Introduction
Chronic kidney disease (CKD) is a well-known risk factor for cardiovascular disease, with kidney function decline consistently linked to increased mortality and cardiovascular events [1]. An estimated glomerular filtration rate (eGFR) below 60 mL/min/1.73 m2 is commonly used to diagnose CKD, as this threshold is associated with significant adverse outcomes [2]. However, this fixed threshold does not account for the natural, age-related decline in eGFR seen in healthy older adults, often resulting in CKD diagnoses that may not accurately reflect disease progression in this age group [3–5]. Distinguishing physiological kidney aging from pathological CKD is essential to avoid overdiagnosis and unnecessary interventions in elderly populations.
Moreover, epidemiological evidence suggests that older adults with a reduced glomerular filtration rate (GFR) tend to have a more stable course than younger individuals with the same GFR levels [6]. As such, whether to apply the fixed GFR threshold (60 mL/min/1.73 m2) regardless of age as a definition of CKD has been debatable [7–9]. To improve the prognostication of older adults, some experts have proposed a lower threshold of 45 mL/min/1.73 m2 for diagnosing CKD in individuals over 65 years-old, as significant mortality risks appear to increase only below this level in older adults [10].
The question of whether age-adapted CKD criteria are suitable for high-risk groups, such as elderly individuals with diabetes mellitus (DM), remains unresolved. Considering the early occurrence of glomerular hyperfiltration in DM [11], eGFR levels affecting long-term prognosis may reasonably differ between older patients with and without DM. This is of great importance considering the high prevalence of DM in older adults and the frequent coexistence of CKD in patients with DM [12,13]. In particular, as nearly half of patients with CKD with eGFR between 45 and 60 mL/min/1.73 m2 do not exhibit signs of albuminuria among the older patients with DM [14], whether mild eGFR reduction without albuminuria represents a normal consequence of aging or an early indicator of diabetic CKD needs to be discerned. This study examines the prognostic impact of traditional eGFR categories on mortality and progression to end-stage kidney disease (ESKD) in older, normoalbuminuric patients with type 2 diabetes mellitus (T2DM). Our findings may inform CKD diagnostic criteria for elderly diabetic populations, supporting more precise risk stratification and improved clinical decision-making.
Methods
This was a retrospective cohort study, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected in a priori approval by the Institutional Review Board (IRB) of Samsung Medical Center (No. 2017-02-016). Owing to the retrospective nature of the study with no identifying data, the IRB waived the requirement for informed consent. STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines were used to ensure the reporting of this observational study [15].
Study population and data collection
We identified all older adult patients (aged ≥65 years) with T2DM who visited the outpatient diabetes center of a 2,000-bed tertiary referral hospital in Seoul, Korea between January 1, 2009 and December 31, 2009. T2DM was defined using the International Classification of Diseases 10th Revision (ICD-10) codes (E11 and E14), and the definition was tightened using the presence of at least three ICD-10 codes during outpatient visits. The exclusion criteria were as follows: patients who received renal replacement therapy including dialysis and kidney transplantation, with eGFR <15 mL/min/1.73 m2, were hospitalized within a year prior to study enrollment and did not have a serum creatinine or urine albumin-to-creatinine ratio (ACR) measurement. Finally, patients with a urine ACR of ≥30 μg/mgCr were excluded from the study. The study participants were followed up from their enrollment to death or until July 26, 2017, whichever occurred first.
After identifying older patients with T2DM, the following data were collected at the time of enrollment: body weight, height, systolic blood pressure (SBP), serum creatinine (standardized against isotope dilution mass spectrometry), urine ACR, hemoglobin A1c (HbA1c), low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, duration of DM, and medications including insulin, metformin, renin-angiotensin system (RAS) blockers including angiotensin-converting enzyme inhibitors (ACEI) and angiotensin receptor blockers (ARBs), and statins. Body mass index (BMI) was calculated by dividing weight in kilograms by the square of height in meters. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [16] and then was categorized into the following four categories: ≥90, 60–89, 45–59, and 15–44 mL/min/1.73 m2.
Comorbidity burden was assessed for each patient using the Charlson Comorbidity Index (CCI), which is defined using 19 major disease categories [17,18]. Cardiovascular disease was defined as cardiovascular events including myocardial infarction, cerebrovascular accident, and peripheral arterial disease.
Assessment of outcomes
The primary outcomes of interest were the composite events of all-cause mortality and progression to ESKD. In cases where death followed the development of ESKD, it was considered as a single event for composite outcomes. Data on the occurrence of death, irrespective of progression to ESKD, were collected based on the statistics of the Korean National Death Registry from the Ministry of the Interior and Safety of Korea. All deaths were linked and confirmed using center databases. ESKD was defined as the requirement for maintenance dialysis or kidney transplantation, which was identified by combining the center database and the Korean Society of Nephrology ESRD Registry.
Statistical analysis
Patient characteristics are described as medians and interquartile ranges (IQRs) for continuous variables and frequencies and proportions for categorical variables. They were compared according to baseline eGFR categories using the Jonckheere–Terpstra test or the Cochran–Armitage test.
The date of enrollment was on January 1, 2007, and the follow-up ended on July 26, 2017. The 7-year cumulative incidence of composite outcomes for each eGFR category is described. Survival analyses were performed using the log-rank test for all follow-up data.
The Cox proportional hazards model was used to investigate the association between the eGFR and outcomes after adjusting for multiple covariates. The final multivariate models were fitted using the backward stepwise method with adjustments for variables including sex, age, BMI, SBP, duration of DM, urine ACR, eGFR categories, HbA1c, LDL cholesterol, HDL cholesterol, CCI, and use of RAS blocker. The linearity assumption was verified for the continuous variables.
All statistical analyses were performed using IBM SPSS for Mac (version 29.0; IBM Corp.). Significance was set at p < 0.05.
Results
Among 3,065 older patients with DM with eGFR ≥15 mL/min/1.73 m2, 1,068 patients with ACR ≥30 μg/mgCr were excluded. Altogether, 1,997 patients were included in the analysis and followed up for a median of 8.1 years (IQR, 7.8–8.3 years) and a mean of 7.5 years (standard deviation, 1.5 years). The clinical characteristics of the study participants according to the eGFR categories are summarized in Table 1. Overall, 159 (8%), 1,411 (71%), 328 (16%), and 99 (5%) exhibited an eGFR of ≥90, 60–89, 45–59, and 15–44 mL/min/1.73 m2, respectively. HDL and LDL cholesterol concentrations were lower in patients with a lower eGFR than in those with a higher eGFR (p < 0.001 and p = 0.009, respectively). The prevalence of comorbid conditions and cardiovascular disease increased progressively with a lower eGFR (p < 0.001 for both). Patients with a lower eGFR used insulin more often than those with a higher eGFR (p = 0.007).
Cumulative incidence of mortality and end-stage kidney disease according to estimated glomerular filtration rate category
Fig. 1 illustrates the numbers and percentages of all-cause deaths and progression to ESKD during 7 years according to eGFR categories. There was no loss to follow-up cases during the 7-year follow-up period. Overall, mortality was higher than incident ESKD cases. Compared to an eGFR of 60–89 mL/min/1.73 m2, the proportion of deaths increased for those with an eGFR of 45–59 mL/min/1.73 m2 (p < 0.001), and the proportion of ESKD increased for those with an eGFR of 45 mL/min/1.73 m2 below (p = 0.001). The 7-year mortality was 10.7% for eGFR ≥90 mL/min/1.73 m2, 8.2% for 60–89 mL/min/1.73 m2, 14.3% for 45–59 mL/min/1.73 m2, and 25.3% for 15–44 mL/min/1.73 m2, respectively (p < 0.001 for trend) (Fig. 1).
The cumulative incidence plot for the composite outcome according to the eGFR categories is illustrated in Fig. 2. Composite outcome rates were significantly higher in patients with eGFRs of 45–59 and 15–44 mL/min/1.73 m2 than in those with an eGFR of 60–89 mL/min/1.73 m2 (p < 0.001 for both).
Estimated glomerular filtration rate categories and risks of progression to end-stage kidney disease or mortality
We conducted a multivariate analysis to assess the independent effects of eGFR categories on composite events (Table 2). After adjusting for sex, age, BMI, SBP, duration of DM, LDL and HDL cholesterol levels, urine ACR, HbA1c, CCI, DM duration, and ACEI or ARB use, patients with an eGFR of 45–59 mL/min/1.73 m2 demonstrated a 1.30-fold higher risk (95% confidence interval, 1.01–1.76; p = 0.04) compared to those with an eGFR of 60–89 mL/min/1.73 m2. Analysis of interaction terms revealed a significant interaction between eGFR and the CCI score (p = 0.02), indicating that the effect of eGFR on the outcome varies with the CCI score. No significant interaction was found with age (p = 0.13).
Subgroup analyses
To mitigate the potential impact of hyperfiltration that may accompany patients with DM, subgroup analyses were conducted based on the following criteria: BMI of <25 kg/m2 or ≥25 kg/m2, HbA1c levels of <7% or ≥7%, use of RAS blockers, and urine ACR of <10 μg/mgCr or ≥10 μg/mgCr (Fig. 3). The findings revealed that even in subgroups with a lower likelihood of accompanying single-nephron hyperfiltration—specifically those with a BMI of <25 kg/m2 and urine ACR of <10 μg/mgCr—the increased risk associated with an eGFR of 45–59 mL/min/1.73 m2 remained consistent. No significant interactions were observed between an eGFR of 45–60 mL/min/1.73 m2 and any variables regarding their impact on the outcome.
Discussion
This longitudinal study of older diabetic patients without albuminuria demonstrates that an eGFR of 45–59 mL/min/1.73 m2, as well as an eGFR of 45 mL/min/1.73 m2, is associated with an elevated risk of composite events encompassing mortality and ESKD. This finding suggests that the current traditional eGFR threshold used to define CKD is applicable as a criterion for older patients with T2DM. These observations remained consistent across subgroups characterized by well-controlled blood glucose levels, non-obesity, use of RAS blockers, and normal urinary albumin excretion. Hence, even in older patients with DM without factors contributing to glomerular hyperfiltration, a mild decline in eGFR within 45–59 mL/min/1.73 m2 is associated with an elevated risk as is in eGFR under 45 mL/min/1.73 m2.
The diagnostic criteria of CKD have not been changed for the last two decades. Previous studies have established well that GFR declines by approximately 6–7 mL/min/1.73 m2 per decade from 35–40 years of age with normal aging [19,20]. Therefore, using a single absolute GFR threshold, regardless of age, to identify CKD has been a subject of debate because it leads to the diagnostic labeling of CKD in many older adults with renal senescence [21,22]. Although whether this aging-associated GFR decline is “normal” remains controversial [13], a modified threshold of 45 mL/min/1.73 m2 for defining CKD has been proposed for individuals aged ≥65 years [10]. Its supporting data have been based on comparable mortality as an outcome of interest, which is rationalized by an extremely low incidence (<1%) of ESKD in older patients with an eGFR of 45–59 mL/min/1.73 m2 [10]. However, in our study cohort of patients with DM, the composite outcome of mortality and incident ESKD in this eGFR category was higher than those in patients with preserved kidney function.
The reasons for the divergent results of our study compared to those of previous population-based studies [23–26], indicating the benign nature of mildly reduced eGFR, are unclear, but three potential explanations can be considered. First, patients with DM have a higher risk of mortality and CKD progression, even in the absence of albuminuria. Normoalbuminuric renal impairment is common in patients with T2DM regardless of age and has been suggested as a distinct phenotype of diabetic kidney disease in recent studies [27,28]. Therefore, the mildly reduced eGFR in older patients with DM is likely attributed to beyond the age-related decline. Moreover, a substantial proportion of patients with DM have little or no proteinuria despite biopsy-proven diabetic nephropathy [29–31].
Second, although not directly measurable, the presence of DM-related glomerular hyperfiltration might have shifted the age-adjusted normal GFR threshold upward. The supra-physiological increase in GFR reported in 6%–73% of patients with DM could persist at the single-nephron level, despite an overall decrease in the GFR of the whole kidney [32]. Although the mechanism by which DM contributes to glomerular hyperfiltration has not been clearly elucidated, hyperglycemia and overweight are likely to be associated [32,33]. The use of RAS blockers, which reduce intraglomerular pressure, is expected to mitigate this effect. In addition, glomerular hyperfiltration can lead to an increase in urinary albumin excretion [32]. Consequently, we conducted subgroup analyses targeting individuals with a lower likelihood of accompanying hyperfiltration; those with BMI <25 kg/m2, HbA1c <7%, using RAS blockers, and urine ACR <10 μg/mgCr. The results were not significantly different from those of the overall cohort.
Third, older patients included in our study may have had a relatively lower muscle mass, leading to a potential overestimation of GFR using creatinine-based eGFR. Sarcopenia is more frequent in the older Asian population, particularly in those with DM than in those without it [34,35]. Confirming this issue requires further research utilizing alternative methods to assess GFR values. However, considering the current widespread use of creatinine-based eGFR in clinical practice, there is a potential risk of overlooking participants at a high risk when applying the reduced eGFR threshold of 45 mL/min/1.73 m2 as a CKD definition in DM for individuals aged ≥65 years. This concern is particularly relevant considering the high prevalence of DM in this age group, making this finding remarkable.
Our finding showing the worse renal outcomes of mildly reduced eGFR is consistent with the recently reported meta-analysis study [36]. Although the sample size of the present study is relatively smaller compared to the previous study, the strength of our study is the completeness of urine ACR data and concrete outcome measures by combining nationwide death and ESKD registry data and adjusting multiple clinical covariates.
This study has several limitations that require consideration. First, unmeasured confounders may have existed because this was a single-center retrospective study. Nevertheless, a long-term follow-up duration of 8 years was used. Second, our cohort was based on an Asian population and did not include a non-diabetic control group to compare risks. Additionally, since the 2021 CKD-EPI equation was not validated and has been reported to have a higher bias compared to the CKD-EPI 2009 equation for Koreans [37,38], we used the CKD-EPI 2009 equation. These factors may limit generalizability. Third, considering that our center is a tertiary referral university hospital, the study cohort may not represent the general T2DM population. Another important limitation was that both eGFR and ACR were measured at baseline only. As such, the relative effect of eGFR 45–59 mL/min/1.73 m2 vs. 60–89 mL/min/1.73 m2 may be overestimated since patients who had eGFR ≥60 mL/min/1.73 m2 at baseline could have lost kidney function after the baseline and yet remained in the reference group in the analysis. It should be also noted that defining ESKD based on the receipt of dialysis reflected not only disease outcomes but also treatment decisions. Considering that older patients do not always prefer care that prolongs their lives, the probability of dialysis initiation among individuals may vary with age [39]. Some patients with ESKD who rejected dialysis treatment may have been counted as a mortality but not those with ESKD [40]. Another limitation that requires consideration was that, although we excluded hospitalized patients, those who had acute kidney injury might have been included in the study cohort due to the lack of longitudinal laboratory data. Furthermore, some of the patients who developed ESKD might have required dialysis due to irreversible acute kidney injury rather than the progression of CKD.
Nevertheless, the clinical implications of our study are considerable as concrete outcomes were analyzed after adjusting for major comorbidities. Furthermore, unlike some registry-based studies that lacked many clinical covariables such as laboratory data and medication, we adjusted for quantified albuminuria data and multiple clinical covariates that affect patient outcomes.
Given the aging global population and the natural decline in GFR with age, age-adjusted CKD diagnostic criteria have been proposed, with lower eGFR thresholds for older adults to reflect the generally neutral prognosis of mildly reduced eGFR in this group. However, our longitudinal study in older diabetic patients showed an increased risk of adverse outcomes with an eGFR of 45–59 mL/min/1.73 m2. Thus, applying age-adapted GFR criteria universally may overlook high-risk individuals in elderly diabetic populations.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
