Barriers to initiating SGLT2 inhibitors in diabetic kidney disease: a real-world study

Background Sodium-glucose cotransporter 2 inhibitor (SGLT2i) should be considered for patients with type 2 diabetes (T2D) and chronic kidney disease (CKD) having estimated glomerular filtration rate (eGFR) ≥ 30 mL/min/1.73 m2 and urine albumin-to-creatinine ratio (UACR) > 30 mg/g. However, SGLT2i is currently underprescribed among eligible, at-risk patients for CKD progression. We analyzed prescription patterns and barriers to initiating SGLT2i in patients with T2D and CKD in real practice. Methods A total of 3,703 consecutive outpatients with T2D from four teaching hospitals during six months (2019 ~ 2020) were reviewed. Five eGFR categories (G1, ≥ 90; G2, 60–89; G3ab, 30–59; G4-5, < 30 mL/min/1.73 m2) and three UACR categories (A1, < 30; A2, 30–300; A3, > 300 mg/g) were used to define CKD status. Results Overall, 25.8 % patients received SGLT2i in the following eGFR and albuminuria categories: G1 (A1, 31 %; A2, 48 %; A3, 45 %); G2 (A1, 18 %; A2, 24 %; A3, 30%); and G3 (A1, 9 %; A2, 7 %; A3, 13 %). Total prevalence estimate of CKD was 33.8 % (n = 1,253), of whom 25.6 % patients received SGLT2i. We defined eGFR ≥ 45 mL/min/1.73 m2 and UACR ≥ 30 mg/g as high-risk CKD group eligible for SGLT2i (n = 905), of whom 32.9 % patients were treated with an SGLT2i. In this high-risk group, SGLT2i initiation showed negative correlations with age ≥ 65 years and recent hospitalization. Conversely, HbA1c level, body mass index (BMI), presence of diabetic retinopathy, and previous heart failure events were positively correlated with SGLT2i initiation. Conclusions Only 32.9 % of T2D with CKD eligible for SGLT2i is currently treated with SGLT2i in real-world clinical practice. The older patient group and clinical inertia are the main barriers to initiate SGLT2i for eligible patients. Clinicians should change the glucocentric approach and focus on reducing renal events in T2D. Supplementary Information The online version contains supplementary material available at 10.1186/s12882-021-02381-3.


Background
Chronic kidney disease (CKD) attributed to diabetes (diabetic kidney disease) occurs in 20-40 % of patients with diabetes [1,2]. Diabetic kidney disease (DKD) is a clinical diagnosis made based on the presence of albuminuria and/or reduced estimated glomerular filtration rate (eGFR) [2]. It is caused by hyperglycemia, hypertension, aging, and other risk factors of chronic kidney disease. Although the prevalence of DKD is increasing, there are few medications to treat or slow its disease course.
However, real-world evidence demonstrates that SGLT2i is currently underutilized for eligible patients and that the decision to start SGLT2i is typically deferred to the endocrinologist [6][7][8]. Although recent studies have reported that the use of SGLT2i reached 6.2 % in T2D and CKD, they did not clarify the barriers against initiation of SGLT2i, including clinical factors such as the impact of glycemic control [9,10].
Thus, the objective of this study was to determine the proportion of patients with T2D who received SGLT2i and factors limiting the initiation of SGLT2i in patients with eGFR ≥ 45 mL/min/1.73 m 2 and UACR ≥ 30 mg/g in diabetes clinics from teaching hospitals.

Study subjects
The present study was based on data obtained from four teaching hospitals in Seoul Metropolitan Area from September 2019 to May 2020 because drug labeling for SGLT2i was extended to eGFR ≥ 45 mL/min/1.73m 2 in August 2019 after the CREDENCE trial released in June 2019 [3]. We excluded patients with systemic conditions that might affect vascular glomerulonephritis or vasculitis, patients with other reasons of renal dysfunction, and patients with acute kidney injury (AKI).
Of 4,186 consecutive subjects visiting each diabetes clinic, subjects who were prescribed SGLT2i from outside hospitals (n = 74), those who used glucagon-like peptide − 1 receptor agonist (GLP-1 RA) before initiation of SGLT2i (n = 163), and adults without measurements of serum creatinine levels (n = 57) or UACR (n = 189) were excluded. After these exclusions, 3,703 subjects were included in the final analysis (Fig. S1). This retrospective cross-sectional study was approved by the Institutional Review Board (IRB) of Dongguk University Ilsan Hospital and participating hospitals (IRB No. 2020-03-052), and informed consent was waived.

Demographic, anthropometric, and clinical characteristics
To analyze prescription patterns according to CKD status, patients were divided into two groups: SGLT2i nonusers and SGLT2i initiators. Clinical characteristics were collected based on the time of the last outpatient department visit. However, for SGLT2i initiators, age, body mass index (BMI), diabetes duration, duration of atherosclerotic cardiovascular disease or heart failure (collectively called CVD-HF), and recent hospitalization data were collected based on the time of SGLT2i initiation. The definition of recent hospitalization was hospital admission within one year of the investigation, excluding minor procedure hospitalization (for the SGLT2i initiators, before initiation of SGLT2i).
Comorbidities were determined based on chart review. We defined previously documented history of atherosclerotic cardiovascular diseases (ASCVDs), including coronary artery diseases (CAD; chronic coronary syndrome, acute coronary syndrome and/or coronary artery revascularization), cerebrovascular diseases (cerebral infarction and/or transient ischemic attack), peripheral artery diseases (peripheral artery occlusive disease and/ or lower limb revascularization), and heart failure (HF) according to the International Classification of Diseases (ICD-10). The occurrence of diabetic retinopathy (DR) was determined by an ophthalmologist. Medication histories regarding anti-hypertensive drugs including RAASi and lipid-lowering drugs were also reviewed. For glucose-lowering drugs, data of nonusers were based on the last prescription. Glucose-lowering drugs of SGLT2i initiators were based on the last prescription just before switching to SGLT2i.

Laboratory data
Most current laboratory data of subjects were recorded within two years if available. For SGLT2i initiators, HbA1c and eGFR values were based on the time of SGLT2i initiation. Lab data were obtained from each hospital. eGFR was calculated from the serum creatinine level standardized to IDMS using the Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) equation [11].

Statistical analysis
Patients initiating SGLT2i (SGLT2i initiators) during the study period were compared with SGLT2i nonusers. These two groups were mutually exclusive. Data were described as mean ± SD or as percentages and number of cases. Student's t-test or Kruskal-Wallis one-way analysis of variance was used for continuous variables while Pearson Chi-square test was used for categorical variables. Predictive factors for SGLT2i initiation were obtained using univariate and multivariate logistic regression analyses. Analyzed results were described as odds ratios with their 95 % confidence intervals. Statistical significance was considered when p value was less than 0.05. All statistical analyses were performed using R Statistical Software version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).
Cancer patients or recently hospitalized patients were less likely to be started on an SGLT2i, but patients with diabetic retinopathy were more likely to be started on an SGLT2i(37.6 vs. 28.3 %; p = 0.019). Likewise, the SGLT2i initiator group had comparable treatment rates with RAASi (75.5 vs. 68.5 %; p = 0.066) but higher insulin use rates before switching to SGLT2i (30.9 vs. 21.1 %; p = 0.008).
Factors related to failure to start SGLT2i in those with high-risk CKD group eligible for SGLT2i Logistic regression analyses revealed that SGLT2i initiation had negative correlations with age ≥ 65 years and recent hospitalization in the high-risk group. Conversely, SGLT2i initiation was positively correlated with HbA1c level, BMI, presence of DR, and previous HF events (

Trends in baseline characteristics of patients initiating an SGLT2 inhibitor in the high-risk CKD group eligible for SGLT2i
Characteristics of patients receiving SGLT2i changed over time (Table S2) Another substantial change was an increase in patients' mean age of initiating SGLT2i therapy, increasing from 52.3 ± 9.7 years in 2015 to 60.1 ± 10.7 years in 2019 (p = 0.004). The duration of diabetes increased from 9.2 ± 5.6 years in 2015 to 13.0 ± 9.2 years in 2019 (p = 0.008). The proportion of patients with A3 was similarly increased from 35.6 % to 2015 to 73.3 % in 2020 (p = 0.014). When we combined G3a and G3b, the proportion of patients newly initiated on an SGLT2i in G3 category was increased from 6.7 % to 2015 to 16.2 % in 2019 (p = 0.026, data not shown).

Discussion
In this study of real-world data with T2D, SGLT2i was initiated most often for patients with the lowest risk (i.e., young, G1-G2 categories). SGLT2i was initiated for only 32.9 % of patients in the high-risk CKD group eligible for SGLT2i for renal protection (G1-G3a with A2-A3 categories). The proportion of patients who received SGLT2i in the high-risk group was significantly lower in G3a (14.8 %) than in G1 (41.4 %) and G2 (44.1 %) category. Such lower usage of SGLT2i was contrary to the 2020 ADA/EASD (European Association for the Study of Diabetes) consensus guideline to add SGLT2i for patients with CKD (eGFR 30 to ≤ 60 mL/min/1.73 m 2 or UACR > 30 mg/g, particularly UACR > 300 mg/g) [15].
In general, among T2D individuals, 50-65 %, 20-30 %, and 15-25 % have no DKD, albuminuric DKD with preserved eGFR, and reduced eGFR, respectively [16]. In a USA study, the prevalence of CKD was 43.5 % in the T2D population overall and 61.0 % in those aged ≥ 65 years [17]. In this study, total prevalence of CKD was estimated to be 33.8 % overall and 46.9 % in those aged ≥ 65 years. Our study had more patients in G3-G5 categories (13.1 %) and A3 category (11.8 %) than those in a Korean nationwide survey. In the KNHANES 2011-2013 on diabetes, the total prevalence of CKD was estimated to be 27.6 %. For each eGFR category, prevalence results were: G1, 45.2 %; G2, 45.1 %; G3, 8.8 %; and G4-5, 0.9 % [18]. Albuminuria prevalence was: A1,77.7 %; A2,18.3 %; and A3,4.0 %.   Values are presented as mean ± standard deviation or number (%) ASCVD atherosclerotic cardiovascular disease, BMI body mass index, CVD-HF atherosclerotic cardiovascular disease or heart failure, DBP diastolic blood pressure, eGFR estimated glomerular filtration rate, HbA1c glycosylated hemoglobin, HDL-C high-density lipoprotein cholesterol, HF heart failure, LDL-C lowdensity lipoprotein cholesterol, RAASi renin-angiotensin-aldosterone system inhibitor, SBP systolic blood pressure, SGLT2i sodium-glucose cotransporter-2 inhibitor In our study, 8.5 % of patients in G3-G5 categories initiated SGLT2i. On the contrary, in an Australian study regarding T2D and G3-G5 CKD, SGLT2i use was 6.2 % [9]. In a USA study, 7.2 % of patients with diabetes overall initiated an SGLT2i regardless of their CKD stage. In that report, diabetic nephropathy was present in 12.7 % of SGLT2i users and 16.1 % of nonusers. CKD was present in 6.7 % of SGLT2i users and 11.7 % of nonusers [10]. However, the definition of chronic kidney disease or diabetic nephropathy was not clear. These low percentages of SGLT2i use in T2D and CKD were similar to T2D and cardiovascular disease (CVD) data. For high-risk CVD patients eligible for recent cardiovascular outcome trials, only 5.2 % received SGLT2i in the USA and 11.1 % in the UK [19,20]. Since the permitted prescribing range for SGLT2i is for those with eGFR ≥ 45 mL/min/1.73 m 2 in Korea, patients with G3b category have no opportunity to receive SGLT2i. However, there is an additional limitation of starting SGLT2i for those in eGFR ≥ 45 mL/min/1.73 m 2 and UACR ≥ 30 mg/g. This study indicated that patients with SGLT2i initiators were younger than SGLT2i nonusers in the highrisk CKD group eligible for SGLT2i. In the elderly group, the prevalence of high-risk CKD eligible for SGLT2i was higher (28.6 %) than that in those aged < 65 years (21.8 %). However, older patients were less likely to start SGLT2i therapy (17.6 %) than those aged < 65 years (45.8 %) in the high-risk group. Trials in the SGLT2i class show that older patients have similar or greater benefits than younger patients [21][22][23]. These mutually opposed results are based on ADA recommendations that glycemic goals for some older adults might reasonably be relaxed as part of individualized care [24]. Since older patients are more likely to have various complications, Bold values denote statistical significance at the p < 0.05 level ASCVD atherosclerotic cardiovascular disease, BMI body mass index, CAD coronary artery disease, CI confidence interval, CVD-HF atherosclerotic cardiovascular disease or heart failure, DBP diastolic blood pressure, HbA1c glycosylated hemoglobin, HDL-C high-density lipoprotein cholesterol, HF heart failure, LDL-C lowdensity lipoprotein cholesterol, OR odds ratio, PAOD peripheral arterial occlusive disease, RAASi renin-angiotensin-aldosterone system inhibitors, SBP systolic blood pressure, SGLT2i sodium-glucose cotransporter-2 inhibitor they are cautious about using newly available drugs. Besides, while understanding of clinical benefits of SGLT2i is evolving, side effects such as volume depletion might be more common for older patients [24,25]. It is also difficult to distinguish whether chronic kidney disease in the elderly is due to aging or diabetes, although the DAPA-CKD study has shown that dapagliflozin can reduce the risk of worsening kidney function or death in patients with CKD without T2D [26]. Notably, for subjects older than 65 years without albuminuria, CKD has been proposed as eGFR below 45 mL/min/1.73 m² to distinguish age-related from disease-related changes in eGFR [27]. There were HbA1c differences in SGLT2i initiation. This is due to clinical inertia of physicians not to change medications when glycemic control is in the target range despite the presence of CVD-HF or CKD [6]. This is contrary to the ADA/EASD consensus recommending SGLT2i should be considered independently of baseline HbA1c or individualized HbA1c target in appropriate high-risk individuals with established T2D [15]. Furthermore, those who received SGLT2i showed significantly higher eGFR, BMI, and shorter diabetes duration. Indeed, it has been reported that obesity is a greater driver of treatment options when comparing SGLT2i, GLP-1 RA, and dipeptidyl peptidase 4 inhibitors [28]. Efforts to provide treatment following guidelines and extension of an SGLT2i 's official prescription drug label for eGFR may help improve renal outcomes of patients with T2D.
Trends in clinical characteristics for SGLT2i have also demonstrated potential impacts of clinical practice guidelines or formulary information on rates of SGLT2i initiation [29,30]. In fact, the presence of HF and CAD significantly increased the odds of SGLT2i initiation. Drug labeling for eGFR to start SGLT2i also influenced the rise in prescription. In Korea, SGLT2i was introduced in 2014 and approved for those with eGFR ≥ 60 ml/min/1.73m 2 . Empagliflozin 10 mg and dapagliflozin were approved in 2015 and 2019, respectively, for those with eGFR ≥ 45 ml/min/1.73m 2 . Hence, we can notice the rise of SGLT2i prescriptions in G3 category recently. Notably, safety and efficacy results from the DAPA-CKD trial can broaden the population of patients eligible for SGLT2i to those in the upper range of stage 4 CKD [26]. Physicians should identify patients with T2D who may benefit from renal function protection by using SGLT2i.
The strength of the present study was that it addressed underutilized SGLT2i data in the real-world for less renal benefit control among outpatients with CKD in diabetes. Our study was conducted at four diabetes centers with a larger sample size in a considerably more diverse and complex outpatient population (including G4-G5 category) than previous studies. Furthermore, it uses CKD data from Korea where the prevalence of CKD and CVD in patients with T2D is expected to be different from that in a randomized clinical trial or the West. Besides, we examined the impact of glycemic control and clinical characteristics on the likelihood of SGLT2i initiation.
Our study has some limitations. First, the study sample was not representative of the Korean population. Therefore, results should be interpreted with caution. Its main limitations derive from its retrospective cross-sectional observational nature. Since GLP-1 RA is considered a renoprotective glucose-lowering drug, we excluded GLP-1 RA users before initiation of SGLT2i in this study (3.9 %). The determination of the presence of chronic kidney disease based on eGFR was made using a single random sample of laboratory values [31]. The albuminuria category was based on two elevated UACR levels regardless of the continuity or interval. This condition might have overestimated the proportion of A2 category since increased albuminuria should be confirmed on repeat testing over 3 to 6 months. Besides, we started the study shortly after the publication of CREDENCE trial that showed renal benefits of SGLT2i in patients with eGFR > 30 mL/min/1.73 m 2 [3]. Therefore, if the study was started late, more patients with CKD would have initiated SGLT2i.

Conclusions
In the high-risk CKD group eligible for SGLT2i, younger patients with poor glycemic control in eGFR G1-G2 categories received more SGLT2i than in those without each corresponding characteristic. Thus, physicians should be aware of limitations of this glucocentric approach and focus on a personalized approach to CKD in diabetes management. Also, health plans supporting these evidence-based treatment strategies could improve renal outcomes of patients with this disease.
To overcome therapeutic inertia, the medical community needs several improvements. First, an accurate diagnosis of chronic kidney disease is necessary in order to identify eligible patients. Second, regulatory authorities should extend the eGFR values above 30 mL/min/1.73 m 2 . Finally, since more than half of CKD patients are older adults in this study, conservative clinical guidelines on the treatment of T2D for the elderly should be individualized according to CKD status.