Association between the difference in estimated GFR based on cystatin C versus creatinine in coronary artery diseases

Abstract

Background

The difference in estimated glomerular filtration rate (eGFR) derived from creatinine and cystatin C (eGFR_diff) has been noticed recently and the relationship with poor cardiovascular prognosis has been proven. However, primary prevention of the risk of coronary artery disease (CAD) is equally important but there is a lack of studies specifically investigating this implication.

Methods

This prospective cohort study utilized data from the UK Biobank, including 437,536 participants without CAD at baseline. The primary outcome was defined as CAD. The eGFR_diff was calculated by subtracting creatinine-based eGFR from cystatin C-based eGFR. Participants were then categorized into a negative, intermediate range, and positive group based on thresholds of −15 mL/min/1.73 m² and 15 mL/min/1.73 m². Cox proportional risk models were used to evaluate the associations of eGFR_diff with CAD and the relationship among different genetic risks of CAD.

Results

During a median follow-up of 13.8 years, CAD occurred in 36,797 participants. In the fully adjusted model, compared to midrange eGFR_diff, participants with a positive eGFR_diff had a lower risk of CAD (HR 0.717, 95%CI 0.675-0.762), while with a negative eGFR_diff had a higher risk (HR 1.433, 95%CI 1.399-1.468). When eGFR_diff was treated as a continuous variable, a statistically significant trend toward a lower risk of CAD as eGFR_diff increased (HR 0.982, 95% CI 0.981-0.982). Moreover, this relationship is independent of genetic susceptibility.

Conclusions

eGFR_diff was associated with CAD risk, where a high eGFR_diff corresponded to a decreased likelihood of CAD onset no matter genetic susceptibility.

Introduction

As a global health problem and the leading cause of mortality, it is undoubted that coronary artery disease (CAD) is gaining more and more attention [1,2]. Approximately 18.2 million adults in the United States over age 20 suffer from CAD, leading to nearly 365,000 deaths annually, while globally, CAD affects around 126 million individuals, accounting for 610,000 deaths (estimated 1 in 4 deaths) [3–5]. Chronic kidney disease (CKD) is an independent risk factor for coronary artery disease [6,7]. A decline in estimated glomerular filtration rate (eGFR) signals impaired kidney function, and each reduction is closely tied to an increased risk of CAD, whereas albuminuria and high blood urea nitrogen, also raise the likelihood of cardiovascular events [8,9].

In clinical practice, eGFR is usually calculated by serum creatinine, named eGFR_cr (creatinine-based eGFR), and has been reported may be a strong predictor of CAD and cardiovascular outcomes, however, it can be affected by muscle mass, age, and sex, with great individual differences [10–12]. The use of cystatin C-based eGFR (eGFR_cys) for monitoring kidney disease has attracted attention since 1994 [13]. When compared with serum creatinine, cystatin C has demonstrated higher sensitivity to small changes in GFR and higher diagnostic accuracy for reduced GFR [14–16]. Estimates of GFR based solely on serum cystatin C had more practical implications for assessing kidney function in diabetic patients and those recovering from stage 3 acute kidney injury, with higher sensitivity and specificity for detecting early CKD in hypertensive patients [17–19]. It has also shown better predictive effect in specific patient populations such as kidney transplantation, chemotherapy, cirrhosis, and autoimmune diseases. Interestingly, eGFR_cys and eGFR_cr values differ substantially even within the same individual, in one-third of participants of more than 15 mL/min/1.73 m², thus coming up with the variable eGFR_diff, defined as the difference between cystatin C-based and creatinine-based eGFR [20].

In patients with discordances between creatinine and cystatin C-based estimations, combining both biomarkers could improve accuracy compared to using either one alone, which applies to individuals with cardiovascular diseases, heart failure (HF), diabetes mellitus (DM), liver disease, or cancer [16,21]. Cystatin C testing has been implemented in Sweden for over a decade, and they found that many patients have a lower eGFR_cys than eGFR_cr, who have a higher risk of various adverse outcomes [22]. Further, a negative eGFR_diff, regardless of the level of renal function, is associated with an increased risk of frailty, cardiovascular disease, end-stage kidney disease, and all-cause mortality [20,23]. The relationship between eGFR_diff and cardiovascular diseases like HF or atrial fibrillation (AF) has already been explored [24,25], however, research specifically linking eGFR_diff to CAD is still lacking. Thus, we analyzed data from the UK Biobank (UKB) to estimate whether eGFR_diff is independently associated with CAD.

Methods

Study design and participants

The UKB data utilized in this study constitute a substantial prospective cohort comprising more than 500,000 participants aged 37 to 73 years, recruited from 22 assessment centers across the UK between 2006 and 2010. At the initial study visit, participants underwent a nursing interview covering medical history, substance use, sociodemographic factors, and lifestyle, along with a physical assessment involving blood and urine sample testing [26]. Our study involved participants with baseline serum cystatin C and creatinine measurements, excluding those diagnosed with CAD at baseline. We assessed incident CAD using self-reports from baseline questionnaires, primary care records, hospital inpatient records, death registry data, and hospital operation/procedure codes. Additionally, participants without a polygenic risk score (PRS) for CAD were excluded. Ultimately, 437,536 participants were enrolled in this analysis. Detailed inclusion and exclusion criteria are presented in Figure 1.

Figure 1.

Study flow charts.

Abbreviations: CAD, coronary artery disease; Cys, cystatin C; PRS, polygenic risk scores Scr, serum creatinine.

The database contains more detailed information online (www.ukbiobank.ac.uk). It undergoes regular updates, with multiple subsequent surveys conducted. For the detailed study protocol, the web link can be attached in Supplementary Method S1. All participants provided written informed consent, and the study received approval from the North West Multicenter Research Ethics Committee (REC reference for UKB 11/NW/0382).

Measure of exposure

Serum cystatin C and creatinine levels were assessed using serum samples obtained at the study’s onset. Serum cystatin C levels were determined via a latex-enhanced immunoturbidimetric assay performed on the Siemens Advia 1800 system, demonstrating an inter-assay coefficient of variation of 1.1%. An enzyme-based assay was conducted using the Beckman Coulter AU5800 system for serum creatinine levels, showing a coefficient of variation of 2.0%. The eGFR_cr was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2021 equations without race coefficients [27]. The eGFR_cys was calculated using the CKD-EPI 2012 equations without differentiating between races [28]. We used race-free eGFR_diff as the primary exposure defined as eGFR_cys minus race-free eGFR_cr. Subjects were categorized into 3 groups: negative (<−15 mL/min/1.73 m²), intermediate range (−15 –15 mL/min/1.73 m²), and positive (≥15 mL/min/1.73 m²). In addition, eGFR_diff (every 10 mL/min/1.73 m² increase) was analyzed as a continuous variable.

Definitions of outcome

The primary outcome of this study was CAD, defined as the first occurrence of the International Classification of Diseases, Tenth Edition (ICD-10) I20-25 in any hospitalization data and cause of death registry records from eight English hospitalization data, Scottish morbidity record data from Scotland, and patient data databases from Wales.

The follow-up cutoff was recorded as June 1, 2024, and the difference between the date of enrollment in the UKB study and the date of the first report of CAD or the follow-up cutoff was the time of onset or follow-up. Patients with baseline CAD were recorded if the time of the first report of CAD was earlier than the subject’s enrollment in the UKB study. In addition, in the subgroup analyses, a history of CKD was defined as an ICD N17, N18, and N19 with the above information reported.

Definitions of covariates

The covariates that were included at baseline were age, sex, ethnicity, body mass index (BMI), Townsend-derived index (TDI), employment statement, college/university degree, smoking, alcohol use, DM, hyperlipidemia (HLP), systolic blood pressure, diastolic blood pressure, healthy dietary score (HDS), sleep duration, C-reactive protein, cholesterol, creatinine, cystatin C, glucose, low-density lipoprotein direct, triglycerides, urate, urea, vitamin D.

The TDI assesses material deprivation within the population, where higher scores indicate increased poverty levels. Regarding alcohol use, "no" encompassed individuals who reported "never drinking" or consuming alcohol only on "special occasions", while "yes" included frequencies such as "Daily or almost daily", "Three or four times a week", "Once or twice a week", and "One to three times a month". Daily sleep duration was self-reported by patients on the questionnaire. HDS criteria were based on consuming a median of at least 4 tablespoons of vegetables, 3 pieces of fruit, at least two servings of fish weekly, no more than two servings of unprocessed red meat weekly, and no more than two servings of processed meat weekly. Each favorable dietary factor contributed one point toward a cumulative score of 5 [29]. Genetic risk groups were categorized based on PRS tertiles provided by UKB for CAD. Additional details on covariates are available in Supplementary Table S1.

Statistical analyses

Baseline characteristics were described in groups according to the three categorical situations of eGFR_diff. For missing data, we used median substitution for variables with a missing rate of <5%, and for variables with a missing rate of ≥5%, we labeled the missing data as a separate category as “unknown”. The missing data for each variable can be obtained in Supplementary Table S2.

Baseline characteristics of continuous variables with normal distribution were expressed as mean ± standard error, and categorical variables were expressed as frequency (percentage). Two-sample t-tests or analysis of variance were used to analyze statistical differences for continuous variables, and χ² tests were used to analyze statistical differences for categorical variables.

The cumulative risk of CAD among the 3 groups was demonstrated by Kaplan-Meier curves with the log-rank test. Then, Cox proportional risk models were used to calculate hazard ratios (HR), and 95% confidence intervals (CI) for corrected CAD incidence. Model 1 was adjusted for demographic factors and lifestyle factors, like age, sex, TDI, ethnicity, BMI, college/university, alcohol, and smoking; model 2 was further adjusted for laboratory parameters, specifically the effects of SBP, DBP, HLP, HDS, DM, sleep duration, C-reactive protein, creatinine, LDL direct, eGFR_cr, cholesterol, urate, urea, triglycerides, and vitamin D. In addition, we applied model 2 to stratified analyses in the high genetic risk group, medium genetic risk group, and low genetic risk group separated by PRS.

As sensitivity analyses, after excluding the baseline cancer population, exploration of the relationship between eGFR_diff and CAD incidence risk was continued. As a further sensitivity analysis, the statistical analysis process described above was performed again for samples excluding non-white. Subgroup analyses were conducted by age group (<65 years or ≥65 years), sex, obesity (BMI ≥30 kg/m²; yes or no), DM, HLP, and kidney disease history.

The statistical analyses were conducted using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria), and a significance level of p < 0.05 was set for two-sided testing.

Results

Baseline characteristics

A total of 437,536 participants were included in the study, including 242,890 (55.5%) females. The participants were divided into three groups according to the eGFR_diff, which were the negative group representing less than −15 mL/min/1.73 m² including 150,662 individuals, the midrange group presenting −15 to 15 mL/min/1.73 m² including 262,546 individuals, and the positive group representing more than 15 mL/min/1.73 m² including 24,328 individuals. Participants in the positive group were more likely to be nonwhite, have an older age, lower BMI, less smoking, have higher HDS, as well as drink more alcohol, and have less HLP, DM, and hypertension. Table 1 provides a comprehensive overview of the baseline characteristics of the participants, categorized according to the aforementioned stratification.

Table 1.

Baseline characteristics grouped by eGFR_diff.

	Negative	Midrange	Positive	p
n	150,662	262,546	24,328
Age (years)	55.32 (8.21)	57.13 (7.96)	57.48 (7.92)	<0.001
Male (%)	52,911 (35.1)	131,839 (50.2)	9896 (40.7)	<0.001
BMI (kg/m2)	28.09 (5.88)	26.79 (4.45)	26.07 (4.01)	<0.001
TDI	−1.09 (3.17)	−1.51 (2.99)	−1.38 (3.10)	<0.001
Employment (%)	79,722 (52.9)	17,8557 (68.0)	19,420 (79.8)	<0.001
College/university (%)	42,029 (27.9)	93,406 (35.6)	9340 (38.4)	<0.001
Ethnicity (%)				<0.001
White	143,135 (95.0)	250,000 (95.2)	21,571 (88.7)
Non-White	7527 (5.0)	12,546 (4.8)	2757 (11.3)
Smoking (%)				<0.001
Current	23,136 (15.4)	21,010 (8.0)	1290 (5.3)
Never	76,942 (51.0)	15,1912 (57.6)	15,583 (63.8)
Previous	50,584 (33.6)	89,624 (34.1)	7455 (30.6)
Alcohol (%)				<0.001
Current	134,693 (89.4)	246,167 (93.8)	22,983 (94.5)
Never	8981 (6.0)	8988 (3.4)	804 (3.3)
Previous	6988 (4.6)	7391 (2.8)	541 (2.2)
Sleep duration (hours)	7.15 (1.17)	7.16 (1.05)	7.14 (1.02)	0.021
SBP (mmHg)	141.51 (19.46)	138.70 (18.74)	134.55 (18.02)	<0.001
DBP (mmHg)	82.79 (10.40)	82.16 (10.26)	80.43 (10.13)	<0.001
HDS	2.47 (1.09)	2.51 (1.10)	2.54 (1.10)	<0.001
DM (%)	9,378 (6.2)	9,932 (3.8)	595 (2.4)	<0.001
HLP (%)	18,560 (12.3)	26,854(10.2)	1762 (7.2)	<0.001
eGFRcr (mL/min/1.73 m2)	108.75 (18.66)	93.73 (13.30)	78.29 (10.75)	<0.001
CRP (mg/L)	3.36 (5.15)	2.17 (3.71)	1.72 (3.23)	<0.001
Cholesterol (mmol/L)	5.82(1.17)	5.73(1.09)	5.61(1.06)	<0.001
Creatinine (µmol/L)	64.65 (13.25)	74.63 (17.72)	85.71 (13.86)	<0.001
LDL direct (mmol/L)	3.65 (0.88)	3.58 (0.84)	3.45 (0.81)	<0.001
Triglycerides (mmol/L)	1.88 (1.07)	1.68 (0.99)	1.50 (0.91)	<0.001
Urate (µmol/L)	303.80 (81.71)	309.20 (78.45)	304.73 (76.62)	<0.001
Urea (mmol/L)	5.24 (1.31)	5.42 (1.38)	5.63 (1.31)	<0.001
Vitamin D (nmol/L)	46.47 (20.26)	49.58 (20.59)	50.56 (21.37)	<0.001

Abbreviations: BMI: body mass index; CRP: C-reactive protein; DBP: diastolic blood pressure; DM: diabetes mellitus; eGFR_cr: creatinine-based eGFR; HDS: healthy dietary score; HLP: hyperlipidemia; SBP: systolic blood pressure; TC: Triglyceride; TDI: Townsend-derived index. Values expressed with % or mean ± SD, as applicable.

Association between eGFR_diff and CAD

During a median follow-up time of 13.8 years, a total of 36,797 cases of CAD occurred in the enrolled participants, including 15,982 cases in the negative group, 19,630 cases in the midrange group, and 1,185 cases in the positive group.

We found that higher eGFR_diff was associated with a lower risk of CAD in all multivariable analysis. To further validate this correlation, we observed a statistically significant correlation between the positive group compared to the midrange group and a reduction in the risk of CAD in the fully adjusted model (HR 0.717, 95%CI 0.675-0.762). And the HR for negative eGFR_diff compared to midrange eGFR_diff was 1.433 (95%CI 1.399-1.468). Table 2 shows in detail the HRs of the groups of eGFR_diff values with the risk of CAD under different adjusted models. When eGFR_diff was treated as a continuous variable, the HR (95% CI) per 1 mL/min/1.73 m² increase in eGFR_diff was 0.982 (0.981-0.982).

Table 2.

Cox analysis of the relationship between eGFR_diff and CAD.

	Crude Model		Model 1		Model 2
eGFRdiff	HR (95% CI)	p value	HR (95% CI)	p value	HR (95% CI)	p value
Categorical value
Negative (<-15 mL/min/1.73 m2)	1.444 (1.415,1.475)	<0.001	1.435 (1.404,1.467)	<0.001	1.433 (1.399,1.468)	<0.001
Midrange (-15 to 15 mL/min/1.73 m2)	1.00 (Reference)	–	1.00 (Reference)	–	1.00 (Reference)	–
Positive (≥15 mL/min/1.73 m2)	0.643 (0.606,0.681)	<0.001	0.749 (0.706,0.795)	<0.001	0.717 (0.675,0.762)	<0.001
Continuous
Per 1 mL/min/1.73 m2 increase	0.992 (0.992,0.992)	<0.001	0.990 (0.99,0.99)	<0.001	0.982 (0.981,0.982)	<0.001

Model 1: adjusted for age, sex, TDI, ethnicity, BMI, college/university, alcohol, and smoking.

Model 2: adjusted further for the effect of SBP, DBP, HLP, HDS, DM, sleep duration, C-reactive protein, creatinine, LDL direct, eGFR_cr, cholesterol, urate, urea, triglycerides, and vitamin D.

Abbreviations: BMI, body mass index; CI, confidence intervals; DBP, diastolic blood pressure; DM, diabetes mellitus; eGFR_cr, creatinine-based eGFR; HDS, healthy dietary score; HLP, hyperlipidemia; HR, hazard ratio; SBP, systolic blood pressure; TDI, Townsend-derived index.

In addition, the Kaplan-Meier curves of the crude model yielded results consistent with the above findings. Throughout the follow-up period, the risk of CAD was consistently lower in the positive group compared to the midrange group, and consistently higher in the negative group (Figure 2).

Figure 2.

Kaplan-Meier plots for the risk of CAD grouped by eGFR_diff.

Abbreviations: CAD, Coronary artery disease.

Furthermore, the relationship between eGFR_diff and the risk of incident CAD has also been observed in various strata of genetic susceptibility for CAD in Figure 3, with the risk of CAD in the positive group reduced by 35% (HR 0.648, 95% CI 0.571-0.736), 31% (HR 0.687, 95% CI 0.618-0.764), and 23% (HR 0.770, 95% CI 0.704-0.841) in populations with low, intermediate, and high genetic risk, respectively, compared with those in midrange group. As for the patients in the Negative group, they showed a higher risk of CAD in all the different genetic risk groups (low genetic risk: HR 1.518, 95% CI 1.445-1.594; intermediate genetic risk: HR 1.449, 95% CI 1.390-1.511; high genetic risk: HR 1.399, 95% CI 1.349-1.451).

Figure 3.

Association of eGFR_diff with the risk of CAD among different levels of PRS.

Abbreviation: CI, confidence intervals.

Subgroups and sensitivity analyses

Figure 4 displayed the association between eGFR_diff and CAD among subgroups. In terms of the incidence of CAD, a significant interaction was noted in gender (p for interaction < 0.001), HLP statuses (p for interaction < 0.001), and obesity statuses (p for interaction < 0.001), but no significant interaction was observed across age, DM statuses, and history of kidney disease.

Figure 4.

Forest plots of the relationship between eGFR_diff and CAD in different subgroups.

Abbreviations: CI, confidence intervals; DM, diabetes mellitus; HLP, hyperlipidemia.

Sensitivity analyses were performed by methods that excluded participants with cancer and non-White. Table 3 demonstrates that across sensitivity analyses, patients in group negative exhibited a higher risk of CAD, and group positive exhibited a lower risk of CAD compared to group midrange, demonstrating the robustness of the main findings.

Table 3.

Sensitivity analyses for the association between eGFR_diff and CAD.

	Crude Model		Model 1		Model 2
eGFRdiff	HR (95% CI)	p value	HR (95% CI)	p value	HR (95% CI)	p value
Sensitivity analyses for excluding cancer
Categorical value
Negative	1.459 (1.428,1.492)	<0.001	1.450 (1.417,1.484)	<0.001	1.444 (1.408,1.480)	<0.001
Midrange	1.00 (Reference)	–	1.00 (Reference)	–	1.00 (Reference)	–
Positive	0.643 (0.605,0.684)	<0.001	0.752 (0.708,0.800)	<0.001	0.720 (0.676,0.767)	<0.001
Continuous
Per 1 mL/min/1.73 m2 increase	0.989 (0.988,0.989)	<0.001	0.988 (0.987,0.988)	<0.001	0.981 (0.980,0.982)	<0.001
Sensitivity analyses for excluding non-White
Categorical value
Negative	1.445 (1.415,1.477)	<0.001	1.438 (1.406,1.471)	<0.001	1.435 (1.400,1.470)	<0.001
Midrange	1.00 (Reference)	–	1.00 (Reference)	–	1.00 (Reference)	–
Positive	0.667 (0.627,0.709)	<0.001	0.772 (0.726,0.821)	<0.001	0.736 (0.691,0.783)	<0.001
Continuous
Per 1 mL/min/1.73 m2 increase	0.992 (0.992,0.992)	<0.001	0.990 (0.990,0.990)	<0.001	0.983 (0.982,0.985)	<0.001

Abbreviations as in Table 2.

Discussion

This prospective cohort study used the UKB database to explore the relationship between eGFR_diff and CAD. We observed that participants in the eGFR_diff-positive group had a lower risk of CAD, whereas those in the eGFR_diff-negative group had a higher risk of new-onset CAD, and when eGFR_diff was used as a continuous variable, there was a statistically significant trend toward a lower risk of CAD as eGFR_diff increased. Furthermore, this relationship was still stable no matter adjusting for covariates or PRS.

Some pathophysiologic mechanisms can explain the relationship between eGFR_diff and CAD risk. First, eGFR_diff can provide a more refined assessment of kidney function. Recent studies suggested that cystatin C is an earlier indicator of mild renal failure and is better at detecting changes in GFR in critically ill patients [30,31]. Among 41 elderly patients with no evidence of renal disease, 11 had GFRs measured by inulin clearance below the 95% reference interval, all of which had increased CysC but normal SCr [32]. These may be described as shrunken pore syndrome. This phenomenon was described in studies about a decade ago and has been suggested to be associated with poor prognosis in the elderly, and high mortality in patients with cardiovascular disease, and the assessment variable used in some of these studies was eGFR_cys/eGFR_cr-ratio [33–37]. As the difference between eGFR_cys and eGFR_cr, eGFR_diff has also been used in recent studies as a way to assess shrunken pore syndrome and has the same ability to estimate mortality [38–40]. Inflammation and oxidative stress induced by hyperglycemia could cause the contraction of capillary endothelial cells, potentially leading to a reduction in filtration pore size [41]. In patients with CAD, inflammatory and oxidative stress mechanisms may also affect the structure of glomerular capillaries [42]. Endothelial dysfunction plays an important role in the early stages of atherosclerosis, facilitating both the initiation and progression of plaque formation [43]. It may involve the glomerular vascular endothelium, leading to alterations in infiltration. Given that eGFR_cys is more sensitive to detecting kidney damage than eGFR_cr, a negative eGFR_diff may indicate subtle kidney impairment that might be missed by eGFR_cr alone. This makes it valuable for identifying the hidden renal impairment and the associated CAD risks.

Furthermore, creatinine levels are affected by factors such as age, gender, and muscle mass [10]. In populations like the elderly, children, and frail individuals with lower muscle mass, baseline serum creatinine levels tend to be naturally lower due to decreased creatinine production [44]. As a result, even if eGFR_cr values fall within a normal range, actual kidney function may be impaired. Importantly, these populations, characterized by reduced muscle mass, are also predisposed to higher risk for cardiovascular diseases [45]. However, in such scenarios, eGFR_diff remains at a lower level in that cystatin C is not influenced by muscle mass [10], allowing for a more accurate assessment of true kidney function. Additionally, when eGFR_cr values are within the normal range in these groups, with an overestimation of renal function, and patients are prescribed medications such as anticoagulants and antitumor drugs, which are primarily cleared through the kidneys, failure to accurately assess kidney function and reduce the drug dosage may result in levels that exceed the kidneys’ capacity to clear them, leading to toxic side effects, potentially harming the cardiovascular system.

Moreover, eGFR_diff is not only related to renal function but is also associated with cardiovascular risk factors, such as endothelial dysfunction, inflammation, oxidative stress, and coronary artery calcification [46–48]. Recent studies have revealed that certain serum proteins, such as interleukin-6, monocyte chemotactic protein-3, tumor necrosis factor receptor, and osteoprotegerin, would accumulate in patients with shrunken pore syndrome [49]. These proteins, as important inflammatory factors, are linked to atherosclerosis occurrence and accelerate atherosclerosis [47]. When there is a large positive difference in eGFR_cr minus eGFR_cys, patients had more baseline coronary artery calcification and faster accelerated progression, which may also be related to the accumulation of atherosclerosis-promoting proteins [48]. Negative eGFR_diff indicates reduced selectivity for glomerular filtration and endothelial dysfunction. Studies have shown that eGFR_cys is associated with endothelial-dependent vasodilation also in persons with normal kidney function [46]. The development of atherosclerosis is a part of a general vascular disorder. Endothelial dysfunction can lead to early vascular injury, contributing to the development of atherosclerotic cardiovascular diseases [50]. In short, negative eGFR_diff often uncovers hidden kidney dysfunction and related cardiovascular risks, making it a potential risk factor for identifying patients at elevated risk of cardiovascular events.

Previous studies have described similar conclusions. Cohort analysis of SPRINT revealed that the difference between eGFR_cys and eGFR_cr was associated with lower adverse events, including cardiovascular events, when positive [23], while it has not specifically analyzed specific cardiovascular events. The association of differences in estimated GFR by cystatin C versus creatinine with incident HF or AF has been reported [24,25], but there is no literature for CAD. Meanwhile, larger values obtained by subtracting eGFR_cys from eGFR_cr have also been found to be associated with major adverse cardiovascular events and coronary artery calcification progression, but the cohort is specific to patients with CKD [48]. Moreover, some studies confirmed that elevated plasma cystatin C is associated with both the occurrence and severity of CAD [51]. Some showed its predictive value in long-term all-cause and cardiovascular mortality in patients undergoing coronary angiography [52]. Considering that cystatin C may be affected by a variety of factors, our study started with the eGFR differences between eGFR_cys and eGFR_cr and specifically analyzed the correlation and predictive value of eGFR_diff for CAD in the general population, regardless of the prevalence of CKD.

Our study has several strengths. We included large sample sizes and had a long-term follow-up period. At the same time, all study participants performed standardized measurements of creatinine and cystatin C, and we also performed adjusted analyses for confounders. However, we also recognize some limitations of the study. Firstly, the data we included were all from the UKB, which may have led to bias in the selection of ‘healthy volunteers’ as participants were generally younger and healthier, and although the database included multiple ethnicities, it was still predominantly white, so extrapolation to other ethnicities or regions and at-risk populations should be done with caution thus affecting the broad applicability of the findings. Secondly, given the limitations of the database, we did not collect creatinine and cystatin C levels data for dynamic testing, and while admittedly changes in these data in relation to CAD would be stronger evidence, the baseline data also reflect a relationship with the onset of CAD in the basal state, which could likewise provide some clinical value. Third, we only found significant correlations in our study and interpreted them in the context of existing studies with possible mechanisms, but there is still the possibility of reverse causality in the study, which is a problem that is difficult to avoid in nonintervention cohorts, and thus the present study only suggests the potential significance and applied value of eGFR_diff, which needs to be further confirmed by subsequent large-scale randomized controlled trials and Mendelian randomization studies. Forth, while eGFR_diff may have shown promise in CAD risk assessment, its direct clinical application remains limited. Routine cystatin C testing and the integration of eGFR_diff into clinical practice have not been widely implemented globally. However, in Sweden, cystatin C-based eGFR assessment has been integrated for over a decade, providing real-world evidence of its clinical significance. Despite this, further studies are needed to focus on standardizing its application and validate its role in clinical decision-making.

Conclusion

In this large UKB-based retrospective cohort, we found a correlation between eGFR_diff and CAD, and the large difference between eGFR_cys and eGFR_cr conveys important information about the risk of CAD, which may be used as a clinical marker to predict and aid in the judgment of CAD risk. Future research is needed to further evaluate its predictive value in CAD or multiple diseases, and better establish disease prediction models.

Funding Statement

This work was supported by grants from CAMS Innovation Fund for Medical Sciences (No. 2021-I2M-1-008).

Ethics approval and consent to participate

Ethical approval for the UK Biobank study was granted by the North West Multi-center Research Ethics Committee (REC reference: 11/NW/03820). Prior to enrollment, written informed consent was obtained from all participants, affirming their understanding and agreement to the study procedures, and the research was conducted in strict adherence to the ethical guidelines set forth in the Declaration of Helsinki.

Consent for publication

All authors declare no conflict of interest for this contribution.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Authors’ contribution

Concept and design: Kefei Dou, Weihua Song, Zechen Liu, and Wangying Jiang. Acquisition of data: Zechen Liu. Formal analysis and interpretation of data: Zechen Liu. Statistical analysis: Zechen Liu. Drafting of the manuscript: Zechen Liu, Wangying Jiang, Yanjun Song, Weihua Song. Critical revision of the manuscript for important intellectual content: Kefei Dou and Weihua Song. Obtained funding: Kefei Dou. Administrative, technical, or material support: Kefei Dou. Supervision: Kefei Dou and Weihua Song. All authors share the responsibility of deciding whether or not to contribute to the study.

Data availability statement

Access to the UK Biobank data utilized in this study is available to researchers through the application process specified on the UK Biobank website, under application number 97155. (https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access).