Introduction
Chronic kidney disease (CKD) is highly prevalent in the elderly population and is frequently accompanied by multiple comorbidities, frailty, and progressive decline in renal function [1,2]. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) have emerged as a cornerstone in the management of CKD due to their robust nephroprotective and cardiovascular (CV) benefits. These agents are generally well-tolerated and remain effective even in older adults [3–5]. Beyond their glucose-lowering effects, SGLT2i significantly reduce proteinuria and attenuate the progression of renal impairment, both of which are critical for lowering CV risk in this vulnerable population [6,7].
The benefits of SGLT2i in the elderly population are especially critical as older patients typically are more susceptible to the adverse effects of CKD and may experience more pronounced benefits from interventions that can halt the progression of both renal and CV deterioration. However, prescription of SGLT2i in elderly patients necessitates careful consideration, particularly for those with frailty such as low body mass index (BMI), low blood pressure, compromised oral intake, or a heightened risk of volume depletion [8–11]. Notably, the initiation of SGLT2i therapy is associated with an initial dip in estimated glomerular filtration rate (eGFR), a phenomenon primarily reflecting hemodynamic effect in the kidney rather than actual renal damage [12]. This early eGFR decline, while generally transient in most cases, might cause consistent renal deterioration in those vulnerable elderly populations [13].
Our previous research highlighted that elderly patients might experience a more pronounced initial decline in eGFR following SGLT2i initiation. However, renal function generally stabilizes over the subsequent year, suggesting no overall long-term deterioration [14].
Nevertheless, that study did not evaluate which subgroups of patients are more vulnerable to rapid decline in renal function. To address this gap, the present study aimed to investigate the impact of frailty-associated factors on rapid renal function decline in elderly CKD patients treated with SGLT2i.
Methods
This retrospective study was approved by the Ethics Committee of Hallym University Sacred Heart Hospital with the waiver for informed consent (No. 2024-01-007). It was conducted following the Declaration of Helsinki. Medical records were anonymized and de-identified before analysis.
Study design and subjects
This study utilized the clinical data warehouse, which has been operational since 2016 and employs the QlikView Elite Solution (QlikTech Inc.) for advanced big data analytics. As our center reports eGFR higher than 90 mL/min/1.73 m2 as just >90 mL/min/1.73 m2, patients with an eGFR less than 90 mL/min/1.73 m2 were included in this study. It was calculated with the creatinine-based CKD-EPI (CKD Epidemiology Collaboration) equation. Eligibility for the study required patients to have at least three recorded eGFR measurements: at baseline, 3 months, and 1 year. Initially, data were collected from 465 CKD patients aged 65 years or older who received SGLT2i therapy between June 2018 and December 2022. Among these, patients with incomplete data on critical variables such as height, body weight, serum albumin, or proteinuria (n = 163) were excluded. The final cohort for analysis included 302 elderly patients. To assess the risk of selection bias, we compared baseline characteristics between included and excluded patients. As shown in Supplementary Table 1 (available online), no significant differences were observed in age, sex, diabetes mellitus (DM) status, baseline eGFR, or medication use, suggesting that the final analytic sample was representative of the original cohort.
Covariates
Demographics data and baseline comorbidities were collected at the initiation of SGLT2i. A comprehensive panel of serum biochemical parameters, including hemoglobin, glucose, hemoglobin A1c, albumin, uric acid, blood urea nitrogen, creatinine, total cholesterol, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglyceride levels, was assessed within 1 month prior to or on the date of SGLT2i initiation. Proteinuria was assessed using urine dipstick results, with proteinuria defined as ≥1+. In a subset of patients (n = 168), urine protein-to-creatinine ratio (uPCR, g/g) data were also available. Anemia was defined as hemoglobin <13.0 g/dL in male and <12.0 g/dL in female, consistent with World Health Organization criteria. Given the potential impact on renal hemodynamics, we also evaluated the effects of antihypertensive medications (renin-angiotensin system [RAS] blockers, beta-blockers, calcium channel blockers [CCBs], and diuretics) and antidiabetic medications (metformin, sulfonylureas, dipeptidyl peptidase-4 [DPP-4] inhibitors, and insulin).
All patients’ nutrition condition was evaluated according to Geriatric Nutritional Risk Index (GNRI) formula: GNRI = [1.489 × albumin (g/L)] + [41.7 × (weight/WLo)], where WLo means ideal weight and was calculated from the Lorentz equations: for male: H – 100 – [(H – 150)/4]; for female: H – 100 – [(H – 150)/2.5] (H, height) [15,16]. Even for patients with a height of less than 150 cm, the formula was used without modification. Based on previous literature, a GNRI value of ≤98 has been shown to identify patients with at least nutritional risk, while scores >98 indicate no risk [15,17]. Since this study analyzed patients in a chronic stable condition, we used a threshold of 98 for comparison.
In this study, we selected four objective and clinically relevant variables, more advanced age ≥75 years, anemia, proteinuria, and low GNRI (≤98), as frailty-related clinical factors, based on prior literature showing their associations with vulnerability and adverse outcomes in elderly CKD patients [18–21].
Changes in estimated glomerular filtration rate and primary outcome
We calculated both the absolute and relative changes in eGFR at 3 and 12 months compared to baseline. In accordance with previous randomized controlled trials, a significant initial dip was defined as a >10% decline in eGFR at 3 months following SGLT2i initiation.
The primary renal outcome was defined as a composite of 1) an absolute decline in eGFR >10 mL/min/1.73 m2 within 12 months or 2) a ≥40% reduction from baseline. While a ≥40% decline is a well-validated surrogate endpoint for CKD progression [22–24], we also included the absolute decline criterion to better capture early and clinically meaningful renal deterioration during the 12-month follow-up period. This approach was supported by previous data and reflects the clinical importance of sustained eGFR loss with SGLT2i initiation [25]. We have included a scatter plot (Supplementary Fig. 1, available online) that visually shows the concordance between these two renal outcome definitions
Statistical analysis
Statistical analyses in this study were conducted using IBM SPSS version 25.0 (IBM Corp.). Continuous variables were expressed as mean ± standard deviation or median with interquartile range, depending on the data distribution. Categorical variables were presented as counts and percentages. The differences in eGFR changes between the two GNRI groups were assessed using the independent samples t test for normally distributed data, while the Mann-Whitney U test was applied for non-normally distributed variables. These tests were utilized to compare the absolute and percentage changes in eGFR at 3 months and 12 months between the GNRI >98 and ≤98 groups. Odds ratios (ORs) were calculated using logistic regression models to assess the impact of multiple frailty-associated factors on the rapid decline in renal function. The results were expressed as ORs with 95% confidence intervals (CIs). A p-value <0.05 was considered statistically significant.
Results
Baseline clinical characteristics
This study included 302 elderly patients aged ≥65 years, comprising 191 males and 111 females. Among them, 246 (81.5%) had DM. The mean baseline eGFR was 48.1 ± 10.1 mL/min/1.73 m2, with 47.4% and 38.4% classified as having stage 3 and stage 4 CKD, respectively. Nutritional status appeared generally favorable, with a mean serum albumin level of 4.3 ± 0.4 g/dL and a mean GNRI of 112.5 ± 10.0. Based on a GNRI cutoff of 98, 37 patients (12.3%) were classified as having low GNRI. Table 1 compares the baseline characteristics between patients with low and high GNRI. Those in the low GNRI group were significantly older and had lower body weight and BMI (all p < 0.001). They also exhibited lower levels of hemoglobin, total protein, albumin, phosphate, and triglycerides. Importantly, proteinuria was more prevalent and severe in this group, with a mean uPCR of 1.8 ± 2.4 g/g, compared to 0.7 ± 1.1 g/g in the high GNRI group (p = 0.01). No significant differences were found in lipid profiles, including total cholesterol, LDL-C, and HDL-C.
Regarding medication use, 233 patients (77.1%) were receiving RAS blockers, 157 (52.0%) were on CCBs, and 88 (29.1%) were using diuretics, with no significant differences between GNRI groups. However, patients with GNRI ≤98 were more likely to be prescribed loop diuretics compared to those with GNRI >98 (32.4% vs. 17.0%, p = 0.07), suggesting a potential trend toward higher diuretic burden in the nutritionally vulnerable group. Metformin was more frequently prescribed in the high GNRI group, while the use of other antidiabetic medications, including sulfonylureas, DPP-4 inhibitors, and insulin, was comparable between the groups.
Estimated glomerular filtration rate changes at 3 and 12 months by Geriatric Nutritional Risk Index group
Table 2 presents the differences in eGFR across baseline, 3 months, 6 months, and 12 months, stratified by GNRI groups. This study only included patients who had mandatory eGFR data at baseline, as well as at 3 and 12 months; eGFR at the 6-month interval was available for only 180 participants. At baseline, eGFR values were comparable between the low and high GNRI groups. From 3 months onwards, however, those with a GNRI score ≤98 experienced a greater decrease in eGFR.
We evaluated both absolute and percentage changes in eGFR at 3 and 12 months after SGLT2i initiation. Higher GNRI scores were associated with more stable renal function (Fig. 1A, B), while patients with GNRI ≤98 showed significantly greater declines at both time points (Fig. 1C–F). For 12 months, those with GNRI >98 had a modest decline (–2.0 mL/min/1.73 m2), consistent with previous studies. In contrast, those with GNRI ≤98 experienced a more substantial reduction (–4.0 mL/min/1.73 m2). This suggests that SGLT2i are generally well-tolerated by most elderly patients with CKD; however, those with a low GNRI score may be more vulnerable to renal function decline. Additionally, an initial eGFR dip >10% at 3 months was more common in the low GNRI group (51.4% vs. 31.3%, p = 0.01) (Table 2).
Frailty-related factors and rapid decline in renal function
Rapid decline in renal function occurred in 29 patients (9.6%). Compared to those without decline, these patients were older (81.6 years vs. 78.3 years, p = 0.045), with lower hemoglobin, serum albumin, and GNRI scores (all p < 0.001), and a higher prevalence of proteinuria (p = 0.001). Based on these findings, we evaluated the impact of four frailty-related factors: advanced age (≥75 years), anemia, low GNRI (≤98), and proteinuria.
In univariate logistic regression (Table 3), each factor was significantly associated with rapid eGFR decline: age ≥75 years (OR, 2.34; 95% CI, 1.03–5.32), anemia (OR, 3.61; 95% CI, 1.50–8.75), proteinuria (OR, 4.80; 95% CI, 1.78–10.15), and low GNRI (OR, 8.43; 95% CI, 3.62–16.66). Diuretic use was also associated (OR, 4.08; 95% CI, 1.86–8.98). However, in the multivariable model, proteinuria (OR, 3.84; 95% CI, 1.34–11.02; p = 0.01), low GNRI (OR, 2.75; 95% CI, 1.03–7.35; p = 0.04), and diuretic use (OR, 3.75; 95% CI, 1.60–8.80; p = 0.002) remained significant predictors. Anemia showed a borderline association (OR, 2.51; 95% CI, 0.91–6.95; p = 0.07), whereas age ≥75 years was no longer significant (OR, 1.35; 95% CI, 0.51–3.59; p = 0.60), suggesting that frailty-related comorbidities rather than age per se were more predictive of renal decline.
Given the strong predictive value of low GNRI in the multivariable model, we further assessed the comparative performance of nutritional markers. Additional models incorporating GNRI, serum albumin, or BMI were evaluated to determine which parameter better captured nutritional risk. Both GNRI and serum albumin were significantly associated with rapid decline in renal function, whereas BMI was not. The area under the curve was similar for GNRI (0.833) and albumin (0.832), and lower for BMI (0.789) (Supplementary Table 2, available online).
A cumulative risk assessment (Fig. 2) demonstrated a clear stepwise increase in the odds of rapid decline in renal function with the accumulation of frailty-related factors. Age ≥75 years alone was associated with increased risk (OR, 2.34; 95% CI, 1.03–5.32), which rose with the addition of anemia (OR, 3.86; 95% CI, 1.76–8.47) and further with proteinuria (OR, 4.85; 95% CI, 2.19–10.77). Notably, patients exhibiting had the highest risk (OR, 9.70; 95% CI, 3.82–24.81). These findings highlight a graded association between the number of frailty-related clinical factors and the likelihood of rapid renal decline.
Discussion
This study demonstrated that SGLT2i are generally effective and well-tolerated in elderly patients with CKD, with a mean annual eGFR decline of –2.0 mL/min/1.73 m2, consistent with the –2.16 mL/min/1.73 m2 reported in the EMPA-KIDNEY trial [26]. However, patients with frailty-related factors, defined by more advanced age, anemia, proteinuria, and low GNRI, exhibited a substantially greater decline, with a median annual eGFR loss of –4.0 mL/min/1.73 m2. These findings suggest that, despite the overall renoprotective effects of SGLT2i, frailty-associated factors may confer additional vulnerability to rapid renal function loss. While prior studies have demonstrated the safety of SGLT2i even in frail individuals [27], our results underscore the importance of optimizing modifiable risk factors such as anemia, use of diuretics, and nutritional status before and during SGLT2i therapy to minimize adverse outcomes.
The KDIGO (Kidney Disease: Improving Global Outcomes) 2024 guidelines recommend SGLT2i as first-line therapy for CKD, regardless of DM status [28]. Major trials such as VERTIS-CV and DECLARE-TIMI 58 have confirmed their efficacy in patients aged ≥65 years, with no significant increase in adverse events [3–5]. However, a recent systematic review in frail or older adults with DM and heart failure (HF) reported no significant renal or macrovascular benefit, despite reductions in all-cause mortality and CV risk [29]. These discrepancies underscore the need for real-world data focused on renal outcomes in elderly patients with frailty-related factors.
In this elderly CKD population (mean age, 75 years), nutritional status emerged as a key determinant of eGFR changes following SGLT2i initiation. This finding aligns with prior studies showing that malnutrition is common in older adults with CKD and is strongly associated with adverse clinical outcomes [30,31]. The hemodynamic effects of SGLT2i, particularly the reduction of intraglomerular pressure, may be compromised in the presence of hypoalbuminemia, a key contributor to low GNRI scores [32]. This aligns with previous studies showing that malnutrition and hypoalbuminemia, both common in older adults, can lead to volume depletion and impaired renal hemodynamics, thereby accelerating eGFR decline [33,34]. More importantly, low GNRI remained independently associated with rapid decline in renal function even after adjusting for proteinuria, suggesting that GNRI captures additional prognostic dimensions such as sarcopenia and physiologic vulnerability, not reflected by albuminuria alone.
Although our findings may appear to contrast with a DAPA-CKD subgroup analysis that found no excess renal risk among severely frail patients [27], key methodological differences exist. The frailty index in DAPA-CKD did not incorporate nutritional markers such as GNRI, and their study evaluated long-term outcomes in a controlled trial setting.
And although patients aged ≥75 years exhibited greater eGFR decline in univariate analysis, age did not remain an independent predictor in multivariable models. This attenuation may reflect confounding by nutritional and hemodynamic factors, including diuretic use. These findings are consistent with prior data demonstrating that chronological age alone does not modify the renal benefits of SGLT2i, and reinforce the notion that clinical frailty, rather than age per se, may better capture the heterogeneity of treatment response in older adults [35,36].
We also found that diuretic use emerged as another independent predictor of rapid decline in renal function. This likely reflects both the hemodynamic effects of combined natriuresis as well as the underlying clinical indication for the use of diuretics (e.g., fluid overload, HF), which is itself a marker of poor prognosis. Caution should be exercised when prescribing concurrent diuretics and SGLT2i in elderly CKD patients, with close attention to volume status and comorbidity burden.
Another notable observation in this study is the possible role of anemia in predicting renal function decline in elderly patients treated with SGLT2i. While SGLT2i are known to enhance erythropoiesis via modulation of hypoxia-inducible pathways [37,38], the presence of baseline anemia may still portend adverse renal outcomes. Mechanisms may include impaired erythropoietin production, inflammation-driven hepcidin elevation, and renal hypoxia [39–41]. While prior studies have focused on hemoglobin improvements during therapy, few have explored the prognostic implications of pre-existing anemia in this context. Further studies are needed to delineate the prognostic role of anemia in this setting and its implications for individualized treatment strategies.
It is important to note that all patients in our study were treated with SGLT2i; thus, we could not assess whether baseline factors such as anemia or proteinuria modified the therapeutic effect of SGLT2i. Rather, our findings suggest that these clinical parameters serve as prognostic indicators, helping to identify patients who remain at increased risk of renal deterioration despite therapy.
Despite these strengths, this study has several limitations. First, it was a single-center retrospective analysis, which may limit external generalizability. However, the inclusion of over 300 elderly patients offers valuable real-world insights. Second, the proportion of patients with low GNRI was relatively small (12.3%), potentially limiting the broader applicability of nutritional risk findings. Third, while malnutrition and hypoalbuminemia were hypothesized to contribute to renal decline, we were unable to assess inflammation-related markers such as C-reactive protein or interleukin-6, which may underlie the interplay between frailty, malnutrition, anemia, and renal progression. We also acknowledge the clinical relevance of volume status, particularly in patients receiving SGLT2i, due to the drug’s natriuretic and hemodynamic effects. However, due to the retrospective design and lack of standardized assessments, we could not adequately account for this variable. Lastly, a validated frailty index could not be applied, as functional and cognitive measures were not consistently available in the electronic medical records. Instead, we used objective clinical surrogates including age, anemia, GNRI, and proteinuria to reflect frailty-related risk. While these proxies are supported by prior literature, they do not fully capture the multidimensional construct of frailty.
In summary, SGLT2i appears to be generally safe and effective in elderly CKD patients, particularly among those with preserved nutritional status. While older age is often considered a risk factor for adverse outcomes, our findings suggest that age alone is not the primary driver of early renal function loss. Rather, the presence of frailty-related factors such as anemia, proteinuria, and low GNRI more accurately identifies those at heightened risk despite SGLT2i therapy. These results underscore the importance of comprehensive baseline assessment beyond chronological age.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
