Efficacy and Safety of Dapagliflozin in the Elderly: Analysis From the DECLARE–TIMI 58 Study

OBJECTIVE:Data regarding the effects of sodium–glucose cotransporter 2 inhibitors in the elderly (age ‡65 years) and very elderly (age ‡75 years) are limited.RESEARCH DESIGN AND METHODS:The Dapagliflozin Effect on Cardiovascular Events (DECLARE)–TIMI 58 assessed cardiac and renal outcomes of dapagliflozin versus placebo in patients with type 2 diabetes. Efficacy and safety outcomes were studied within age subgroups for treatment effect and age-based treatment interaction.RESULTS:Of the 17,160 patients, 9,253 were <65 years of age, 6,811 ‡65 to <75 years, and 1,096 ‡75 years. Dapagliflozin reduced the composite of cardiovascular death or hospitalization for heart failure consistently, with a hazard ratio (HR) of 0.88 (95% CI 0.72, 1.07), 0.77 (0.63, 0.94), and 0.94 (0.65, 1.36) in age-groups <65, ‡65 to <75,and ‡75 years, respectively (interaction P value 0.5277). Overall, dapagliflozin did not significantly decrease the rates of major adverse cardiovascular events, with HR 0.93 (95% CI 0.81, 1.08), 0.97 (0.83, 1.13), and 0.84 (0.61, 1.15) in age-groups <65, ‡65 to <75, and ‡75 years, respectively (interaction P value 0.7352). The relative risk reduction for the secondary prespecified cardiorenal composite outcome ranged from 18% to 28% in the different age-groups with no heterogeneity. Major hypoglycemia was less frequent with dapagliflozin versus placebo, with HR 0.97 (95% CI 0.58, 1.64), 0.50 (0.29, 0.84), and 0.68 (0.29, 1.57) in age-groups <65, ‡65 to <75,and ‡75 years, respectively (interaction P value 0.2107). Safety outcomes, including fractures, volume depletion, cancer, urinary tract infections, and amputations were balanced with dapagliflozin versus placebo, and acute kidney injury was reduced, all regardless of age. Genital infections that were serious or led to discontinuation of the study drug and diabetic ketoacidosis were uncommon, yet more frequent with dapagliflozin versus placebo, without heterogeneity (interaction P values 0.1058 and 0.8433, respectively).CONCLUSIONS: The overall efficacy and safety of dapagliflozin are consistent regardless of age. INTRODUCTION Type 2 diabetes mellitus (T2DM) is a prevalent disorder in the elderly, with approximately one-quarter of people age .65 years with diabetes and an expected increase in rates of diabetes in the upcoming years (1). Diabetes care in the elderly is challenging due to high rates of concomitant comorbidities, func- tional disability, frailty, cognitive impair- ment, and polypharmacy. The complexity of treatment, side effects, and drug inter- actions are important considerations when choosing the appropriate glucose-lowering pharmacotherapy for older patients with diabetes (1,2). However, data regarding the efficacy and safety of glucose-lowering agents are often lacking, particularly in the very elderly, age $75 years. TheU.S. Food and Drug Administration aswell as the European Medicines Agency recommended collecting comprehen- sive data especially in very elderly patients with diabetes to enable ap- propriate assessment of their drug responses (2,3).Sodium–glucose cotransporter 2 (SGLT2) inhibitors have been available for the treatment of diabetes since 2012. Mul- tiple clinical benefits beyond glucose lowering have been established with this drug class. These include reduced hospitalizations for heart failure, renal protection, and improvements in weight and blood pressure (4–8). Furthermore, the drugs are taken orally and at any time of the day and have no known significant drug interactions (8). Considering their multiple favorable effects, minimal in- cremental risk for hypoglycemia, and simplicity of administration, they appear to be an attractive therapeutic option for older adults addressing the many comor- bidities more prevalent in this popula- tion. Nevertheless, there has been some hesitance in clinical practice prescribing these agents to the elderly, mostly due to insufficient long-term safety data (1). Older patients are more prone to the development of fractures and acute kid- ney injury (AKI), and safety alerts re- garding these potential risks with some SGLT2 inhibitors have been issued (9). Moreover, due to the limited therapeutic experience in patients age 75 years and older, initiation of SGLT2 inhibitor ther- apy at this age has so far not been rec- ommended by some authorities (10).The Dapagliflozin Effect on Cardiovas-cular Events (DECLARE)–TIMI 58 study is a cardiovascular (CV) outcome study that ascertained the CV and renal effects of dapagliflozin on a large patient pop- ulation both with and without established cardiovascular disease, including a large cohort of elderly and very elderly pa- tients (4). In the present analysis, we studied the efficacy and safety of dapa- gliflozin stratified by age.RESEARCH DESIGN AND METHODSStudy OverviewIn the DECLARE–TIMI 58 trial, a total of 17,160 patients, including 7,907 age $65 years and 1,096 age $75 years, with T2DM and established atherosclerotic cardiovas- cular disease or risk factors, 41% and 59%, respectively, were randomly assigned to receive dapagliflozin or placebo in addi- tiontostandard ofcare andfollowedfor a median period of 4.2 years. The study enrolled patients at least 40 years old, with HbA1c 6.5%212.0% and creatinine clearance $60 mL/min. Patients who re- mained eligible after a 4–8-week placebo run-in period were randomized in a 1:1 double-blind fashion to dapagliflozin 10 mg daily or matched placebo. All patients were to be treated according to regional stand- ards of care for CV risk factors: blood pressure, lipids, antithrombotic treat- ment, and HbA1c. The design, baseline characteristics, and principal results of this study have been published (4,11,12).Assessment of Outcomes The dual primary composite efficacy end points were CV death or hospitalization for heart failure (CVD/HHF) and major adverse cardiovascular events (MACE; the composite of CV death, MI, or ische- mic stroke). A secondary prespecified cardiorenal composite outcome was a sustained decrease of 40% or more in estimated glomerular filtration rate (eGFR) to ,60 mL/min/1.73 m2, new end-stage renal disease (ESRD), or death from renal or CV causes. A renal-specific composite outcome included a 40% decrease in eGFR to ,60 mL/min/1.73 m2, ESRD, or death from renal cause.Safety end points were assessed in all patients who received at least one dose of study drug (the safety population). For amputations, fractures, and malignancies, events ocurring from first dose until the end of trial were included in the analyses. The remaining safety outcomes were as- sessed on treatment, which included all events that occurred after the first dose of the study drug to the earlier of 30 (for serious adverse events [SAEs]) or 7 (for nonserious AEs) days after the last dose of the study drug or the closing visit. Major hypoglycemia was defined as symptomatic events requiring external assistance due to the severe impairment of consciousness with prompt recovery after glucose or glucagon administration. The urinary tract and genital infections collected were only those that were either serious or led to the discontinuation of the study drug.Prespecified age-groups were ,65, $65 and ,75, and $75 years. The data pre- sented in the main article are for three age-groups (,65, $65 to ,75, and $75 years) to enable the description of data from all age subgroups, including 65 to ,75 years, which comprised 39.7% of the study population. Efficacy and safety data limited to the prespecified age- groups are included in the Supplementary Material.Baseline characteristics are reported as frequencies and percentages for cat- egorical variables and as median and in- terquartile range (IQR) for continuous variables. Incidence rates and log-rank test for trend P values in efficacy and safety end points are reported for the three age- groups.Analyses were performed on an intention- to-treat basis. Hazard ratios (HRs) and 95% CIs were determined from Cox regression models with stratification factor (athero- sclerotic cardiovascular disease or multiple CV risk factors and hematuria status) as strata in models comparing treatment in age-groups.Mixed models for repeated measures in HbA1c, weight, systolic blood pressure, and diastolic blood pressure were analyzed to produce least squares mean estimates and 95% CIs in each treatment and age-group. Attainment of glycemic and weight targets was compared between groups using logistic regression models. P values for the covariate of interest were adjusted for baseline HbA1c or weight accordingly. There was no statistical adjustment for multiple comparisons. All analyses were performed using SAS software, version9.4 (SAS Institute Inc., Cary, NC) and Stata version 14.2 (College Station, TX). RESULTS Baseline Characteristics The study included 9,253 patients age ,65 years, 6,811 patients age $65 to ,75 years, and 1,096 patients age $75 years. Their baseline characteristics are shown in Supplementary Table 1. Chronic kidney disease and history of heart failure were more prevalent with increasing age as was use of ACE inhibitors or angiotensin receptor blockers (ARBs), diuretics, loop diuretics, and antiplatelet therapy.Overall incidence rates of the dual pri- mary composite efficacy end points and of their individual components were higher with increasing age. Incidence rates of CVD/HHF were 10.9, 14.8, and26.7 (P , 0.0001) and those of MACE were 20.9, 24.7, and 37.4 cases per 1,000 person-years in age-groups ,65, $65 to ,75, and $75 years, respectively (P , 0.0001).Dapagliflozin reduced the composite CVD/HHF consistently, with HR 0.88 (95% CI 0.72, 1.07), 0.77 (0.63, 0.94), and 0.94(0.65, 1.36) in age-groups ,65, $65 to ,75, and $75 years, respectively (interaction P value 0.5277). The HR for dapagliflozin for MACE was 0.93 (95% CI 0.81, 1.08), 0.97 (0.83, 1.13), and 0.84(0.61, 1.15) in age-groups ,65, $65 to ,75, and $75 years, respectively (interaction P value 0.7352). Rates of HHF were reduced with dapagliflozin, with HR 0.88 (95% CI 0.68, 1.15), 0.60(0.46, 0.79), and 0.81 (0.50, 1.30), re-spectively, in age-groups ,65, $65 to ,75, and $75 years (interaction P value 0.1402). The other components of the primary end points as well as all-cause mortality were unchanged, with effects consistent across all age-groups (Fig. 1). The cardiorenal secondary composite outcome (sustained decrease of 40% or more in eGFR to ,60 mL/min/1.73 m2, new ESRD, or death from renal or CV causes) was reduced with dapagliflozin versus placebo, with HR 0.72 (95% CI 0.59, 0.88), 0.80 (0.65, 0.98), and 0.82(0.52, 1.29) in age-groups ,65, $65 to ,75, and $75 years, respectively (interaction Pvalue 0.7299). Similar results were ob- served for the renal-specific composite outcome (Fig. 1). Efficacy end points by dichotomous age-groups showed similar results with no age-based treatment interactions (Supplementary Fig. 1).Changes in HbA1c, weight, and blood pressure by age subgroups and treat- ment allocation are shown in Fig. 2. Baseline HbA1c in the elderly and very elderly was lower compared with the younger population (median levels 8.2 [IQR 7.5, 9.3], 7.9 [7.3, 8.7], and 7.8 [7.2,8.5] in age-groups ,65, $65 to ,75, and $75 years, respectively; P-trend,0.0001). Nevertheless, significant and similar declines in HbA1c with dapagli- flozin versus placebo were observed for all age-groups (Fig. 2). At 1 year, least squares mean difference between the treatment groups was 20.58 (95% CI 20.63, 20.53), 20.46 (20.51, 20.41), and 20.51(20.63, 20.40) in age-groups ,65, $65 to ,75, and $75 years, respectively (all P , 0.0001). At 1, 2, and 3 years, patientsallocated to dapagliflozin versus placebo, at all age subgroups, were statistically Figure 1—Efficacy outcomes by age-groups. CVD, cardiovascular death. Figure 2—Changes in metabolic parameters over time. Change in HbA1c (A), weight (B), systolic blood pressure (C), and diastolic blood pressure (D) with dapagliflozin vs. placebo according to age-group. Dapa, dapagliflozin. more likely to attain an HbA1c of ,7.0% (excluding age $75 years at year 3) or ,8.0% or to reduce their HbA1cby $0.5% (Fig. 3). The effect was somewhat atten- uated at year 4, particularly in the very elderly.Dapagliflozin yielded a greater reduc- tion in weight versus placebo, and this was maintained in all age-groups during the entire study period (all P , 0.0001) (Fig. 2). Patients allocated to dapagliflo- zin versus placebo were more likely to attain a 5% weight loss at all age-groups, which was sustained throughout the study (Fig. 3). In the 4th year, a 5% weight reduction was observed in 37.0% vs. 23.2%, 40.5% vs. 24.6%, and 52.9% vs.31.2% with dapagliflozin versus placebo in age-groups ,65, $65 to ,75, and $75 years, respectively (all P , 0.0001).SAEs in the overall safety population were more common in the elderly and very elderly compared with the younger patients, with incidence rates of 107.3, 131.2, and 191.1 cases per 1,000 person- years in age-groups ,65, $65 to ,75, and $75 years, respectively (P , 0.0001). The incidence of SAEs was lower with dapagliflozin versus placebo overall in the trial, and this pattern was consistent regardless of age, with HR 0.93 (95% CI 0.86, 1.00), 0.88 (0.81, 0.95), and 1.02(0.85, 1.21) in age-groups ,65, $65 to ,75, and $75 years, respectively, with no age-based treatment interac- tion (interaction P value 0.2667) (Fig. 4). Moreover, no heterogeneity across age-groups was observed for any of the outcomes assessed, although the num- ber of events in the very elderly was often quite small, yielding wide CIs in this age category.Major hypoglycemia events increased with increasing age in the overall safety population, with incidence rates of 1.7, 2.6, and 6.5 cases per 1,000 person-years in age-groups ,65, $65 to ,75, and $75 years, respectively (P , 0.0001). Major hypoglycemia was less frequent with da- pagliflozin versus placebo, with the effect more predominantly observed in the age- group $65 vs. ,65 years (HR 0.53 [95%CI 0.34, 0.83] vs. 0.97 [0.58, 1.64], re-spectively; interaction P value 0.0896) (Supplementary Fig. 2). Overall fractures were more common in the elderly and very elderly, with incidence rates of 11.2, 15.8, and 17.4 cases per 1,000 person- years in age-groups ,65, $65 to ,75, and $75 years, respectively (P , 0.0001), yet events were balanced between the dapagliflozin and placebo groups at all age subgroups studied, with no hetero- geneity. Events of volume depletion in the overall safety study population in- creased with increasing age, with inci- dence rates of 5.6, 7.8, and 14.9 cases per 1,000 person-years in age-groups ,65,$65 to ,75, and $75 years, respectively(P , 0.0001). Similarly, AKI was reported overall at higher rates with increasing age, with incidence rates of 4.2, 5.4, and 9.3 cases per 1,000 person-years in age- groups ,65, $65 to ,75, and $75 years, respectively (P 5 0.0001). Vol- ume depletion events were balanced between the dapagliflozin and placebo groups, and AKI events were overall fewer with dapagliflozin versus placebo, with no age-based treatment interaction. The amputation rate did not differ by age (P 5 0.3201) and was balanced between dapagliflozin and placebo, with no age- based treatment interaction. Diabetic ke- toacidosis was rare, but more events were observed with dapagliflozin versus pla- cebo, consistently across age-groups. Genital infections that led to the discon- tinuation of the study drug were more common with dapagliflozin versus placebo, and there were two SAE genital infections in each treatment arm, with no heteroge- neity. There was no statistically significant increase in urinary tract infections (serious Figure 3—Attainment of HbA1c and weight targets. Percentage of patients attaining HbA1c ,8 (A), HbA1c ,7 (B), reducing HbA1c by .0.5% (C), or reducing weight by .5% (D) with dapagliflozin vs. placebo by age-group. Black bars, dapagliflozin; white bars, placebo. *P , 0.0001; 1P , 0.05. or leading to drug discontinuation) with dapagliflozin versus placebo in the over- all population or in any of the age-groups. Overall malignancies were balanced be- tween treatment arms across all age- groups (Fig. 4).Safety end points by dichotomous age- groups revealed consistent results with no age-based treatment interaction for any ofthe outcomes (Supplementary Fig. 2). CONCLUSIONS In this manuscript, we present analyses of data from the DECLARE–TIMI 58 study that establish the beneficial CV and renal effects of dapagliflozin in a robust num- ber of elderly and very elderly partici- pants. The overall pattern of efficacy and safety of dapagliflozin was consistent regardless of age.Care of older patients with diabetes represents an ongoing challenge. Rates of HHF, CV disease, and MI are increased in those age $65 years and are markedly increased in patients with age $75 years. The robust reduction in HHF observed with dapagliflozin is thus of great clinical significance, particularly when viewed in light of the fact that many of these patients were already treated with stan- dard of care, including ACE inhibitors, ARBs, b-blockers, and diuretics. Although fewer events of MACE were observed in patients treated with dapagliflozin, it did not result in a significant reduction in the incidence of MACE. The study met its primary safety noninferiority end point for MACE across all studied age-groups. SAEs, though generally more frequent in older versus younger individuals, were not increased in the elderly or the very elderly with dapagliflozin versus placebo, and there was no age-based treatment interaction for any of the safety outcomes assessed. Several adverse outcomes are of particular concern in older patients with diabetes. Events of volume deple- tion increase with increasing age, and there is greater concern related to pos- sible adverse consequences of volume depletion in older patients, including falls and kidney injury. In that respect, the observed reduction in HHF and renal ben- efit with dapagliflozin with no excess risk of volume depletion and fewer events of AKI is reassuring. Older adults are at higher risk of hypoglycemia from both insulin and sulfonylurea treatment as a result of insulin deficiency, progressive renal insufficiency, and higher rates of cognitive deficits, which may cause diffi- culty in disease management, i.e., glucose monitoring and adjustment of insulin dos- ing (1). Hypoglycemia should be particu- larly avoided in older patients due to their greater risk of other major adverse out- comes secondary to hypoglycemia, such as falls or fractures (1). Patients ran- domized to dapagliflozin versus placebo in addition to standard of care attained superior glycemic control with lower rates of hypoglycemia, irrespective of age, high- lighting the benefit of the drug.The greater morbidity of the elderly and very elderly population in our study, as reflected by higher rates of baseline HF and chronic kidney disease, strengthens the clinical impact of our results, which demonstrate consistent efficacy and safety of dapagliflozin extending across all age- groups.The paradigm of diabetes treatment has shifted from a glucose-focused ap- proach to the pursuit of a therapeutic regimen that will yield a reduction in morbidity and mortality. This is particu- larly true in the elderly, whose life ex- pectancy is shorter and for whom event rates are higher (13). Treatment goals for T2DM in the elderly should be individ- ualized; thus, in healthy patients with good functional status, few comorbid- ities, and intact cognitive function, goals Figure 4—Safety outcomes by age-groups. DKA, diabetic ketoacidosis. may be similar to those of younger adults. However, in patients with an intermediate remaining life expectancy, targets should be less stringent as part of individualized care and a target HbA1c ,8.0% may be acceptable (13). Dapagliflozin enabled more patients to attain a clinically sig- nificant HbA1c reduction, whether striving for the standard or less stringent target.Weight loss is generally not a treat- ment goal in the very elderly population; nevertheless, in our analysis, we observed sustained weight loss across all age- groups. This did not appear to lead to any untoward effects in the elderly during the time frame of the study, although notably, median BMI was 31.1 and30.2 kg/m2 at baseline in the elderly and very elderly patients in the study, re- spectively. SGLT2 inhibitors have been shown to reduce adipose tissue mass while maintaining lean body mass (14), changes which are probably beneficial at all ages.Few studies have been published to date on the use of SGLT2 inhibitors in the elderly. A randomized, double-blind, age- stratified trial of dapagliflozin versus pla- cebo demonstrated a beneficial effect of dapagliflozin onglucose, weight, andblood pressure across all age-groups studied, with no outstanding safety issues in the elderly or very elderly, although not many very elderly patients were included(15). Postmarketing reports from Japan reported no age-related safety issues with tofogliflozin, ipragliflozin, and can- agliflozin, yet these studies are limited by the lack of comparator and dependence upon physician reporting of AEs (16–18). The efficacy and safety of other glucose- lowering agents in the elderly have been studied. Dipeptidyl peptidase 4 (DPP- 4) inhibitors showed no CV benefit in any age-group studied, and safety outcomes of DPP-4 inhibitors were shown to be similar in older versus younger patients (19,20). GLP-1 receptor agonists have also shown efficacy and safety in the elderly that are comparable to that of younger patients, and post hoc analysis of the Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial pro- posed greater benefit in the elderly (21). Our study is the first to substantiate the efficacy and safety of an SGLT2 inhibitor in the elderly and very elderly and may pave the way for relaxing the warning placed on this drug in the geriatric pop- ulation, although individualization of therapy is pivotal, particularly in this vulnerable population.Some limitations of our study should be noted. Creatinine clearance ,60 mL/ min was an exclusion criterion in the study, and this may have led to exclusion of the frailer elderly patients, which are more prone to volume depletion, AKI, fractures, and other adverse outcomes. Moreover, there were no assessments of cognitive function, functional capacity, or frailty at baseline or at any time during the study; however, as a randomized trial, one would expect that these un- measured features would be balanced between groups. Additionally, the very elderly subgroup consisted of a small subset (6.4%) of the total population of 17,160, and the number of events was small, yet it still accounted for 1,096 participants. Finally, the analyses of met- abolic outcomes were post hoc and should be considered exploratory. In conclusion, our study establishes the CV and renal benefits of dapagliflozin in the elderly and very elderly. The overall efficacy and safety of dapagliflozin were consistent regardless of age; thus, this drug may be considered a valuable glucose- lowering agent regardless of age.