FormalPara Key Summary Points

Why carry out this study?

Even though the long-term effectiveness and acceptable safety profile of SGLT2 inhibitors were reported in previous RCTs, there is a lingering concern about the safety profile of these drugs in real-world practice, particularly in the more elderly (≥ 75 years old) patient population.

What was learned from the study?

The long-term therapeutic effectiveness of SGLT2 inhibitors was comparable with prior observations in RCTs, but the frequency of adverse events and the discontinuation rate in the elderly (≥ 75 years old) population with T2D were significantly higher than those in adults (< 75 years old) with T2D.

Introduction

Sodium-glucose co-transporter 2 (SGLT2) inhibitors have dramatically changed the treatment strategy of patients with type 2 diabetes (T2D) following the completion of several seminal clinical trials, including EMPA-REG OUTCOME (Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients) and DECLARE-TIMI 58 (Dapagliflozin Effect on Cardiovascular Events-Thrombolysis in Myocardial Infarction 58). These studies provided clear evidence for SGLT2 inhibitors to promote protective cardiovascular (CV) effects in patients with T2D [1, 2], while providing more optimal control of their glycemic index with a persistent reduction in their body weight [3, 4]. In addition, SGLT2 inhibitors could also slow the progression of diabetic kidney disease and reduce the hospitalization rate as a result of heart failure (HHF) [5, 6]. Because of these beneficial findings attributed to SGLT2 inhibitors, the prescription of this class of drugs is now the first-line choice to reduce cardiorenal adverse events in high-risk patients with T2D [7].

The drug approval process is based on the results obtained through the design and implementation of the randomized controlled trial (RCT), which is dependent upon its excellent internal validity. Patients are randomized to minimize bias and ensure that the only variable between treatment arms is their exposure to the treatment of interest. For these reasons, the drug efficacy and safety profiles in a specific population can be highly reliable in this setting. On the other hand, the findings from these RCTs may not be directly applicable to the real-world clinical practice, because of the potential differences in the patient pool and practitioners. In addition, the patients with one or more comorbidities, particularly in the elderly, are generally not included in RCTs [8, 9].

As a result, there have been several real-world clinical studies, including those from our lab, demonstrating some distinct differences in the safety profile of SGLT2 inhibitors between those described in the RCTs. Notably, the most common type of adverse events (AEs) was hypoglycemia in the real-world setting compared to genitourinary infections in the previously published RCT [10,11,12,13]. The differences in safety profiles may be compounded in other special populations, especially in the more elderly patients with T2D. Moreover, our lack of knowledge about the dropout rate or the reason for drug discontinuation during the course of treatment needs further investigation.

In this present study, we performed a retrospective, electronic medical record (EMR)-based, real-world clinical analysis to investigate the efficacy and safety of SGLT2 inhibitors in a population of varying aged adults (< 75 years old) or elderly (≥ 75 years old) patients with T2D.

Methods

Study Design and Participants

In this study, we evaluated medical records of 715 patients with T2D whose prior drug regimen was modified with the addition of SGLT2 inhibitors (either dapagliflozin or empagliflozin) to improve blood glucose control in the Endocrinology Department at Chungbuk National University Hospital between 2015 and 2019. Among this group of candidates, patients were excluded if they failed one of the following criteria: (1) no follow-up visit occurred after the initial meeting, (2) refusal by the patient in the follow-up period of 4–12 weeks, (3) lack of any follow-up laboratory data, (4) ongoing anticancer treatment, (5) patients diagnosed with type 1 diabetes. After assessing the criteria, 119 patients were excluded and the remaining 596 (n = 596) patients with T2D were allowed to be included in this study. Basal data used for stratification were age, sex, duration of diabetes, family history of diabetes and smoking, type of SGLT2 inhibitor being administered, presence of hypertension (HTN) or dyslipidemia, history of coronary artery disease (CAD), heart failure (HHF), or cerebral stroke, topographic measurement including height (Ht), body weight (Bwt), systolic blood pressure (SBP), diastolic blood pressure (DBP), and current use of hypoglycemic agents, including metformin, sulfonylurea (SU), dipeptidyl peptidase 4 (DPP4) inhibitor, thiazolidinedione (TZD), alpha1-glucosidase inhibitor (AGI), or insulin. The basal laboratory data used for analysis included glycated hemoglobin (HbA1c), fasting plasma glucose (FPG), lipid profile, estimated glomerular filtration rate (eGFR), creatinine, fasting C-peptide and insulin.

Outcome Measures

The primary objective in this study was to monitor changes of the HbA1c from basal (time 0) to 36 months at each 6-month interval visits. Secondary objectives were evaluated to determine (1) changes of FPG or Bwt from basal to every 6-month visit throughout the 36-month period; (2) frequencies of AEs, including hypoglycemia (overall and severe), genitourinary tract infection (GUI), polyuria including nocturia (> 3 times in night), gastrointestinal (GI) troubles; (3) frequencies and causes of discontinuation of SGLT2 inhibitor; (4) frequencies of major adverse cardiovascular events (MACE) or death for whole study period. All of the topographic and laboratory data were obtained from patient records after each visit. AEs and causes for SGLT2 inhibitor discontinuation were collected from notations made from the notes by the committed physician, laboratory data, and/or other notes from the emergency room or other departments (e.g., hemato-oncology or cardiology) during the follow-up period. For the subgroup analysis, all included patients were divided into elderly (≥ 75 years old) or adult (< 75 years old) patients. The frequencies of AEs and discontinuation were compared between elderly and adult patients. On the index day of enrollment, other than the initial addition of the SGLT2 inhibitor, no other drugs were added to prior regimen. During the follow-up period, no new drugs were added to prior regimen and none of the prior drug dosing schedules for metformin (1000–2000 mg), DPP4 inhibitor (maximal dose), SU (tolerable dose: glimepiride 2–8 mg), or insulin were changed except when hypoglycemia developed. The patients taking multiple oral hypoglycemic agent (OHAs) with poor glucose control even after the addition of a SGLT2 inhibitor were generally treated with insulin without administering an SGLT2 inhibitor. In this situation, the cause of SGLT2 discontinuation of the index case was defined as inadequate control.

Statistical Analyses

Statistical analyses were done in SPSS for Windows (version 24.0, IBM Corp., Armonk, NY, USA). Continuous variables and discrete variables are expressed as the mean ± standard deviation or standard error, and number (%), respectively. The Kolmogorov–Smirnov test for normality was performed to achieve the adequate statistical test on continuous variables. For the analysis of the difference of HbA1c, FPG, and Bwt between basal and those of each 6-month visit, paired Student t test was performed and a two-sided significance level of 0.05 was used as being significant. The last observation carried forward approach was used to impute missing data. AEs and causes of discontinuation were summarized as number with percentage calculated as the total number of subgroup patients. The incidence of AEs or discontinuation between elderly (≥ 75 years old) and adult (< 75 years old) patients was compared using the chi-square test and P < 0.05 was considered significant. The incidence of MACE or death in all patients was descriptively expressed.

Ethical Approval

The study was carried out in accordance with the principles stated in the Declaration of Helsinki, as revised in 2013 and the International Conference of Harmonization/Good Clinical Practice guidelines [14, 15]. An institutional review board (IRB) at Chungbuk National University Hospital approved the study (No. 2024-01-010-001, and written informed consent from patients to participate was waived owing to the study being based on retrospective data collection.

Results

Patient Baseline Characteristics

Table 1 summarizes all of the clinical and laboratory basal characteristics in the patients with T2D (n = 596) who were identified being treated with their usual drug regimen plus the addition of either one of the two SGLT2 inhibitors, dapagliflozin or empagliflozin. Of the 596 enrolled patients, dapagliflozin and empagliflozin was added in 279 (46%) and 317 (54%) patients, respectively. In terms of other medications being combined with the SGLT2 inhibitors, insulin was used in 10% of the patients (n = 62) whereas the majority of the patients (90%) used various number(s) of distinct OHAs. More specifically, 5% of the patients used two OHAs (n = 31), 18% used three OHAs (n = 111), 64% (n = 384) used four OHAs, and only 1% (n = 5) used five OHAs. The mean age of the patients was approximately 60 years old with the majority being male (60%). The mean body mass index (BMI) was 27 kg/m2, the mean duration of diabetes was over 14 years, and more than half of the patients also exhibited hypertension or dyslipidemia. Among the patients, 153 (25%) patients had existing cardiovascular diseases; 24 (4%) patients showed heart failure and 40 (6%) patients had a stroke.

Table 1 Baseline clinical and laboratory characteristics of study subjects (n = 596)
Full size table

Effectiveness

Over the follow-up period, there was a significant reduction in the mean HbA1c levels starting at the first 6-month follow-up visit, 7.5 ± 1.1% (P < 0.01), compared to the basal (initial visit level of 8.8 ± 1.3%), and remained at this lower HbA1c level at each subsequent 6-month visit throughout the 36-month observation period. More specifically, the mean HbA1c levels at 12 months (7.6 ± 1.1%, P < 0.01), 18 months (7.7 ± 1.1%, P < 0.01), 24 months (7.7 ± 1.1%, P < 0.01), 30 months (7.6 ± 1.0%, P < 0.01), and 36 months (7.7 ± 1.0%, P < 0.01) were significantly lower than the basal measurement of the HbA1c levels (Fig. 1a). Similar to the HbA1c levels, the FPG levels also decreased from basal (180 ± 60 mg/dL) to the 6-month (135 ± 42 mg/dL, P < 0.01) visit, and remained significantly lower until the end of the 36-month study period; 12 months (138 ± 45 mg/dL, P < 0.01), 18 months (136 ± 36 mg/dL, P < 0.01), 24 months (140 ± 40 mg/dL, P < 0.01), 30 months (140 ± 42 mg/dL, P < 0.01), and 36 months (138 ± 38 mg/dL, P < 0.01) (Fig. 1b). Another beneficial effect that was observed was a decrement in the patient body weight. The body weight at the initial visit averaged 74 ± 15 kg, which significantly fell to 71 ± 15 kg (P < 0.01) at the 6-month visit. The decreased body weight remained significantly lower than the basal (initial measurement) through each of the subsequent follow-up visits, specifically at 12 months (71 ± 15 kg, P < 0.01), 18 months (72 ± 14 kg, P < 0.01), 24 months (72 ± 14 kg, P < 0.01), 30 months (72 ± 16 kg, P < 0.01), and the final visit at 36 months (72 ± 15 kg, P < 0.01) (Fig. 1c).

Fig. 1
figure 1

Mean changes in a HbA1c, b FPG, and c body weight over the entire patient study period. Time = 0 is the baseline point. Data are expressed as a, b mean ± SD and c mean ± SEM. *P < 0.001, significant difference between basal and each time point. N number of tested patients at each point, Rate number of tested patients at each point/number of tested patients at baseline, FPG fasting plasma glucose, HbA1c glycated hemoglobin

Full size image

In a subgroup analysis to determine the efficacy of the SGLT2 inhibitors in the elderly (≥ 75 years old) patients with T2D, the mean HbA1c levels were significantly (P < 0.01) reduced from basal (8.8 ± 1.1%, n = 57) to (8.0 ± 1.3%, n = 26) at the end of the 36-month study period. Similarly, the FPG levels were significantly reduced (P < 0.05) from the basal (165 ± 53 mg/dL) to 36-month period (143 ± 47 mg/dL).

Safety and Tolerability

All reported AEs are summarized in Table 2. The most common AE was hypoglycemia (n = 57), including one patient that was admitted to hospital because of severe hypoglycemia, but recovered without any sequelae. All the cases of hypoglycemia developed only in the patients taking three or more OHAs or only insulin. The second most common AE was genitourinary infection (GUI) (n = 31), which was approximately 45% less frequent than hypoglycemia, in which two patients were admitted to hospital because of severe acute pyelonephritis (APN), treated with antibiotics, and then discharged with complete recovery. The two other AEs that were commonly observed were polyuria (n = 28), including urinary frequency or nocturia, and gastrointestinal problems (n = 19), including constipation. In a subgroup analysis, the incidence of AEs (n = 18, 31%) in elderly (≥ 75 years old) patients (n = 57) was significantly higher than AEs (n = 117, 21%) in adult (< 75 years old) patients (n = 539) (P < 0.001).

Table 2 Adverse events during the patient study period
Full size table

In the follow-up period, 211 (35.4% of total number) patients dropped out of the trial or completely discontinued the SGLT2 inhibitors (Table 3), the most common reason being loss to follow-up (n = 70). The other reasons for their discontinuation were based on their inability to adequately control glucose levels (n = 26), excessive weight loss (n = 20), hypoglycemia (n = 19), and GUI (n = 19). In a subgroup analysis, the frequency of discontinuation (n = 30, 52%) in elderly (≥ 75 years old) patients (n = 57) was significantly higher than (n = 181, 33%) adult (< 75 years old) patients (n = 539) (P < 0.001). Similar to the overall combined patient group, the most common causes of discontinuation in elderly patients were loss to follow-up (n = 8), inadequate ability to control their blood glucose levels (n = 6), excessive weight loss (n = 3), hypoglycemia (n = 3), and GUI (n = 3).

Table 3 Reasons for drug discontinuation during patient follow-up period
Full size table

Table 4 summarizes the causes of death and other CV events observed during all of the follow-up periods. During this study period, five patients died because of a cardiac arrest (n = 2), cancer (n = 2), or hepatic failure (n = 1). Other notable CV events in other patients during this study period that were non-fatal included myocardial infarction (n = 12), stroke (n = 2), and hospitalization due to heart failure (n = 2).

Table 4 Cardiovascular (CV) events and causes of death during patient follow-up period
Full size table

Discussion

The present study provides a retrospective analysis regarding the long-term efficacy and safety of SGLT2 inhibitors in a real-world clinical practice. Consistent with the observations from prior RCTs [16,17,18,19], our overall patient population exhibited promising T2D management outcomes with respect to the long-term control of their glycemia and persistent body weight reduction upon the use of SGLT2 inhibitors. Within our overall patient group, the therapeutic effects of SGLT2 inhibitors in the elderly (≥ 75 years old) patients were equally as efficacious and sustainable in the improvement of glucose control compared to the general adult population, with the caveat for only those who were able to tolerate the SGLT2 inhibitor. This result was consistent to long-term (> 4 year) follow-up data in the elderly from the DECLARE-TIMI 58 study [20], in which HbA1c was reduced and maintained at a lower level for over 4 years in all age groups of < 65, 65 to  < 75, and ≥ 75 years old patients with T2D.

However, the positive therapeutic effects of SGLT2 inhibitors need to be tempered with a cautionary note that the use of these drugs tended to increase the prevalence of AEs or the discontinuation of the drugs in patients with T2D older than 75 years of age. Although the occurrence of AEs is not a new finding for SGLT2 inhibitors in patients, the prevalence regarding the type of AE needs further evaluation beyond those observed in RCTs and now in phase IV post-marketing surveillance studies. In our study, hypoglycemia was the most common AE due to the addition of SGLT2 inhibitors to their existing drug regimen of either three or more OHAs or insulin alone. The appearance of hypoglycemia as the major AE in our study population is consistent with another post-marketing study evaluating tofogliflozin [21]. In that study, patients with T2D already being treated with three different OHAs, namely metformin (47.0% of included patients), sulfonylurea (SU) (28.7%), and DPP4 inhibitor (60.3%), exhibited a significantly (P < 0.001) higher incidence of hypoglycemia versus those patients taking only two different drugs [21]. Other real-world data in patients with uncontrolled T2D have also reported that hypoglycemia is their most common AE compared to GUI (2.6%) in a 24-month study period [11]. Moreover, GUI was not universally observed as the most common AE in all RCTs, and hypoglycemia was noted as their top AE in a randomized trial evaluating long-term efficacy of dapagliflozin versus placebo in insulin-treated patients with diabetes [10]. The reasons for the onset of hypoglycemia remains to be fully described, but in some cases, it may be attributed to the combined use with other OHA, such as DPP4 inhibitor or SU [22]. Unlike the prior RCTs that exclude the use of either DPP4 inhibitors or SU in their treatment arms [17,18,19], the majority of the patients evaluated in the study by Salvo et al. [22] were already being treated for their T2D using a combination of either metformin, a DPP4 inhibitor, or SU prior to the addition of SGLT2 inhibitor. Because of the difference in drug combinations prior to supplementing with SGLT2 inhibitors, this has been postulated as one of the major contributing factors that lead towards the onset of hypoglycemia in inadequately controlled patients with T2D [10,11,12,13]. Even though hypoglycemia was a common AE, nearly all of these cases were only categorized as minor, except for one incidence where the patient was discovered unconscious, transferred to a hospital, and fortunately recovered without any sequelae. In general, it is imperative to immediately adjust the dosing of the other OHAs or insulin and/or discontinue the use of the SGLT2 inhibitor during situations where the patients exhibit severe hypoglycemia.

Upon further examination of specific age groups (adults < 75 versus elderly ≥ 75 years old) in our study population, we determined that there was a notable increase by nearly 50% (P < 0.001) in the frequency of AEs (31%) in the elderly patients (≥ 75 years old) compared to the adult group (21% in < 75 years old) There was a similar rise in the frequency of other AEs listed in Table 2 in the elderly compared to their relatively younger patients. Our retrospective analyses are comparable with other studies using long-term SGLT2 inhibitors in patients with T2D. In the DECLARE-TIMI study using dapagliflozin, elderly (≥ 75 years old) patients with T2D exhibited the incidence of AEs, including major hypoglycemia, volume depletion, UTI, and cancer, at a higher incidence than younger (< 75 years old) adults [20]. To compare the incidence of AEs or causes of discontinuation between our result and previous study, the incidence of hypoglycemia, volume depletion such as weight loss, and GUI was higher in elderly (≥ 75 years old) patients in both studies, but oncogenesis was not observed in our study. In another study evaluating empagliflozin in elderly Asian patients with T2D, the incidence of AEs (8 cases/30 of total patients, 26.6%) in elderly (≥ 75 years old) patients was approximately 70% higher than observed in adult (< 75 years old) patients (6 cases/38 of total patients, 15.7%) [23].

As far as we currently know, there is limited data to report the dropout or discontinuation rates during our study period. In a post-marketing surveillance study of tofogliflozin over a 36-month period in Japanese patients with T2D, the rate of discontinuation treatment was fairly high at 41.0%. The reasons of discontinuation were listed from the highest to lowest reasons as follows: stopped visits (14.3%), development of AEs (7.2%), request by the patient (5.5%), insufficient treatment response (5.4%), medical decision by the clinician (4.3%), cure (2.0%), and other miscellaneous factors (2.0%) [24]. In our study, the total discontinuation rate was relatively close at 35.4% to the published findings in the Japanese T2D study, and the stopped visits to the physician were the same most common cause of discontinuation. Moreover, the discontinuation rate in elderly patients (≥ 75 years old) was apparently higher than that in the adult (< 75 years old) patients with T2D. The reason for elderly people exhibiting a greater prevalence of AEs or deciding upon discontinuing SGLT2 inhibitor use more frequently than younger adults remains to be determined. From our findings in the present study, elderly patients (n = 30) who dropped out during the follow-up period tended to exhibit a higher basal HbA1c (8.9% vs 8.6%), and a longer duration of diabetes (21 vs 18 years) compared to the elderly patients that faithfully maintained follow-up visits throughout the 36-month period (n = 27) although these values were not statistically significant. These findings may suggest that elderly patients with T2D who are willing to maintain regular visits to their physician to monitor their physical conditions while taking SGLT2 inhibitors could better control their glycemia and possibly reduce their likelihood of having an AE or wanting to discontinue this medication.

As with all retrospective study designs and data analysis, a limitation to our proposed conclusions from our findings may be attributed to the relatively small population of patients that were identified and included from our single clinic. In our study, we collected data about AEs or cause of discontinuation from notations made at the time of admission or surgery or directly from the committed physician during the follow-up period. Even though from our knowledge the most common cause of patient discontinuation due to the use of SGLT2 inhibitors was not returning back to our clinic, it remains unknown whether these patients decided to continue with the use of these drugs by acquiring them from other sources. Also, we cannot confirm with absolute certainty whether the use of non-diabetic drugs was an initiating or exacerbating factor in the onset of AEs. With this in mind, a larger prospective multicentered study is necessary to confirm our present results within and outside of our geographical region.

Conclusion

The present study provides a detailed overview of the real-world utilization of SGLT2 inhibitors in a diverse T2D population at a single South Korean hospital clinic. While affirming the sustained efficacy of SGLT2 inhibitors to treat T2D, our findings underscore the potential importance of age-specific considerations when prescribing SGLT2 inhibitors. Our findings have identified that elderly patients (≥ 75 years old) need to have a more rigorous monitoring of their potential AEs to help reduce the chances of developing more serious medical issues or leading to premature death. It will be imperative for healthcare providers to actively engage with these elderly patients to ensure that there is shared decision-making upon the use of these drugs, and that transparent and candid discussions are provided by the patients to their physicians to better develop an individualized drug regimen based on their drug tolerances and individual patient characteristics, including their advanced age, to optimize the personal benefits of SGLT2 inhibitors in T2D management. Further research and long-term follow-up visits are warranted to validate these findings in other clinical practice sites to determine whether other race and ethnicities exhibit similar outcomes, and this would help to refine our understanding of the safety and efficacy of SGLT2 inhibitors in routine clinical practice.