Research Article How Many Patients with Type 2 Diabetes Meet the Inclusion Criteria of the Cardiovascular Outcome Trials with SGLT2 Inhibitors? Estimations from a Population Database in a Mediterranean Area Silvia Canivell,1,2,3 Manel Mata-Cases ,1,4,5 Bogdan Vlacho,1 Mònica Gratacòs,1 Jordi Real,1,4 Dídac Mauricio ,1,4,6 and Josep Franch-Nadal 1,4,7 1 DAP-Cat Group, Unitat de Suport a la Recerca Barcelona, Fundació Institut Universitari per a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain 2 Primary Health Care Center Sant Martí de Provençals, Gerència d’Atenció Primaria, Institut Català de la Salut, Barcelona, Spain 3 Health Sciences Research Institute and Hospital Universitari Germans Trias i Pujol, Badalona, Spain 4 CIBER of Diabetes and Associated Metabolic Diseases (CIBERDEM), Instituto de Salud Carlos III (ISCIII), Barcelona, Spain 5 Primary Health Care Center La Mina, Gerència d’Àmbit d’Atenció Primària Barcelona Ciutat, Institut Català de la Salut, Sant Adrià de Besòs, Spain 6 Department of Endocrinology & Nutrition, Hospital de la Santa Creu i Sant Pau, Autonomous University of Barcelona, Barcelona, Spain 7 Primary Health Care Center Raval Sud, Gerència d’Atenció Primaria, Institut Català de la Salut, Barcelona, Spain Correspondence should be addressed to Dídac Mauricio; didacmauricio@gmail.com and Josep Franch-Nadal; josep.franch@gmail.com Received 18 April 2019; Revised 30 September 2019; Accepted 16 October 2019; Published 11 November 2019 Academic Editor: Kim Connelly Copyright © 2019 Silvia Canivell et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Objective. Regulatory agencies require the assessment of cardiovascular (CV) safety for new type 2 diabetes (T2D) therapies through CV outcome trials (CVOTs). However, patients included in CVOTs assessing sodium-glucose cotransporter-2 inhibitors (SGLT2i) might not be representative of those seen in clinical practice. This study examined the proportion of patients that would have been enrolled into three main SGLT2i CVOTs to determine whether these trials’ eligibility criteria can be applied to a real-world Mediterranean T2D population. Methods. Cross-sectional, retrospective, cohort study of T2D patients registered in primary care centres of the Catalan Institute of Health using medical records from a population database (SIDIAP) that includes approximately 74% of the population in Catalonia (Spain). Eligibility criteria were according to those of three SGLT2i CVOTs: EMPA-REG OUTCOME (empagliflozin), CANVAS (canagliflozin), and DECLARE-TIMI 58 (dapagliflozin). Results. By the end of 2016, the database included 373,185 patients with T2D with a mean age of 70 ± 12 years, 54.9% male, with a mean duration of T2D of 9 ± 6 years, and a mean glycated haemoglobin (HbA1c) of 7:12% ± 1:32 (59% with HbA1c < 7%). Of these, 86,534 (23%) had established CV disease and 28% chronic renal failure (estimated glomerular filtration < 60 ml/min/1:73m2 ). Among all included patients, only 8.2% would have qualified for enrolment into the EMPA-REG OUTCOME trial, 29.6% into the CANVAS program, and 38% into the DECLARE-TIMI 58 trial. The main limiting factors for inclusion would have been a previous history of CV disease and the baseline HbA1c value. Conclusion. The external validity of the analysed CVOTs is clearly limited when applying the same eligibility criteria to a T2D Mediterranean population. 2 Journal of Diabetes Research 1. Introduction between trials regarding the proportion of patients seen in clinical practice that would have met entry criteria in Patients with type 2 diabetes (T2D) have an increased risk these CVOTs, with the DECLARE-TIMI 58 trial as the of renal and cardiovascular disease (CVD) and mortality most generalisable and applicable one. Moreover, the [1]. Therefore, improvement in cardiovascular (CV) health results from the study conducted in Northern Europe were is one of the main goals of diabetes management. While consistent across all four included countries (i.e., Germany, tight good glycaemic control in T2D is associated with The Netherlands, Norway, and Sweden) [18]. However, reduced risk of microvascular disease [2, 3], the benefit there is no published information from Southern Euro- regarding macrovascular disease is less clear [4–6]. Indeed, pean countries so far, although the distribution of CV risk a meta-analysis combining the results of large-scale trials factors as well as the prevalence of CVD in patients with showed that intensive glucose-lowering therapy was associ- diabetes differs across regions in Europe [20, 21]. Based ated with a significant reduction in the overall incidence of on these potential differences, we hypothesised that the CV events and myocardial infarction compared to conven- external validity of the CVOTs could be different when tional therapy (odds ratio (OR) 0.89, P = 0:001; OR 0.84, the general T2D population is estimated in a Mediterra- P < 0:001, respectively) [7]. However, there was no differ- nean country. ence in the incidence of CV mortality [7]. Both the US The aim of the present study was to determine the pro- Food and Drug Administration (FDA) and the European portion of patients with T2D in primary care that would be Medicines Agency (EMA) require, for each new antidia- eligible for inclusion in the CVOTs of SLGT2i in the popula- betic therapy to treat T2D, to show a neutral or beneficial tion served by the Catalonian Health Institute in Catalonia, a effect in CV safety through the conduction of CV outcome Mediterranean area in the northeast of Spain. trials (CVOTs) [8, 9]. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) are 2. Materials and Methods a promising group of new drugs for the treatment of T2D that act by preventing the reabsorption of glucose from the 2.1. Design. This was a cross-sectional retrospective study of proximal renal tubule in the kidney [10]. Additionally, they the T2D population attended at primary care centres of Cat- have numerous pleiotropic effects such as reducing blood alonia, an autonomous region located in the northeast of plasma glucose, body weight, and blood pressure and induc- Spain, corresponding to 12% of the total Spanish population. ing natriuresis [10]. In the particular case of SGLT2i, recent We compared the potential eligibility of patients to those CVOTs have shown renal and CV benefits and further stud- included in publications describing three completed SLGT2i ies are ongoing [11–13]. However, one of the major issues of CVOTs [15–17]. randomised clinical trials (RCTs) is the external validity of The study was approved by the Ethics Committee of the the results, that is, to what extent the overall average effect Primary Health Care University Research Institute (IDIAP) of the treatment can be generalised to a particular group of Jordi Gol in accordance with the Spanish regulations on patients or clinical setting [14]. For instance, the external observational studies. This retrospective study using anon- validity can be challenged by the trial’s setting (e.g., differ- ymised data did not require obtaining informed consent ences between countries regarding the health care system, from the patients. disease management, or natural history of the disease), the inclusion and exclusion criteria, or differences between the 2.2. Data Source. Data from patients were extracted from protocol trial and routine clinical practice, among other the SIDIAP database, which contains anonymised longitu- issues [14]. dinal patient information obtained from electronic clinical The results of the CVOTs of three SLGT2 inhibitors records; it incorporates available information from 288 available in Spain published to date are EMPA-REG OUT- primary care teams of the Catalonian Health Institute COME with empagliflozin [15], CANVAS with canagliflo- (ICS), which serves around 5.6 million people, 74% of zin [16], and DECLARE-TIMI 58 with dapagliflozin [17]. the total population in Catalonia. The SIDIAP includes The EMPA-REG OUTCOME trial included only patients demographic, clinical, and pharmacy-invoicing data pro- with established CV disease (CVD), i.e., secondary preven- vided by the CatSalut general database, and it has already tion [15]. The other two trials included secondary preven- been used for epidemiological research purposes and real- tion patients and also patients with CV risk factors who world evidence [22–24]. have not yet developed CVD (primary prevention): with ≥1 CV risk factors in the DECLARE-TIMI 58 trial [17] 2.3. Inclusion and Exclusion Criteria. The study population and with ≥2 CV risk factors in the CANVAS study [16]. consisted of patients aged 18 years or older with a diagno- Since the eligibility criteria varied among these SGLT2i sis of T2D (International Classification of Diseases (ICD- CVOTs, it was expected that the external validity of the 10) codes E11, E11.0-E11.9, E14, and E14.0-E14.9) as of different studies might also differ; thus, the trial popula- 31 December 2016 (index date). We excluded patients tion does not actually represent the general T2D popula- with a diagnosis of type 1 diabetes, gestational diabetes tion. Indeed, the external validity of CVOTs regarding mellitus, and any other type of diabetes. After this initial SGLT2i has been assessed by two recently published stud- selection, we applied the inclusion and exclusion criteria ies using clinical routine data from the US and Northern from the CVOTs for the SGLT2i commercialised up to date Europe [18, 19]. Both studies found large differences in Spain, namely, empagliflozin (EMPA-REG OUTCOME Journal of Diabetes Research 3 Table 1: Clinical characteristics of patients with T2D registered in the SIDIAP database. All population Secondary prevention patients∗ Primary prevention patients∗∗ Characteristics (N = 373,185) (N = 86,534) (N = 286,651) Gender, male, n (%) 204,707 (54.9) 56,882 (65.7) 147,825 (51.6) Age (years), mean (SD) 70.1 (12.3) 74.8 (10.5) 68.7 (12.5) Current smokers, n (%) 52,744 (14.1) 11,058 (12.8) 41,686 (14.5) BMI (kg/m2), n (%) ≥30 144,592 (44.9) 30,366 (40.3) 114,226 (46.4) >45 3,905 (1.2) 521 (0.7) 3,384 (1.4) Duration of diabetes (years), 9.3 (6.2) 10.9 (6.7) 8.8 (6.0) mean (SD) HbA1c (%)†, mean (SD) 7.12 (1.32) 7.16 (1.32) 7.10 (1.33) HbA1c ≤ 7%, n (%) 194,751 (59.0) 43,664 (56.6) 151,087 (59.8) Hypertension, n (%) 268,394 (71.9) 70,026 (80.9) 198,368 (69.2) Dyslipidemia, n (%) 223,785 (60.0) 56,194 (64.9) 167,591 (58.5) eGFR (ml/min/1.73m2)†, n (%) ≥60 241,958 (72) 45,598 (58) 196,360 (76) 30-60 80,978 (24) 27,651 (35) 53,327 (21) <30 13,262 (4) 5,909 (7) 7,353 (3) UACR ≥ 30 mg/g, n (%) 51,429 (13.8) 17,126 (19.8) 34,303 (12.0) T2D treatment, n (%) No antidiabetic medication 68,681 (18.4) 12,178 (14.1) 56,503 (19.7) NIAD monotherapy 138,615 (37.1) 28,582 (33.0) 110,033 (38.4) NIADs in combination 86,508 (23.2) 18,557 (21.4) 67,951 (23.7) Insulin±NIAD 79,381 (21.3) 27,217 (31.5) 52,164 (18.2) BMI: body mass index; eGFR: estimated glomerular filtration rate; NIAD: noninsulin antidiabetic drug; SD: standard deviation; T2D: type 2 diabetes; UACR: urine albumin-to-creatinine ratio; HbA1c: glycated haemoglobin. ∗ Secondary prevention: patients with established cardiovascular disease (ICD-10 codes for coronary heart disease, cerebrovascular disease, or peripheral arteriopathy). ∗∗ Primary prevention: patients without any ICD-10 code for cardiovascular disease. †There were 12% of missing data in the registration of HbA1c and 10% in the registration of eGFR. trial), canagliflozin (CANVAS trial), and dapagliflozin tered population. Data were summarised as mean (standard (DECLARE-TIMI 58 trial) (Supplementary Table 1) [15–17]. deviation (SD)) or n (%). 2.4. Study Variables. The following variables from the 3. Results SIDIAP database at the end of December 2016 were analysed: age and gender; duration of T2D; the most recent value (clos- A total of 373,185 patients with T2D were registered in the est to 31 December 2016) of glycated haemoglobin (HbA1c); SIDIAP database as of 31 December 2016 (Table 1). The mean and presence of risk factors, including hypertension (ICD-10 age was 70:1 ± 12:3 years, and 54.9% were male. The mean codes I10 and I15, or systolic blood pressure ðSBPÞ ≥ 140 T2D duration was 9:3 ± 6:2 years, and the mean HbA1c was and/or diastolic blood pressure ðDBPÞ ≥ 90 mmHg, or use of 7:12% ± 1:32. More than half (59%) of the patients had good antihypertensive medications), dyslipidemia (ICD-10 codes glycaemic control (HbA1c ≤ 7%), about one-third (28%) had E780 to E785, or LDL cholesterol ðLDLcÞ ≥ 160 mg/dl, or chronic kidney disease (eGFR < 60 ml/min/1:73m2 ), and 4% use of lipid-lowering drugs), smoking status, body mass had severe renal insufficiency (eGFR < 30 ml/min/1:73m2 ). index (BMI), estimated glomerular filtration rate (eGFR) Overall, 77% of patients (n = 286,651) did not have any using the Chronic Kidney Disease Epidemiology Collabora- ICD-10 code of established CVD recorded, thus consid- tion (CKD-EPI) equation, albumin-to-creatinine ratio ered primary prevention cases; 23% (n = 86,534) had an (UACR), and history of CVD (stroke, peripheral artery dis- ICD-10 code of an established CVD, so they were con- ease, ischaemic heart disease, and heart failure) (Supplemen- sidered secondary prevention cases. Patients with estab- tary Table 2). Noninsulin antidiabetic drugs (NIADs) and lished CVD were more often men, older, and with a insulin-active electronic prescriptions on the index date longer T2D duration and had more CV risk factors were also considered. (Table 1). Applying the EMPA-REG OUTCOME eligibility cri- 2.5. Statistical Analysis. The eligibility was determined by teria, only 8.2% (n = 30,559) of the patients included in dividing the number of patients fulfilling each of the CVOTs the SIDIAP database would have qualified for entry in key inclusion and exclusion criteria by the total T2D regis- the trial (Table 2), while 29.6% (n = 110,551) could have 4 Journal of Diabetes Research Table 2: Eligibility criteria for the EMPA-REG OUTCOME (empagliflozin) trial and number of patients in the SIDIAP database that would have met criteria for enrolment. Eligibility criteria in EMPA-REG OUTCOME trial Potentially eligible patients from the SIDIAP database (N = 373,185) Inclusion criteria n (%) Age ≥ 18 years 373,185 (100) Preexisting CV event: CHD, angina, MI, stroke, and PAD 86,534 (23.2) HbA1c level 7.0%-≤10.0% 33,270 (8.9) Main exclusion criteria n (%) eGFR < 30 ml/min/1:73m2 2,488 (0.7) 2 BMI > 45 kg/m 223 (0.06) Total eligible, n (%) 30,559 (8.2) CHD: coronary heart disease; CV: cardiovascular; eGFR: estimated glomerular filtration rate; MI: myocardial infarction; PAD: peripheral artery disease; HbA1c: glycated haemoglobin; BMI: body mass index. All the percentages refer to the proportion from the total number of eligible subjects of the SIDIAP database (n = 373,185). Table 3: Eligibility criteria for the CANVAS program (canagliflozin) and number of patients in the SIDIAP database that would have met criteria for enrolment. Potentially eligible patients from the SIDIAP Eligibility criteria in the CANVAS program database (n = 373,185) Inclusion criteria In PP, n (%) In SP, n (%) Age: ≥50 years in PP 349,896 (93.8) — ≥30 years in SP — 372,764 (99.9) Primary prevention cohort ≥50 years and ≥2 CVRF: (i) T2D duration ≥10 years (ii) Hypertension 189,969 (50.9) — (iii) Current smoker (iv) Micro- or macroalbuminuria (v) HDLc − <39 mg/dl (1 mmol/l) Secondary prevention cohort ≥30 years and history of CV events (CHD, angina, MI, stroke, and PAD) — 86,531 (23.2) HbA1c: 7.0%-10.5% 83,537 (22.4) 34,320 (9.2) Main exclusion criteria In PP, n (%) In SP, n (%) eGFR < 30 ml/min/1:73m2 4731 (1.27) 2575 (0.69) Total eligible in PP and SP, n (%) 78,806 (21.1) 31,745 (8.5) Total eligible, n (%) 110,551 (29.6) CHD: coronary heart disease; CV: cardiovascular; CVRF: cardiovascular risk factors; eGFR: estimated glomerular filtration rate; MI: myocardial infarction; PP: primary cardiovascular prevention; SP: secondary cardiovascular prevention; PAD: peripheral artery disease; T2D: type 2 diabetes; HbA1c: glycated haemoglobin; HDLc: high-density lipoprotein cholesterol. All the percentages refer to the proportion from the total number of eligible subjects of the SIDIAP database (n = 373,185). been enrolled in the CANVAS program (Table 3) and 38% CANVAS, only patients with severe renal failure (n = 141,653) would have been eligible for the DECLARE- (eGFR < 30 ml/min/1:73m2 ) were excluded (4% in SIDIAP). TIMI 58 trial (Table 4). In Figure 1, these results are shown in comparison to those reported by two studies conducted in 4. Discussion the US and Northern Europe [18, 19]. Compared with the characteristics of the patients In this population-based study using routine clinical data, we enrolled in the three studied CVOTs (Table 5), the main lim- estimated the proportion of patients with T2D that would iting factors for inclusion would have been the absence of a have been eligible for inclusion in the main CVOTs regarding history of CVD and the value of HbA1c. Specifically, only the use of SGLT2i in our environment (Catalonia, Spain). We 23% of patients in our T2D population had a preexisting found that the DECLARE-TIMI 58 CVOT trial had the high- CVD and only 41% had HbA1c values ≥ 7%. Finally, chronic est representativeness, covering 38% of the T2D patients in renal failure (present in 28% of our T2D patients) was a our general T2D population, which is in line with previously restrictive criterion in DECLARE, while in EMPAREG and reported studies conducted in Northern Europe and the US Journal of Diabetes Research 5 Table 4: Eligibility criteria for the DECLARE-TIMI 58 trial (dapagliflozin) and number of patients in the SIDIAP database that would have met criteria for enrolment. Potentially eligible patients from the SIDIAP Eligibility in the DELCLARE-TIMI 58 trial database (n = 373,185) Inclusion criteria In PP, n (%) In SP, n (%) Age: (i) In PP: ≥55 years; ≥60 in women 175,092 (46.9); 140,941 (37.8) — (ii) In SP: ≥40 — 368,963 (98.9) Primary prevention cohort ≥55 years (≥60 in women) and ≥1 CVRF: (i) Dyslipidemia 289,126 (77.5) — (ii) Hypertension (iii) Current smoker Secondary prevention cohort ≥40 years and history of CV events (CHD, angina, MI, stroke, and PAD) — 86,468 (23.2) HbA1c 6.5%-<12% 165,777 (44.4) 50,872 (13.6) Main exclusion criteria In PP In SP eGFR < 60 ml/min/1:73m2 53,851 (14.4) 21,145 (5.7) Total eligible in PP and SP, n (%) 111,926 (30) 29,727 (7.9) Total eligible, n (%) 141,653 (38%) CHD: coronary heart disease; CV: cardiovascular; CVRF: cardiovascular risk factors; eGFR: estimated glomerular filtration rate; MI: myocardial infarction; PAD: peripheral artery disease; PP: primary cardiovascular prevention; SP: secondary cardiovascular prevention; HbA1c: glycated haemoglobin. All the percentages refer to the proportion from the total number of eligible subjects of the SIDIAP database (n = 373,185). EMPAREG, 8.5% for the CANVAS program, and 7.9% for Percentage of patients meeting eligibility criteria the DECLARE trial. Indeed, the main differences in the 70 design of the different CVOT trials that we included were 60 59 the enrolment of patients at high CV risk at baseline: EMPA-REG OUTCOME included only patients with T2D 50 and established CV disease (i.e., secondary prevention), while 39.8 38 both the CANVAS and the DECLARE-TIMI 58 also 40 34 included primary prevention patients. In our real-world 30 29.6 T2D population, the main limiting factors for inclusion in 21 the different CVOTs analysed were the absence of a history 20 of CV disease (present in the 23% of the patients) and glycae- 8.8 8.2 10 4.1 mic control: only 41% of the patients had a HbA1c ≥ 7% 0 (entry criteria for the EMPA-REG OUTCOME and CAN- DECLARE-TIMI 58 CANVAS EMPA-REG OUTCOME VAS trials), and 65.3% were above 6.5% threshold (entry cri- Northern Europe terion in the DECLARE-TIMI 58 study). US Concern regarding poor external validity (i.e., generalisa- Mediterranean area (Catalonia) bility) of RCTs is largely known and has implications for the use (or underuse) of treatments in routine clinical practice Figure 1: Graphical representation of the representativeness of [14, 26]. This topic has also been addressed in the past patients in SGLT2i CVOTs when compared to the general type 2 regarding RCTs in the field of diabetes. For instance, in a pre- diabetes population from four European countries, the US, and vious literature review, the external validity of large trials the present Mediterranean population. assessing the impact of glycaemic control on CVD in patients with T2D was reported as limited when applied to a T2D [13, 17, 25]. This result is not surprising since the DECLARE- population-based cohort [27]. Another recent study evalu- TIMI 58 CVOT trial included similar proportions of patients ated the population representativeness of 1691 registered in primary CV prevention and secondary CV prevention T2D trials [28] and found that in 51.4% of cases (and (59% and 41%, respectively) [17], which ensures the highest 53.1% of phase 2 and 3 interventional trials), the population representativeness. The CVOT with the lowest generalisabil- representativeness was <5%. Of note, and in line with our ity was the EMPA-REG OUTCOME trial because it included results, the eligibility criterion that had the largest effect on only secondary prevention patients [15]. However, if we only the population representativeness was HbA1c [28]. Finally, take into account patients with cardiovascular disease as eli- the study showed that the greater the number of eligibility gible to enter into the trials, the resulting proportion would criteria, the lower the representativeness was, and the authors have been similar among the three trials: 8.2% for the concluded that the low representativeness of T2D trials could 6 Journal of Diabetes Research Table 5: Summary of characteristics of patients enrolled in SGLT2i CVOTs and of patients from the SIDIAP database that would have met the corresponding criteria. EMPA-REG Characteristics CANVAS DECLARE-TIMI 58 General population (SIDIAP database) OUTCOME Drug Empagliflozin Canagliflozin Dapagliflozin Empagliflozin/canagliflozin/dapagliflozin Participants, n 7,020 10,142 17,160 373,185 Male (%) 71.5 64.2 62.6 54.9 Age (years), mean (SD) 63.1 (8.7) 63.3 (8.3) 63.9 (6.8) 70.1 (12.3) EMPA-REG OUTCOME criteria: Patients with established 86,534 (23.2) 7,020 (>99) 6,656 (66) 6,974 (41) CVD, n (%) CANVAS criteria: 86,531 (23.2) DECLARE-TIMI 58 criteria: 86,468 (23.2) CANVAS criteria: 189,969 (50.9) CVRFs, n (%) — 3486 (34) 10,186 (59) DECLARE-TIMI 58 criteria: 289,126 (77.5) HbA1c (%), mean (SD) 8.1 (0.8) 8.2 (0.9) 8.3 (1.2) 7.12 (1.32) eGFR (ml/min/1.73 m2), n (%) ≥60 5,199 (74.1) 8,114 (79.9) 15,959 (92.6) 241,958 (72) <60-30 1,819 (25.9) 2,028 (20.1) 1,201 (7.4) 80,978 (24) <30 0 0 0 13,262 (4) CVD: cardiovascular disease; CVRF: cardiovascular risk factors; eGFR: estimated glomerular filtration rate; SD: standard deviation; CVOTs: cardiovascular outcome trials. be attributed to safety concerns when designing the study, for the primary composite outcome of end-stage kidney which may lead to overly restrictive eligibility criteria to pre- disease, doubling of serum creatinine, or renal or CV vent adverse events [28]. death = 0:70; 95% CI, 0.59–0.82) [38]. All the patients Chronic renal failure was a restrictive criterion in included in this trial had an eGFR between 30 and DECLARE, while EMPA-REG OUTCOME and CANVAS 90 ml/min/1.73m2 and albuminuria, and in a subgroup trials only excluded patients with severe renal insufficiency analysis, the renal benefit of canagliflozin was higher among (<30 ml/min/1.73 m2), present in 4% of our population. patients with eGFR between 45 and 60 ml/min/1.73m2 However, renal benefits of SGLT2i have been shown in differ- (HR 0.52; 95% CI, 0.38–0.72) [38]. ent studies and one meta-analysis [29], particularly in the Indeed, a recent systematic review and trial-level meta- EMPA-REG OUTCOME and CANVAS trials [15, 16, 30– analysis of SGLT2i CVOT trials concludes that SGLT2i 32], although these results were from secondary outcomes. reduce the risk of worsening eGFR in a broad spectrum of New trials are ongoing with the primary renal endpoint (kid- T2D patients [39]. Considering the facts mentioned above, ney outcome trials) focusing on patients with T2D and estab- if SGLT2i were allowed to be prescribed to patients with lished chronic kidney disease (CKD) [33, 34]. The latest eGFR>30 ml/min/1.73 m2, the proportion of patients eligible recommendations from the 2018 ADA/EASD Consensus for inclusion in the DECLARE TIMI 58 in the SIDIAP data- Guidelines suggest to consider the use of a SGLT2i in patients base would have increased. with T2D and CKD [35]. In fact, the FDA has recently The results of the present study show that the patterns of modified this limitation for patients with moderate renal external validity of SGLT2i CVOTs were limited, in line with impairment (i.e., eGFR45 60 ml/min/1.73 m2) for dapagli- the studies conducted in Northern Europe or the US [18, 19]. flozin, as previously done with canagliflozin 100 mg and However, the proportion of eligible patients would have been empagliflozin 10 mg [36]. In summary, in the US, SGLT2i less in our population than in Northern Europe for all three are still not recommended when eGFR is less than trials but higher than that in the US for the CANVAS and 45 ml/min/1.73m2 and remain contraindicated in patients EMPAREG studies as shown in Figure 1. This could be with severe renal impairment (eGFR < 30 ml/min/1:73 m2 ), explained by differences in the prevalence of CVD between end-stage renal disease, or on dialysis [36]. Conversely, the regions and countries. Indeed, the prevalence of CVD is EMA has not yet modified the general restriction of eGFR higher in countries from the north of Europe than from the < 60 ml/min/1:73 m2 . However, it is likely that recommen- south of Europe [21], but it is also probable that differences dations by drug agencies for SGLT2i in CKD will change in in the prevalence of specific CV risk factors by country and the future after the publication of the results from the ongo- region have impacted our results. The low eligibility for ing kidney outcome trials (CREDENCE, DAPA-CKD, and CANVAS and EMPA-REG OUTCOME in the US study is EMPAGLIFLOZIN RENAL) [37]. In addition, recent results striking, although the percentage of patients having CVD of the CREDENCE trial showed that the risk of kidney failure was similar to ours (23.7 and 23%, respectively). One expla- and renal or CV mortality was 30% lower in the group of nation for this discordance could be related to the different patients receiving canagliflozin compared to placebo (HR methodologies used in the US study, which is an estimation Journal of Diabetes Research 7 based on data from two NHANES surveys. In this study, data iterranean region where the prevalence of CV risk factors and from 20,293 volunteer subjects, 2,395 of whom had T2D, CVD in patients with T2D is expected to be different from were extrapolated to 23,941,512 persons having T2D in the that in Northern Europe or the US [21]. US [19]. For instance, these patients were younger than in the other databases: 59 years old in the US, 68 years in 5. Conclusions the European study, and 70 years in our study. As a result, the prevalence of CV risk factors was lower in our database This study shows that there are considerable differences in (13.6% in the US vs. 28%). Conversely, the prevalence of the external validity of the different CVOTs of SGLT2i when CVD in the European study (43.7% in Germany, 34.8% in applying the same eligibility criteria to the T2D population of The Netherlands, 31.4% in Sweden, and 25.1% in Norway) Catalonia. The DECLARE-TIMI 58 CVOT was the most was higher in our database (23%) and produced higher per- generalisable, while the EMPA-REG OUTCOME and CAN- centages of inclusion in the CANVAS and EMPA-REG VAS trials were much less representative of real-world T2D OUTCOME trials [18]. patients. However, the clinical applicability of SGLT2i in rou- On the other hand, the applicability of SGLT2i is cer- tine practice goes beyond the strict inclusion criteria of tainly much wider than the strict entry criteria of CVOTs and it is important to consider all patient-centered CVOTs, as shown in several of studies and subanalyses aspects before decision-making in T2D management. reporting the additional benefits of SGLT2i, particularly in chronic kidney disease and heart failure [40]. For Data Availability instance, real-world evidence (RWE) studies have also confirmed reductions in mortality (43% to 49%) and hos- The data used to support the findings of this study are avail- pitalization for heart failure (40% to 51%) in hundreds of able from the corresponding authors upon request. thousands of patients [40]. So far, the main RWE studies published are CVD-Real, CVD-Real 2, EASEL, and Disclosure EMPRISE; these studies compared SGLT2i with other The funding source had no role in the design and conduct antidiabetic drugs, especially against DPP-4 inhibitors of the study; collection, management, analysis, and inter- (CVD-Real 2 and EMPRISE). However, our study aimed pretation of the data; or preparation, review, or approval at determining how many patients from our Mediterra- of the manuscript. nean database would be eligible to enter in each CVOT, but not the applicability of SGLT2i in the whole diabetic population. Thus, even though we found a low percent- Conflicts of Interest age of patients that would have been enrolled in each M.M.-C. has received advisory honorarium from Astra- of the cardiovascular outcome trials, we should not dis- Zeneca, Bayer, Boehringer Ingelheim, GSK, Lilly, MSD, miss the additional benefits of using SGLT2i in the whole Novartis, Novo Nordisk, and Sanofi; he has received diabetic population in terms of heart failure benefit. speaker honorarium from Astra-Zeneca, Bayer, Boehringer Hence, our results should be taken with caution and Ingelheim, GSK, Lilly, Menarini, MSD, Novartis, Novo should not drive the decision to prescribe or not an Nordisk, and Sanofi; he has received research grants to SGLT2i in routine practice without considering all clinical the institution from Astra-Zeneca, GSK, Lilly, MSD, aspects and patient preferences. Finally, we should point Novartis, Novo Nordisk, and Sanofi. J.F.-N. has received out that, at the time of deciding whether to prescribe advisory and/or speaking fees from Astra-Zeneca, Ascen- or not an SGLT2i, clinicians should also refer to the lat- sia, Boehringer Ingelheim, GSK, Lilly, MSD, Novartis, est available evidence and updated guidelines for T2DM Novo Nordisk, and Sanofi; he has received research grants management [35, 41]. In this line, the 2018 Consensus to the institution from Astra-Zeneca, GSK, Lilly, MSD, Report of the American Diabetes Association (ADA) Novartis, Novo Nordisk, Sanofi, and Boehringer. D.M. has and the European Association for the Study of Diabetes received advisory and/or speaking fees from Astra-Zeneca, (EASD) state that, in the presence of cardiovascular dis- Ascensia, Boehringer Ingelheim, GSK, Lilly, MSD, Novartis, ease, empagliflozin or canagliflozin should be recom- Novo Nordisk, and Sanofi; he has received research grants mended if HbA1c levels are above target, usually above to the institution from Astra-Zeneca, GSK, Lilly, MSD, 7% [35]. Moreover, the recently published 2019 ESC Novartis, Novo Nordisk, Sanofi, and Boehringer. S.C., J.R., Guidelines in collaboration with the EASD recommend B.V., and M.G. have no conflicts of interest to declare. the use of SGLT2i in case of established cardiovascular disease or in subjects at high or very high CV risk, with- Authors’ Contributions out establishing any specific threshold for HbA1c [41]. This study has some limitations. The main limitations Silvia Canivell and Manel Mata-Cases contributed equally to derive from its retrospective observational nature, which this work. are common to all similarly designed studies using real- world databases. For instance, 12% and 10% of patients had Acknowledgments no available HbA1c or eGFR values during the evaluated year, respectively. On the other hand, the strength of the The authors acknowledge Amanda Prowse (Lochside Medi- present study is that it involves real-world data from a Med- cal Communications Ltd.) for providing support in editing 8 Journal of Diabetes Research the paper. This study was funded by the Fundació Institut cal considerations in type 2 diabetes management,” Diabetes Universitari per a la recerca a l’Atenció Primària de Salut Care, vol. 41, no. 8, pp. 1543–1556, 2018. Jordi Gol i Gurina (IDIAPJGol). CIBER of Diabetes and [11] T. A. Zelniker, S. D. Wiviott, I. Raz et al., “SGLT2 inhibitors for Associated Metabolic Diseases (CIBERDEM) is an initiative primary and secondary prevention of cardiovascular and renal from Instituto de Salud Carlos III, Madrid, Spain. outcomes in type 2 diabetes: a systematic review and meta- analysis of cardiovascular outcome trials,” The Lancet, vol. 393, no. 10166, pp. 31–39, 2019. Supplementary Materials [12] J. H. Y. Wu, C. Foote, J. Blomster et al., “Effects of sodium- glucose cotransporter-2 inhibitors on cardiovascular events, Supplementary Table 1: eligibility criteria of the three studied death, and major safety outcomes in adults with type 2 diabe- CVOTs completed for SGLT2i in T2D patients. Supplemen- tes: a systematic review and meta-analysis,” The Lancet Diabe- tary Table 2: ICD-10 codes for cardiovascular diseases con- tes & Endocrinology, vol. 4, no. 5, pp. 411–419, 2016. sidered in the study. (Supplementary Materials) [13] K. I. Birkeland, M. E. Jørgensen, B. Carstensen et al., “Cardio- vascular mortality and morbidity in patients with type 2 diabe- References tes following initiation of sodium-glucose co-transporter-2 inhibitors versus other glucose-lowering drugs (CVD-REAL [1] Emerging Risk Factors Collaboration, N. Sarwar, P. Gao et al., Nordic): a multinational observational analysis,” The Lancet “Diabetes mellitus, fasting blood glucose concentration, and Diabetes & Endocrinology, vol. 5, no. 9, pp. 709–717, 2017. risk of vascular disease: a collaborative meta-analysis of 102 [14] P. M. Rothwell, “External validity of randomised controlled prospective studies,” The Lancet, vol. 375, no. 9733, trials: "To whom do the results of this trial apply?",” Lancet pp. 2215–2222, 2010. (London, England), vol. 365, no. 9453, pp. 82–93, 2005. [2] I. M. Stratton, A. I. Adler, H. A. Neil et al., “Association of gly- [15] B. Zinman, C. Wanner, J. M. Lachin et al., “Empagliflozin, car- caemia with macrovascular and microvascular complications diovascular outcomes, and mortality in type 2 diabetes,” The of type 2 diabetes (UKPDS 35): prospective observational New England Journal of Medicine, vol. 373, no. 22, pp. 2117– study,” BMJ, vol. 321, no. 7258, pp. 405–412, 2000. 2128, 2015. [3] ADVANCE Collaborative Group, A. Patel, S. MacMahon [16] B. Neal, V. Perkovic, K. W. Mahaffey et al., “Canagliflozin and et al., “Intensive blood glucose control and vascular outcomes cardiovascular and renal events in type 2 diabetes,” The New in patients with type 2 diabetes,” New England Journal of Med- England Journal of Medicine, vol. 377, no. 7, pp. 644–657, icine, vol. 358, no. 24, pp. 2560–2572, 2008. 2017. [4] N. Marx, J. Rosenstock, S. E. Kahn et al., “Design and baseline [17] S. D. Wiviott, I. Raz, M. P. Bonaca et al., “Dapagliflozin and characteristics of the CARdiovascular Outcome Trial of LINA- cardiovascular outcomes in type 2 diabetes,” The New England gliptin versus glimepiride in type 2 diabetes (CAROLINA®),” Journal of Medicine, vol. 380, no. 4, pp. 347–357, 2019. Diabetes and Vascular Disease Research, vol. 12, no. 3, pp. 164–174, 2015. [18] K. I. Birkeland, J. Bodegard, A. Norhammar et al., “How repre- sentative of a general type 2 diabetes population are patients [5] M. H. Secrest, J. A. Udell, and K. B. Filion, “The cardiovascular safety trials of DPP-4 inhibitors, GLP-1 agonists, and SGLT2 included in cardiovascular outcome trials with SGLT2 inhibi- inhibitors,” Trends in Cardiovascular Medicine, vol. 27, no. 3, tors? A large European observational study,” Diabetes, Obesity pp. 194–202, 2017. and Metabolism, vol. 21, no. 4, 2019. [6] M. A. Nauck, J. J. Meier, M. A. Cavender, M. Abd el Aziz, and [19] E. T. Wittbrodt, J. M. Eudicone, K. F. Bell, D. M. Enhoffer, D. J. Drucker, “Cardiovascular actions and clinical outcomes K. Latham, and J. B. Green, “Eligibility varies among the 4 with glucagon-like peptide-1 receptor agonists and dipeptidyl sodium-glucose cotransporter-2 inhibitor cardiovascular out- peptidase-4 inhibitors,” Circulation, vol. 136, no. 9, pp. 849– comes trials: implications for the general type 2 diabetes US 870, 2017. population,” The American Journal of Managed Care, vol. 24, 8 Suppl, pp. S138–S145, 2018. [7] E. Mannucci, M. Monami, C. Lamanna, F. Gori, and N. Marchionni, “Prevention of cardiovascular disease through [20] T. Tamayo, J. Rosenbauer, S. H. Wild et al., “Diabetes in glycemic control in type 2 diabetes: a meta-analysis of ran- Europe: an update,” Diabetes Research and Clinical Practice, domized clinical trials,” Nutrition, metabolism, and cardiovas- vol. 103, no. 2, pp. 206–217, 2014. cular diseases : NMCD, vol. 19, no. 9, pp. 604–612, 2009. [21] T. R. Einarson, A. Acs, C. Ludwig, and U. H. Panton, “Preva- [8] US Food and Drug Administration (FDA), “Guidance for lence of cardiovascular disease in type 2 diabetes: a systematic Industry. Diabetes mellitus — evaluating cardiovascular risk literature review of scientific evidence from across the world in in new antidiabetic therapies to treat type 2 diabetes. FDA 2007-2017,” Cardiovascular Diabetology, vol. 17, no. 1, p. 83, website,” March, 2019, http://www.fda.gov/downloads/ 2018. Drugs/.../Guidances/ucm071627.pdf. [22] M. Mata-Cases, D. Mauricio, J. Real, B. Bolíbar, and J. Franch- [9] European Medicines Agency, (EMA), “Guideline on clinical Nadal, “¿Se registra y clasifica correctamente la diabetes melli- investigation of medicinal products in the treatment or pre- tus en atencion primaria? Estudio poblacional en Cataluña, vention of diabetes mellitus,” March 2019, https://www.ema España,” Endocrinología y Nutrición, vol. 63, no. 9, pp. 440– .europa.eu/en/documents/scientific-guideline/draft-guideline- 448, 2016. clinical-investigation-medicinal-products-treatment- [23] M. Mata-Cases, J. Franch-Nadal, J. Real et al., “Therapeutic prevention-diabetes-mellitus_en.pdf. inertia in patients treated with two or more antidiabetics in [10] M. J. B. van Baar, C. C. van Ruiten, M. H. A. Muskiet, L. van primary care: factors predicting intensification of treatment,” Bloemendaal, R. G. IJzerman, and D. H. van Raalte, “SGLT2 Diabetes, Obesity and Metabolism, vol. 20, no. 1, pp. 103– inhibitors in combination therapy: from mechanisms to clini- 112, 2018. Journal of Diabetes Research 9 [24] I. Vinagre, M. Mata-Cases, E. Hermosilla et al., “Control of [39] T. A. Zelniker, S. D. Wiviott, I. Raz et al., “Comparison of the glycemia and cardiovascular risk factors in patients with type effects of glucagon-like peptide receptor agonists and sodium- 2 diabetes in primary care in Catalonia (Spain),” Diabetes glucose Cotransporter 2 inhibitors for prevention of major Care, vol. 35, no. 4, pp. 774–779, 2012. adverse cardiovascular and renal outcomes in type 2 diabetes [25] S. V. Arnold, S. E. Inzucchi, F. Tang et al., “Real-world use and Mellitus,” Circulation, vol. 139, no. 17, pp. 2022–2031, 2019. modeled impact of glucose-lowering therapies evaluated in [40] G. Schernthaner, H. Drexel, E. Moshkovich et al., “SGLT2 recent cardiovascular outcomes trials: an NCDR® Research inhibitors in T2D and associated comorbidities - differentiat- to Practice project,” European Journal of Preventive Cardiol- ing within the class,” BMC Endocrine Disorders, vol. 19, ogy, vol. 24, no. 15, pp. 1637–1645, 2017. no. 1, p. 64, 2019. [26] P. M. Rothwell, “Factors that can affect the external validity of [41] F. Cosentino, P. J. Grant, V. Aboyans et al., “2019 ESC Guide- randomised controlled trials,” PLoS Clinical Trials, vol. 1, lines on diabetes, pre-diabetes, and cardiovascular diseases no. 1, p. e9, 2006. developed in collaboration with the EASD,” European Heart [27] C. Saunders, C. D. Byrne, B. Guthrie et al., “External validity of Journal, 2019. randomized controlled trials of glycaemic control and vascular disease: how representative are participants?,” Diabetic Medi- cine, vol. 30, no. 3, pp. 300–308, 2013. [28] A. Sen, A. Goldstein, S. Chakrabarti et al., “The representative- ness of eligible patients in type 2 diabetes trials: a case study using GIST 2.0,” Journal of the American Medical Informatics Association, vol. 25, no. 3, pp. 239–247, 2017. [29] S. Seidu, S. K. Kunutsor, X. Cos, S. Gillani, K. Khunti, and For and on behalf of Primary Care Diabetes Europe, “SGLT2 inhibitors and renal outcomes in type 2 diabetes with or with- out renal impairment: a systematic review and meta-analysis,” Primary Care Diabetes, vol. 12, no. 3, pp. 265–283, 2018. [30] V. Vallon and S. C. Thomson, “Diabetes mellitus: cardiovascu- lar and renal benefits of SGLT2 inhibition: insights from CAN- VAS,” Nature Reviews Nephrology, vol. 13, no. 9, pp. 517-518, 2017. [31] C. Lo, T. Toyama, Y. Wang et al., “Insulin and glucose- lowering agents for treating people with diabetes and chronic kidney disease,” The Cochrane Database of Systematic Reviews, vol. 9, 2018. [32] C. Wanner, S. E. Inzucchi, J. M. Lachin et al., “Empagliflozin and progression of kidney disease in type 2 diabetes,” The New England Journal of Medicine, vol. 375, no. 4, pp. 323– 334, 2016. [33] M. J. Jardine, K. W. Mahaffey, B. Neal et al., “The Canagliflozin and Renal Endpoints in Diabetes with Established Nephropa- thy Clinical Evaluation (CREDENCE) study rationale, design, and baseline characteristics,” American Journal of Nephrology, vol. 46, no. 6, pp. 462–472, 2017. [34] H. Jakher, T. I. Chang, M. Tan, and K. W. Mahaffey, “Canagli- flozin review - safety and efficacy profile in patients with T2DM,” Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy, vol. Volume 12, pp. 209–215, 2019. [35] M. J. Davies, D. A. D’Alessio, J. Fradkin et al., “Management of hyperglycemia in type 2 diabetes, 2018. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD),” Diabetes Care, vol. 41, no. 12, pp. 2669–2701, 2018. [36] M. E. Tucker, “FDA approves expanded uses for type 2 diabe- tes drugs. Medscape Medical News,” 2019, March 2019, https://www.medscape.com/viewarticle/909713. [37] R. Pecoits-Filho and V. Perkovic, “Are SGLT2 inhibitors ready for prime time for CKD?,” Clinical Journal of the American Society of Nephrology, vol. 13, no. 2, pp. 318–320, 2018. [38] V. Perkovic, D. de Zeeuw, K. W. Mahaffey et al., “Canagliflozin and renal outcomes in type 2 diabetes: results from the CAN- VAS program randomised clinical trials,” The Lancet Diabetes and Endocrinology, vol. 6, no. 9, pp. 691–704, 2018.