflicts of interest

Tae-Hyun Yoo is the Editor-in-Chief of Kidney Research and Clinical Practice and was not involved in the review process of this article. All authors have no other conflicts of interest to declare.

Funding

The KNOW-CKD study was supported by Research Programs funded by the Korea Disease Control and Prevention Agency (2011E3300300, 2012E3301100, 2013E3301600, 2013E3301601, 2013E3301602, 2016E3300200, 2016E3300201, 2016E3300202, 2019E320100, 2019E320101, 2019E320102, and 2022‐11‐007) and by the Bio and Medical Technology Development Program of the National Research Foundation, funded by the Korean government (No. 2017M3A9E4044649).

Data sharing statement

The data presented in this study are available on request from the corresponding author.

Authors’ contributions

Conceptualization: JHK, MK, EK, KHO

Data curation: JHK, HR, YJ, JK, SKP

Formal analysis: JHK, YJ, JK

Investigation, Methodology: JK, SKP

Supervision: JCJ, THY, HL

Validation: JHK, THY, YK, YCK, SSH, HL

Writing–original draft: JHK

Writing–review & editing: JHK, YCK, SSH, HL, KHO

All authors read and approved the final manuscript.

References

1. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med 1999;130:461–470.
2. Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604–612.
3. Levey AS, Titan SM, Powe NR, Coresh J, Inker LA. Kidney disease, race, and GFR estimation. Clin J Am Soc Nephrol 2020;15:1203–1212.
4. Peralta CA, Risch N, Lin F, et al. The association of African ancestry and elevated creatinine in the Coronary Artery Risk Development in Young Adults (CARDIA) study. Am J Nephrol 2010;31:202–208.
5. Kramer H, Palmas W, Kestenbaum B, et al. Chronic kidney disease prevalence estimates among racial/ethnic groups: the multi-ethnic study of atherosclerosis. Clin J Am Soc Nephrol 2008;3:1391–1397.
6. Vyas DA, Eisenstein LG, Jones DS. Hidden in plain sight: reconsidering the use of race correction in clinical algorithms. N Engl J Med 2020;383:874–882.
7. Warren E, Wyden R, Booker C, Lee B. Agency for Healthcare Research and Quality (AHRQ) requesting a review of the use of race-based clinical algorithms in standard medical practices Congress of the United States; 2020.
8. Inker LA, Eneanya ND, Coresh J, et al. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med 2021;385:1737–1749.
9. Delgado C, Baweja M, Crews DC, et al. A unifying approach for GFR estimation: recommendations of the NKF-ASN Task Force on reassessing the inclusion of race in diagnosing kidney disease. Am J Kidney Dis 2022;79:268–288.e1.
10. Gregg LP, Richardson PA, Akeroyd J, Matheny ME, Virani SS, Navaneethan SD. Effects of the 2021 CKD-EPI creatinine eGFR equation among a national US veteran cohort. Clin J Am Soc Nephrol 2022;17:283–285.
11. Meeusen JW, Kasozi RN, Larson TS, Lieske JC. Clinical impact of the refit CKD-EPI 2021 creatinine-based eGFR equation. Clin Chem 2022;68:534–539.
12. Delanaye P, Cavalier E, Mariat C, Krzesinski JM, Rule AD. Estimating glomerular filtration rate in Asian subjects: where do we stand? Kidney Int 2011;80:439–440.
13. Jankowski J, Floege J, Fliser D, Böhm M, Marx N. Cardiovascular disease in chronic kidney disease: pathophysiological insights and therapeutic options. Circulation 2021;143:1157–1172.
14. Wright J, Hutchison A. Cardiovascular disease in patients with chronic kidney disease. Vasc Health Risk Manag 2009;5:713–722.
15. Oh KH, Park SK, Park HC, et al. KNOW-CKD (KoreaN cohort study for Outcome in patients With Chronic Kidney Disease): design and methods. BMC Nephrol 2014;15:80.
16. Grubb A, Blirup-Jensen S, Lindström V, Schmidt C, Althaus H, Zegers I. First certified reference material for cystatin C in human serum ERM-DA471/IFCC. Clin Chem Lab Med 2010;48:1619–1621.
17. Thompson S, James M, Wiebe N, et al. Cause of death in patients with reduced kidney function. J Am Soc Nephrol 2015;26:2504–2511.
18. Major RW, Cheng MRI, Grant RA, et al. Cardiovascular disease risk factors in chronic kidney disease: a systematic review and meta-analysis. PLoS One 2018;13e0192895.
19. O’Neill WC, Lomashvili KA. Recent progress in the treatment of vascular calcification. Kidney Int 2010;78:1232–1239.
20. Amdur RL, Feldman HI, Dominic EA, et al. Use of measures of inflammation and kidney function for prediction of atherosclerotic vascular disease events and death in patients with CKD: findings from the CRIC study. Am J Kidney Dis 2019;73:344–353.
21. Shah R, Matthews GJ, Shah RY, et al. Serum fractalkine (CX3CL1) and cardiovascular outcomes and diabetes: findings from the Chronic Renal Insufficiency Cohort (CRIC) study. Am J Kidney Dis 2015;66:266–273.
22. Ritchie J, Rainone F, Green D, et al. Extreme elevations in blood pressure and all-cause mortality in a referred CKD population: results from the CRISIS study. Int J Hypertens 2013;2013:597906.
23. Martínez-Castelao A, Górriz JL, Portolés JM, et al. Baseline characteristics of patients with chronic kidney disease stage 3 and stage 4 in Spain: the MERENA observational cohort study. BMC Nephrol 2011;12:53.
24. Tanaka K, Watanabe T, Takeuchi A, et al. Cardiovascular events and death in Japanese patients with chronic kidney disease. Kidney Int 2017;91:227–234.
25. Orlandi PF, Huang J, Fukagawa M, et al. A collaborative, individual-level analysis compared longitudinal outcomes across the International Network of Chronic Kidney Disease (iNETCKD) cohorts. Kidney Int 2019;96:1217–1233.
26. Ryu H, Kim J, Kang E, et al. Incidence of cardiovascular events and mortality in Korean patients with chronic kidney disease. Sci Rep 2021;11:1131.

Article information Continued

Figure 1.

Flow diagram of patient enrollment.

KNOW-CKD, KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease.

Figure 2.

Bland-Altman plot of differences in eGFR among the three equations.

(A) Difference between 2009 eGFRcr and 2021 eGFRcr equation. (B) Difference between 2009 eGFRcr and 2021 eGFRcr-cysC equation. (C) Difference in eGFR between 2021 eGFRcr and 2021 eGFRcr-cysC equation.

eGFR, estimated glomerular filtration rate; eGFRcr, creatinine-based eGFR; eGFRcr-cysC, creatinine-cystatin C–based eGFR; SD, standard deviation.

Figure 3.

Receiver operating characteristic curve analysis for CVE (A) and the composite of all-cause mortality and CVE predictability (B).

AUC, area under the curve; CVE, cardiovascular event.

Table 1.

Baseline characteristics of study population according to CVE

Characteristic Total CVE (+) group CVE (–) group p-value
No. of patients 2,207 200 2,007
Age (yr) 55.0 (45.0–63.0) 61.0 (55.0–67.5) 54.0 (44.0–63.0) <0.001
Male sex 1,353 (61.3) 145 (72.5) 1,208 (60.2) 0.001
BMI (kg/m2) 24.4 (22.3–26.5) 24.4 (22.6–26.0) 24.4 (22.3–26.5) 0.65
Hb (g/dL) 12.8 (11.3–14.3) 12.5 (10.8–14.1) 12.8 (11.3–14.4) 0.09
BUN (mg/dL) 24.0 (17.0–35.0) 27.6 (19.9–35.5) 23.6 (17.0–35.0) 0.01
Creatinine (mg/dL) 1.5 (1.1–2.2) 1.7 (1.2–2.3) 1.5 (1.0–2.2) 0.02
Cystatin C (mg/dL) 1.5 (1.1–2.2) 1.7 (1.3–2.2) 1.5 (1.0–2.2) 0.003
eGFRa (mL/min/1.73 m2) 46.2 (28.3–73.0) 41.8 (28.2–60.3) 47.4 (28.4–75.0) 0.006
Sodium (mEq/L) 143.0 (140.0–148.0) 142.0 (139.0–148.0) 143.0 (140.0–148.0) 0.046
Potassium (mEq/L) 6.1 (5.1–9.0) 6.2 (5.3–8.6) 6.1 (5.1–9.0) 0.95
Calcium (mg/dL) 9.4 (9.1–9.8) 9.3 (9.0–9.6) 9.5 (9.1–9.8) 0.004
Phosphorus (mg/dL) 5.0 (4.2–6.3) 4.9 (4.1–6.2) 5.1 (4.2–6.3) 0.29
Uric acid (mg/dL) 7.0 (5.8–8.3) 7.2 (5.9–8.5) 7.0 (5.8–8.3) 0.73
PTH (pg/dL) 51.0 (34.8–79.8) 49.4 (33.2–76.4) 51.1 (35.0–79.9) 0.64
hs-CRP (mg/dL) 0.6 (0.2–1.7) 0.8 (0.3–1.8) 0.6 (0.2–1.7) 0.03
Troponin T (ng/mL) 0.01 (0.01–0.02) 0.02 (0.01–0.03) 0.01 (0.01–0.02) <0.001
Fasting glucose (mg/dL) 100.0 (92.0–114.5) 107.0 (93.0–129.5) 99.0 (92.0–113.0) 0.001
HbA1C (%) 6.4 (5.7–7.5) 6.9 (6.1–7.8) 6.4 (5.7–7.3) <0.001
Total cholesterol (mg/dL) 171.0 (146.0–198.0) 161.0 (135.0–191.0) 171.0 (147.0–199.0) 0.002
HDL-C (mg/dL) 47.0 (38.0–58.0) 44.0 (36.0–52.5) 47.0 (38.0–58.0) 0.003
LDL-C (mg/dL) 93.0 (73.0–116.0) 88.0 (70.0–112.0) 94.0 (74.0–116.0) 0.02
Systolic BP (mmHg) 127.0 (118.0–137.0) 128.0 (116.0–139.0) 127.0 (118.0–136.0) 0.34
Diastolic BP (mmHg) 77.0 (70.0–84.0) 76.0 (69.0–82.0) 77.0 (70.0–84.0) 0.03
MAP (mmHg) 93.0 (86.0–101.0) 93.0 (85.0–100.0) 93.0 (87.0–101.0) 0.32
Ejection fraction (%) 64.0 (60.0–68.0) 63.0 (58.9–67.0) 64.0 (60.1–68.0) 0.001
Comorbidity
 Hypertension 2,114 (95.8) 197 (98.5) 1,917 (95.5) 0.07
 Diabetes mellitus 741 (33.6) 105 (52.5) 636 (31.7) <0.001
 Coronary artery disease 133 (6.0) 47 (23.5) 86 (4.3) <0.001
Smoking statusb 0.02
 Current 347 (15.8) 38 (19.1) 309 (15.4)
 Former 676 (30.7) 73 (36.7) 603 (30.1)
 Never 1,179 (53.5) 88 (44.2) 1,091 (54.5)

Data are expressed as median (interquartile range) or number (%).

BMI, body mass index; BP, blood pressure; BUN, blood urea nitrogen; CVE, cardiovascular event; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; hs-CRP, high-sensitivity C-reactive protein; Hb, hemoglobin; LDL-C, low-density lipoprotein cholesterol; MAP, mean arterial pressure; PTH, parathyroid hormone.

a

2009 eGFRcreatinine equation.

b

Patients with missing information on smoking status were excluded.

Table 2.

Comparison of average eGFR, CVE and composite of all-cause mortality and CVE prevalence according to CKD stages in each eGFR equation

Variable 2009 eGFRcr equation
2021 eGFRcr equation
2021 eGFRcr-cysC equation
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Total p-value Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Total p-value Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Total p-value
No. of patients 355 420 827 469 136 2,207 397 462 819 415 114 2,207 397 462 819 415 114 2,207
eGFR (mL/min/1.73 m2) 107.0±11.0 73.1±8.6 43.9±8.7 22.8±4.4 11.6±2.4 53.1±30.7 <0.001 108.7±10.5 73.1±8.8 43.8±8.7 22.8±4.3 11.9±2.4 56.0±31.5 <0.001 108.7±10.5 73.1±8.8 43.8±8.7 22.8±4.3 11.9±2.4 56.0±31.5 <0.001
CVE 12 (3.4) 39 (9.3) 88 (10.6) 53 (11.3) 8 (5.9) 200 (9.1) <0.001 12 (3.0) 46 (10.0) 96 (11.7) 38 (9.2) 8 (7.0) 200 (9.1) <0.001 12 (3.0) 46 (10.0) 96 (11.7) 38 (9.2) 8 (7.0) 200 (9.1) <0.001
All-cause mortality and CVE 14 (3.9) 50 (11.9) 124 (15.0) 100 (21.3) 26 (19.1) 314 (14.2) <0.001 15 (3.8) 61 (13.2) 133 (16.2) 82 (19.8) 23 (20.2) 314 (14.2) <0.001 15 (3.8) 61 (13.2) 133 (16.2) 82 (19.8) 23 (20.2) 314 (14.2) <0.001

Data are expressed as number only, mean ± standard deviation, or number (%).

CKD, chronic kidney disease; CVE, cardiovascular event; eGFR, estimated glomerular filtration rate; eGFRcr, creatinine-based eGFR; eGFRcr-cysC, creatinine-cystatin C–based eGFR.

Table 3.

Predictive value of each eGFR to CVE and composite of all-cause mortality and CVE

Variable 2009 eGFRcr (+10) p-value 2021 eGFRcr (+10) p-value 2021 eGFRcr-cysC (+10) p-value
CVE
 Unadjusted 0.90 (0.86–0.95) <0.01 0.91 (0.87–0.96) <0.01 0.90 (0.85–0.95) <0.01
 Sex-age 0.98 (0.92–1.04) 0.46 0.98 (0.93–1.04) 0.48 0.96 (0.91–1.02) 0.15
 Multivariablea 0.99 (0.91–1.07) 0.75 0.99 (0.91–1.07) 0.73 0.97 (0.90–1.04) 0.41
All-cause mortality and CVE
 Unadjusted 0.84 (0.80–0.88) <0.01 0.85 (0.81–0.89) <0.01 0.84 (0.80–0.87) <0.01
 Sex-age 0.90 (0.86–0.95) <0.01 0.91 (0.87–0.95) <0.01 0.89 (0.85–0.94) <0.01
 Multivariablea 0.94 (0.89–0.99) 0.02 0.94 (0.90–0.99) 0.02 0.92 (0.87–0.97) <0.01

Data are expressed as hazard ratio (95% confidence interval).

CVE, cardiovascular event; eGFR, estimated glomerular filtration rate; eGFRcr, creatinine-based eGFR; eGFRcr-cysC, creatinine-cystatin C–based eGFR.

a

Multivariable model was adjusted for age, sex, diabetes mellitus, smoking, body mass index, mean arterial pressure, low-density lipoprotein cholesterol, ejection fraction, and proteinuria.

Table 4.

NRI for comparison of predictability according to eGFR equations

Variable NRI (%) 95% CI p-value
CVE
 2009 eGFRcr vs. 2021 eGFRcr 0.013 –0.002 to 0.028 0.09
 2009 eGFRcr vs. 2021 eGFRcr-cysC –0.001 –0.031 to 0.029 0.94
 2021 eGFRcr vs. 2021 eGFRcr-cysC –0.015 –0.048 to 0.018 0.38
All-cause mortality and CVE
 2009 eGFRcr vs. 2021 eGFRcr –0.019 –0.039 to 0.000 0.06
 2009 eGFRcr vs. 2021 eGFRcr-cysC –0.002 –0.023 to 0.018 0.82
 2021 eGFRcr vs. 2021 eGFRcr-cysC 0.017 –0.004 to 0.039 0.11

CI, confidence interval; CVE, cardiovascular event; eGFR, estimated glomerular filtration rate; eGFRcr, creatinine-based eGFR; eGFRcr-cysC, creatinine-cystatin C–based eGFR; NRI, net reclassification improvement.