Age-dependent effects of SGLT2 inhibitors on stroke risk in geriatric patients with diabetes and atrial fibrillation

Abstract

Background

Atrial fibrillation (AF) and diabetes mellitus (DM) are associated with an increased risk of ischemic stroke, particularly in geriatric populations. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) have demonstrated cardiovascular benefits, but their effects on stroke risk may vary by age. This study aimed to explore the age-dependent effects of SGLT2i on stroke risk in patients with AF and DM.

Methods

This historical longitudinal follow-up cohort study included 9,669 patients with AF and DM from the National Taiwan University Hospital database (2010–2020). Patients were stratified into three age groups (< 75, 75–89, and ≥ 90 years) to compare SGLT2i users and non-users within each age group. Cox proportional hazards models were used to evaluate stroke risk, adjusting for CHA₂DS₂-VASc score and oral anticoagulant use. Interaction analysis assessed age-specific SGLT2i effects.

Results

In patients aged < 75 years, SGLT2i use significantly reduced stroke risk (HR 0.63, 95% CI 0.44–0.88, P < 0.05). Stroke risk was neutral in patients aged 75–89 years (HR 0.95, 95% CI 0.60–1.50), but significantly increased in those aged ≥ 90 years (HR 5.04, 95% CI 1.20–21.1, P < 0.05). Interaction analysis confirmed a significant age-dependent effect (aged ≥ 90 years x SGLT2i use HR 6.39, 95% CI 1.49–27.40, P < 0.05).

Conclusions

The impact of SGLT2i on stroke risk varies significantly by age. While protective in younger patients, SGLT2i may increase stroke risk in those aged ≥ 90 years. These findings highlight the importance of age-specific considerations in prescribing SGLT2i for patients with AF and DM.

Graphical Abstract

Introduction

Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting approximately 9% of individuals aged 65 years and older, with prevalence increasing to nearly 18% in those over 85 years old [1]. In the elderly population, the risk of thromboembolic events, particularly ischemic stroke, is substantially increased by the coexistence of AF and diabetes mellitus (DM) [2]. It is becoming increasingly essential to comprehend the intricate effects of pharmacological interventions in geriatric patients with these comorbidities as the global population ages.

Sodium-glucose cotransporter-2 inhibitors (SGLT2i) have emerged as a promising class of antidiabetic medications, as evidenced by their significant cardiovascular benefits in many large-scale clinical trials [3]. Our previous research demonstrated that the use of SGLT2i was associated with a 20% reduction in the risk of stroke among patients with AF and DM [4]. Nevertheless, the effects of this study were not explicitly examined in various age categories, emphasizing geriatric individuals.

The geriatric population presents unique challenges in pharmacotherapy due to age-related changes in pharmacokinetics and pharmacodynamics and multiple comorbidities [5]. The potential increase in stroke risk with SGLT2i use in the very elderly (≥ 90 years) raises important questions about the balance of benefits and risks in this vulnerable group. Factors such as frailty, polypharmacy, and altered renal function may contribute to this observed effect and warrant careful investigation [6].

In our previous study, we found that using SGLT2i was associated with a lower risk of stroke [4]. This study aims to elucidate the age-dependent effects of SGLT2i on stroke risk in geriatric patients with AF and DM, with a particular focus on the very elderly population. By stratifying patients into distinct age groups (< 75, 75–89 years, and ≥ 90 years), we seek to comprehensively analyze SGLT2i efficacy and safety across the spectrum of older adults. Our findings may have significant implications for clinical practice guidelines, potentially necessitating age-specific recommendations for SGLT2i use in geriatric patients with AF and DM.

Understanding these age-dependent effects is crucial for optimizing care in the rapidly growing geriatric population, ensuring that the potential benefits of novel therapies are balanced against age-specific risks [7]. This research contributes to the evolving field of geriatric cardiology, aiming to improve outcomes and quality of life for older adults with complex cardiovascular and metabolic conditions.

Methods

Study population

The study population was described in our previous study. In brief, this was a longitudinal follow-up cohort study using medical data collected from the National Taiwan University Hospital (NTUH) Integrative Medical Database [4, 8, 9]. The study was approved by the Research Ethics Committee of the National Taiwan University Hospital (approval numbers: 200911002R and 202011019RINC). The requirement for informed consent was waived due to the retrospective nature of the database analysis and the absence of patient identifiers.

The study population consisted of patients with AF and DM. AF was diagnosed based on patient history, serial electrocardiograms (ECGs), or ambulatory ECG monitoring. Diabetes was diagnosed according to the American Diabetes Association guidelines, defined as fasting glucose > 125 mg/dL, hemoglobin A1c > 6.5%, or the use of oral hypoglycemic agents or insulin [10]. In our dataset, SGLT2i usage for heart failure alone (without diabetes) was too rare to allow meaningful subgroup analysis. Therefore, this study focused on patients with both diabetes and AF.

Out of 29,341 patients with AF identified in the database since January 2010, a total of 9,669 patients with a concomitant diagnosis of DM beginning in January 2016, when SGLT2 inhibitors became available in Taiwan, were included in the analysis. Follow-up started after 2016 and continued until the first ischemic stroke event or the end of the study period (December 2020), whichever came first. Patients without events were censored at the end of follow-up.

To investigate the effects of SGLT2i in geriatric patients, the population was stratified into three age groups: <75 years, 75–89 years, and ≥ 90 years. These thresholds were selected based on clinical relevance and prior geriatric studies [11], which demonstrate that patients aged ≥ 90 years exhibit distinct physiological characteristics, increased frailty, and unique therapeutic needs. In contrast, we acknowledge that categorizing age as a continuous variable may result in some data loss. Our primary goal was to provide actionable insights for clinicians treating the elderly (≥ 90 years), where frailty and vulnerability to adverse outcomes are more pronounced.

Outcome

In the present study, our primary aim was to investigate whether the risk of ischemic stroke was also decreased in significantly older DM patients with AF who were taking SGLT2i. Ischemic stroke was defined as sudden-onset and focal or global neurological deficits that were not explained by other origins, with supporting evidence from imaging studies. Hemorrhagic stroke was excluded from the analysis because it was generally a complication of anticoagulant use and not related to AF per se.

Statistical analysis

Continuous variables are presented as means ± standard deviations (SD), and categorical variables as percentages. The data was compared using the χ² test (categorical variables) or the Student t-test (continuous variables). The time to the first episode of the stroke event was depicted with the Kaplan‐Meier estimate of the survival function. The log‐rank statistics tested the difference between the survival curves. The independent effect of variables to predict thromboembolic events was calculated using a Cox proportional hazards regression model. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated accordingly. The adjusted covariates included potential confounders of estimating AF‐related stroke as the independent covariates: age, sex, hypertension, diabetes, coronary artery disease, myocardial infarction, peripheral artery occlusive disease, congestive heart failure, history of stroke, and use of oral anticoagulants (OAC) [21].

Another multivariable model was also created to adjust for the joint effect of these covariates, as depicted in the CHA₂DS₂-VASc score. P < 0.05 was considered statistically significant. Statistical analysis was performed using STATA version 17 for Windows (STATA, Inc., College Station, TX).

Results

Baseline characteristics

The baseline characteristics of AF patients with DM were stratified by age groups (< 75, 75–89, and ≥ 90 years) and by SGLT2 inhibitor (SGLT2i) use are summarized in Table 1. Among patients aged < 75 years, SGLT2i users were younger (63.0 ± 8.0 vs. 64.6 ± 7.6 years, P < 0.05), had lower rates of chronic kidney disease (CKD) (12% vs. 33%, P < 0.05), and were more likely to use OAC (65% vs. 49%, P < 0.05). Similar trends were observed in the 75–89-year group, where SGLT2i users had lower rates of CKD (16% vs. 40%, P < 0.05) and higher OAC use (65% vs. 42%, P < 0.05). In the ≥ 90-year group, SGLT2i users had lower CKD prevalence (12% vs. 43%, P < 0.05) and greater OAC use (48% vs. 30%, P < 0.05). Across all age groups, SGLT2i users showed a higher prevalence of vascular disease and coronary artery disease compared to non-users. The CHA₂DS₂-VASc scores were generally similar between SGLT2i users and non-users, with no significant differences observed in any group except for the < 75-year group, where SGLT2i users had slightly lower scores (2.6 ± 1.5 vs. 2.7 ± 1.6, P < 0.05).

Table 1 Baseline characteristics of AF patients with DM according to age groups

Use of SGLT2i was associated with a higher risk of stroke in extremely elderly AF patients with DM

Figure 1 shows the Kaplan-Meier survival curve comparing stroke-free survival between SGLT2i users and non-users in patients aged ≥ 90 years. SGLT2i users exhibited a significantly higher stroke risk compared to non-users (log-rank P < 0.05). The divergence in survival curves becomes evident early during the follow-up period. It persists over time, highlighting a consistent pattern of increased stroke risk among SGLT2i users in this very elderly group.

Fig. 1

Stroke-free survival in patients ≥ 90 years with atrial fibrillation and diabetes mellitus by SGLT2i use. SGLT2i users showed significantly higher stroke risk (log-rank P = 0.015). Red line: SGLT2i users; Blue line: non-users; SGLT2i, Sodium-glucose cotransporter-2 inhibitors

Figure 2 illustrates Kaplan-Meier survival curves stratified by age groups (< 75 years and 75–89 years), comparing stroke-free survival between SGLT2i users and non-users. In patients aged < 75 years, SGLT2i users had a lower stroke risk compared to non-users, with significant separation of the curves over time (log-rank P < 0.05). In the 75–89-year group, no significant difference in stroke-free survival was observed between SGLT2i users and non-users, as the survival curves remained closely aligned throughout the follow-up period.

Fig. 2

Stroke-free survival in patients < 90 years with atrial fibrillation and diabetes mellitus by age and SGLT2i use. A Patients < 75 years old: SGLT2i users had lower stroke risk (log-rank P = 0.049). B Patients 75–89 years old: No significant difference in stroke risk between groups. Red line: SGLT2i users; Blue line: non-users; SGLT2i, Sodium-glucose cotransporter-2 inhibitors

An interaction analysis was performed to assess the relationship between age and the effects of SGLT2i use on stroke risk by incoporating an interaction term (aged ≥ 90 years by SGLT2i use) in the multivariable model. The results revealed a significant interaction between age ≥ 90 years and SGLT2i use (HR 6.39, 95% CI 1.49–27.40, P = 0.012), indicating that the effect of SGLT2 inhibitors on stroke risk differs significantly between patients aged ≥ 90 years and those < 90 years. This suggests an age-dependent reversal of SGLT2i effects, with younger patients experiencing reduced stroke risk and very elderly patients showing increased risk.

Multivariable analyses to adjust for CHA2DS2-VASc score

Unadjusted and adjusted models of SGLT2i use effects on stroke risk stratified by age groups were summarized in Table 2. In the unadjusted model, SGLT2i use was associated with reduced stroke risk in patients aged < 75 years (HR 0.71, 95% CI 0.50–1.00), a neutral effect in those aged 75–89 years (HR 1.11, 95% CI 0.71–1.74), and an increased risk in patients aged ≥ 90 years (HR 4.96, 95% CI 1.19–20.7, P < 0.05). These trends persisted after adjustments for individual CHA₂DS₂-VASc components and OAC use in Model 1 and for the overall CHA₂DS₂-VASc score and OAC use in Model 2, with stroke risk significantly reduced in the < 75-year group (HR 0.63, 95% CI 0.44–0.88, P < 0.05) and elevated in the ≥ 90-year group (HR 5.04, 95% CI 1.20–21.1, P < 0.05).

Table 2 Unadjusted and adjusted models of SGLT2i effects on Stroke Risk Stratified by Age groups

Discussion

A critical age-dependent effect of SGLT2i on stroke risk in patients with DM and AF is revealed by our study. Our previous research has shown that SGLT2i generally protects against stroke in this population [4, 12]. However, our current analysis reveals a more complex picture: SGLT2i appears protective against stroke in patients under 75, neutral in those aged 75–90, and potentially detrimental in patients aged 90 and older. These results have substantial implications for using SGLT2i in geriatric patients with DM and AF, and they emphasize the significance of age-specific considerations in pharmacological management.

The protective effect of SGLT2i against stroke in younger patients (< 75 years) is consistent with our previous findings and other large-scale studies [3, 12]. This protection is likely to be achieved through various mechanisms, such as enhanced glycemic control, reduced inflammation, and beneficial effects on cardiovascular risk factors [13]. The neutral effect observed in the 75–90 age group implies that the benefits of SGLT2i may diminish with age, but they do not inherently become detrimental in this cohort. Nevertheless, the risk of stroke that is associated with the use of SGLT2i in patients aged 90 years and older is a novel and alarming discovery.

A variety of factors may influence this age-dependent reversal of effect. Aging is linked to drug metabolism and excretion changes, which may result in elevated drug concentrations and potential adverse effects [14]. The risk of thromboembolism may be elevated in the very elderly due to the dehydrating effects of SGLT2i, although this hypothesis requires further investigation. Furthermore, patients who are 90 years of age or older are more susceptible to frailty and have multiple comorbidities, which may interact with SGLT2i in unexpected ways [15]. This age group is notably affected by the intricate interplay between frailty, multimorbidity, and medication effects. The proposed mechanisms remain speculative as our study design does not allow mechanistic exploration. Randomized controlled trials or physiological studies are needed to elucidate these pathways and confirm the observed age-dependent effects of SGLT2i use.

The physiological reserve of very elderly patients may be reduced, rendering them more susceptible to hemodynamic alterations caused by SGLT2i, including volume depletion and hemoconcentration [16]. These changes could potentially increase the risk of thromboembolism in the context of AF. Furthermore, the ability to maintain homeostasis in response to physiological stress decreases with age [17]. The very elderly may have an impaired capacity to adapt to the metabolic and hemodynamic effects of SGLT2i, potentially leading to adverse outcomes. Advanced age is also associated with changes in coagulation factors and inflammatory markers [18]. SGLT2i may interact differently with these altered profiles in the very elderly, potentially tipping the balance towards a prothrombotic state.

The study has numerous essential implications for clinical practice in geriatric cardiology. This information highlights the need for an age-stratified approach to SGLT2i use in patients with DM and AF. While SGLT2i may benefit younger patients, caution is warranted for those 90 years and older. For patients with DM and AF aged 75–90, where SGLT2i seems to have a neutral impact on stroke risk, other criteria, including glycemic management, cardiovascular benefits, and possible adverse effects, should guide the decision-making [19]. Close monitoring for signs of dehydration, electrolyte imbalances, and thromboembolic events is essential if SGLT2i is used in patients aged 90 and above [20]. Alternative glucose-lowering therapies with established safety profiles in the very elderly should be considered for those 90 years and older. Management of very elderly patients with DM and AF should involve collaboration between geriatricians, cardiologists, and endocrinologists to ensure comprehensive care [11]. Current guidelines for SGLT2i therapy in elderly patients with DM may need to be reevaluated in light of these findings, particularly for the geriatric population [21].

Clinical implication

The optimization of pharmacological management for this complex patient population is becoming increasingly important as the global population ages and the prevalence of DM and AF in older adults continues to rise. Our research contributes to this objective by underscoring the importance of a nuanced, age-specific approach to the use of SGLT2i in geriatric patients with DM and AF. Meticulously assessing the potential benefits and risks of SGLT2i therapy based on the unique characteristics of each patient, such as age.

Limitations

The results of this study should be interpreted with caution due to several limitations. While the findings suggest a possible increase in stroke risk with SGLT2i use in patients aged ≥ 90 years, the small sample size and wide confidence intervals highlight the need for caution. More extensive confirmatory studies are required before definitive conclusions can be drawn. As a non-randomized, observational study, residual confounding and selection bias cannot be ruled out. Patients with coronary artery disease or myocardial infarction were more likely to receive SGLT2i, potentially reflecting physician preference for higher-risk patients. Although multivariable models adjusted for these factors, including the CHA₂DS₂-VASc score, unmeasured confounders such as physician treatment preferences, medication adherence, or undetected frailty may still influence the results.

This study did not differentiate between individual SGLT2i due to the reduced sample size when stratified by drug type. Variability in pharmacokinetics and pharmacodynamics among SGLT2i may influence outcomes and should be explored in future research.

We were also unable to evaluate other cardiovascular outcomes, such as mortality, due to database constraints. These constraints limit a comprehensive assessment of the overall safety and efficacy of SGLT2i, particularly in very elderly patients. Moreover, frailty and unknown multimorbidity, which are prevalent in older adults, were not quantitatively assessed or adjusted for in this study. These factors may interact with SGLT2i use and contribute to the observed outcomes in patients aged ≥ 90 years.

Future studies incorporating larger sample sizes, more complete datasets with mortality and cardiovascular endpoints, frailty indices, or comprehensive geriatric assessments are needed to better evaluate the risks and benefits of SGLT2i in this vulnerable population.

Conclusion

This study suggests that SGLT2i may have an age-dependent impact on stroke risk in patients with DM and AF, with a possible increased risk observed in individuals aged ≥ 90 years. These findings highlight the importance of considering age-specific factors in geriatric cardiology when prescribing SGLT2i. While SGLT2i remains a valuable therapeutic option for younger patients, its use in very elderly individuals should be approached cautiously and with close monitoring. Further studies are needed to confirm these findings and clarify the mechanisms underlying this age-dependent effect.