Acta Diabetol DOI 10.1007/s00592-016-0892-7

ORIGINAL ARTICLE

Effects of SGLT-2 inhibitors on mortality and cardiovascular events: a comprehensive meta-analysis of randomized controlled trials

1 1 1 Matteo Monami • Ilaria Dicembrini • Edoardo Mannucci

Received: 14 June 2016 / Accepted: 21 July 2016 Ó Springer-Verlag Italia 2016

Abstract [0.63–0.94], p \ 0.01), but not stroke (MH-OR 1.09 Aims EMPAREG OUTCOME study showed a reduction in [0.86–1.38], p = 0.50), with no apparent difference across cardiovascular events in patients treated with the sodium- molecules (after excluding cardiovascular outcome trials). glucose transporter 2 inhibitor (SGLT2i) empagliflozin, as Conclusions Available data suggest that the beneficial compared to placebo. Other drugs of the same class are action observed with empagliflozin on all-cause and car- currently been investigated for cardiovascular outcomes. In diovascular mortality in EMPAREG OUTCOME study is a the meanwhile, a re-analysis of data collected in available class effect. The present meta-analysis showed a signifi- studies can add relevant insight. cantly reduction in myocardial infarction, with no Methods A MEDLINE search for SGLT-2 inhibitors (da- increased risk of stroke. pagliflozin, empagliflozin, canagliflozin, ipragliflozin, ertugliflozin, luseogliflozin) was performed, collecting all Keywords SGLT-2 inhibitors Á Cardiovascular events Á randomized trials up to November 16, 2015. All trials with Mortality Á Meta-analysis a duration of treatment C12 weeks, enrolling patients with type 2 diabetes, comparing a SGLT2i with placebo or other comparators were included. The principal outcome was the Introduction effect of SGLT2i on all-cause and cardiovascular mortality. Secondary endpoints were myocardial infarction and The EMPAREG OUTCOME study recently showed a stroke. Mantel–Haenszel odds ratio with 95 % confidence reduction in major cardiovascular events in patients treated interval (MH-OR) was calculated. with the sodium-glucose transporter 2 (SGLT2) inhibitor Results A total of 71 trials were included (31,199 and 16,088 empagliflozin, as compared to placebo [1]. This result was patients in SGLT2i and comparator groups). Treatment with beyond the expectations of most clinicians, considering SGLT2i was associated with a significant reduction in all- that the trial had been designed for safety purposes (with cause mortality (MH-OR 0.70 [0.59–0.83], p \ 0.001), non-inferiority as the primary endpoint) and that previous cardiovascular mortality (MH-OR 0.43 [0.36–0.53], studies in type 2 diabetes with similar design had failed to p \ 0.001), and myocardial infarction (MH-OR 0.77 detect any significant beneficial effect of tested drugs [2–5]. The most striking result of the EMPAREG OUT- Managed by Massimo Federici. COME study was an impressive reduction in all-cause and cardiovascular mortality, which became evident after a few & Matteo Monami weeks from the beginning of treatment [1].Those results matteo.monami@unifi.it could modify current attitudes in the choice of drugs for the Edoardo Mannucci treatment of type 2 diabetes. edoardo.mannucci@unifi.it Landmark studies, while providing essential evidence, often open a number of novel research questions. This is 1 Diabetology, Azienda Ospedaliero-Universitaria Careggi, Careggi Teaching Hospital, University of Florence, Via delle the case of the EMPAREG OUTCOME study [1]. Mech- Oblate 4, 50141 Florence, Italy anisms underlying the observed reduction in cardiovascular 123 Acta Diabetol mortality, which cannot be entirely explained by the events (myocardial infarction, stroke), verifying differ- reduction in the incidence of major cardiovascular events, ences across molecules of the class, and including all remain unknown. Several hypotheses have been formulated available evidence. in this regard [6–8], but the contribution of the proposed mechanisms has not been explored so far. In fact, we do not even know whether the reduction in cardiovascular mor- Methods tality with empagliflozin was determined by the inhibition of SGLT2 or by other, still unidentified, SGLT2-indepen- The review protocol (CRD42015029573) was published on dent mechanisms. Due to this lack of information, we are the University of York (Centre for Reviews and Dissemi- unable to predict whether the benefit of empagliflozin on nation) Web site [14]. mortality is a class effect or a specific feature of that molecule; the ongoing cardiovascular studies with the other Data sources and searches SGLT-2 inhibitors canagliflozin, dapagliflozin, and ertu- gliflozin [9–11] are not bound to produce any results in the A MEDLINE search for SGLT-2 inhibitors (dapagliflozin, next 2 years. empagliflozin, canagliflozin, ipragliflozin, ertugliflozin, In the EMPAREG OUTCOME study, empagliflozin luseogliflozin) was performed, collecting all randomized produced a significant benefit in the incidence of major clinical trials on humans up to November 16, 2015. The cardiovascular events, hospitalization for heart failure, and identification of relevant abstracts, the selection of studies all-cause and cardiovascular mortality. For other cardio- based on the criteria described below, and the subsequent vascular endpoints (e.g., myocardial infarction), a non- data extraction were performed independently by two of significant trend was reported, the only notable exception the authors (I.D., M.M.), and conflicts resolved by the third being the incidence of stroke, with an estimated hazard investigator (E.M.). Completed but still unpublished trials ratio of 1.18 and 1.24 for total and non-fatal cases, were identified through a search of www.clinicaltrials.gov respectively [1]. Interestingly, an estimated hazard ratio of Web site, using the same keywords. In addition, for stroke of 1.46, although with a small number of overall approved drugs, medical reviews were retrieved from the events, had been reported in a pooled analysis of trials Food and Drug Administration (FDA) Web site [15], and submitted to the Food and Drug Administration for the the summary of product characteristics from the European approval of canagliflozin [12]. The safety of SGLT-2 Medicines Agency (EMA) Web site [16], for the identifi- inhibitors with respect to the incidence of stroke deserves cation of further unpublished and otherwise undisclosed further investigation. trials. Other drugs of the same class are currently been investigated for cardiovascular outcomes [9–11], in order Study selection to comply with regulatory requirements. When all ongoing trials are completed, a comprehensive picture of the car- A meta-analysis was performed including all randomized diovascular effects of SGLT-2 inhibitors will be available. clinical trials with a duration of treatment of at least In the meanwhile, a re-analysis of data collected in avail- 12 weeks, enrolling patients with type 2 diabetes, com- able studies, including phase II–III trials, can add relevant paring a SGLT2 inhibitor with placebo or any other non- insight. Recently, such a meta-analysis has been per- SGLT2 inhibitor drug, provided that concurrent treatment formed, showing (for the whole class) a significant reduc- was the same for all treatment arms. tion in the hospitalization for heart failure and cardiovascular and all-cause mortality, which was largely Data extraction and quality assessment driven by the EMPAREG OUTCOME study, with no significant effect on non-fatal myocardial infarction [13]. Results of trials were retrieved from the primary publica- The same meta-analysis also showed a significant increase tion and, if needed, from other publications referring to the in the risk of stroke. Notably, the meta-analysis was per- same trial. When an information was unavailable on pub- formed collecting data only from published trials or regu- lished papers, data were retrieved (in this hierarchical latory submission, without any attempt at retrieving results order) from FDA medical reviews, EMA summaries of from unpublished trials which were not submitted to reg- product characteristics, www.clinicaltrials.gov study ulatory agencies; this means that the database did not results, and trial results on the manufacturers’ company include all available data. Web sites. Retrieved data included all outcomes reported Aim of the present meta-analysis is the assessment of below, plus the main features of each trial (concurrent the effects of SGLT-2 inhibitors on all-cause and cardio- therapy, principal endpoint, and inclusion criteria for age, vascular mortality, and on selected major cardiovascular body mass index [BMI], and HbA1c), and baseline 123 Acta Diabetol characteristics of enrolled patients (age, BMI, duration of The quality of trials (all intention-to-treat) was generally diabetes, HbA1c). The quality of trials was assessed using good (Table 1). The trials fulfilling the inclusion criteria the Cochrane Collaboration’s tool for assessing risk of bias enrolled 31,199 and 16,088 patients in SGLT2 inhibitor in randomized controlled trials; quality was not used as a and comparator arms, respectively, with a mean duration of criterion for the selection of trials, but only for descriptive treatment of 40.2 weeks. The main characteristics of the purposes. selected trials, and the outcomes of interest in each study, are reported in Tables 1 and 2. The search of www.clin Data synthesis and analysis icaltrials.gov Web site allowed the identification of six unpublished and undisclosed, although completed, trials The principal outcome of this analysis was the effect of (Table 3). SGLT-2 inhibitors, compared with placebo or other active drugs, on all-cause and cardiovascular mortality. In addi- All-cause mortality tion, the effect of SGLT-2 inhibitors on the incidence of overall (fatal plus non-fatal) myocardial infarction and Of the 71 trials fulfilling the inclusion criteria, 6 did not stroke was assessed. The composite endpoint of major report information on all-cause mortality, whereas 33 cardiovascular events (MACE) was not considered, reported zero events in all treatment groups. The principal because it is not usually reported as such in trials with non- analysis was therefore performed on 32 trials, enrolling cardiovascular principal endpoints, which were the 26,856 and 14,090 patients in SGLT2 inhibitor and com- majority of available studies. Heterogeneity (on all-cause parator arms, respectively, with a mean duration of mortality) was assessed by using I [2] statistics. In order to 41.3 weeks. The number of reported deaths was 342 for estimate possible publication/disclosure bias, we used SGLT-2 inhibitors and 243 for comparators, with an esti- funnel plots and the Begg’s adjusted rank correlation test mated yearly mortality rate of 1.01 and 1.37 %, respectively. [17, 18], including published and unpublished, but dis- For the principal endpoint, I2 was\0.001, suggesting no closed, trials. Considering the differences across trials in relevant heterogeneity. Funnel plot analysis (Fig. 2) and molecules, treatment schedules, inclusion criteria, and Kendall’s tau (-0.14; p = 0.26) did not suggest any rele- length of follow-up, a random-effects model was applied, vant publication bias. calculating Mantel–Haenszel odds ratio with 95 % confi- Overall, treatment with SGLT-2 inhibitors was associ- dence interval (MH-OR) for all the events defined above, ated with a significant reduction in all-cause mortality on an intention-to-treat basis, excluding trials with zero (MH-OR [95 % CI] 0.70 [0.59–0.83], p \ 0.001; Fig. 3). events. For all-cause mortality, a sensitivity analysis was Similar results were obtained with fixed effects model performed with continuity correction, in order to avoid Peto’s OR (0.70 [0.58–0.83], p \ 0.001). The sensitivity distortions due to the exclusion of trials with zero events; analysis with continuity correction confirmed the same a further sensitivity analysis was performed calculating result (MH-OR 0.67 [0.57–0.78], p \ 0.001). When trials Peto’s OR (fixed effects model). Subgroup analyses were with different molecules were analyzed separately, the performed for all endpoints for different drugs of the results with canagliflozin (N = 7 trials), dapagliflozin class, different classes of comparators, trials with car- (N = 13), and empagliflozin (N = 12) were MH-OR 0.80 diovascular and non-cardiovascular endpoints, and trials [0.45–1.42] (p = 0.45), 0.78 [0.41–1.48] (p = 0.45), and with different inclusion criteria (maximum age, maximum 0.68 [0.56–0.82] (p \ 0.0001). BMI, and maximum HbA1c). Further analyses were per- These results were largely driven by the EMPAREG formed including only trials with molecules and doses OUTCOME study [1], which was one of the two included approved by FDA, EMA, or both (dapagliflozin 10/day, trials with a cardiovascular endpoint, together with an empagliflozin 10 and 25/day, canagliflozin 100 and interim analysis of the CANVAS study with canagliflozin 300/day). All analyses were performed using Compre- [20]. When trials with non-cardiovascular endpoints were hensive Meta-Analysis version 2, Biostat (Englewood, analyzed separately, the reduction in all-cause mortality NJ, USA).The meta-analysis was reported following the with SGLT-2 inhibitors was no longer significant, with no PRISMA checklist [19]. apparent difference across molecules. However, when the sensitivity analysis with continuity correction was per- formed, the reduction in mortality was significant, again Results with similar results for different molecules of the class (Fig. 4). A further analysis was performed including only Out of 139, 235, and 145 items identified through MED- molecules and doses approved by FDA and/or EMA. The LINE, www.clinicaltrials.gov, and FDA/EMA Web sites, results were similar to the previous analysis as reported in respectively, 71 trials were selected, as shown in Fig. 1. Fig. 4. 123 Acta Diabetol

Fig. 1 Trial flow summary

Few of the trials with events compared SGLT-2 inhi- (p \ 0.0001), respectively. When trials with non-cardio- bitors with active comparators (three with DPP4 inhibitors, vascular endpoints were analyzed separately, the reduction three with sulfonylureas, and four with metformin), with no in cardiovascular mortality with SGLT-2 inhibitors was statistically significant differences. The results of the still significant, with no apparent difference across mole- analysis of placebo-controlled trials were similar to those cules (Fig. 7). Similar results were obtained when of the one including all trials, reported above (Fig. 5). restricting the analysis to placebo-controlled trials only Separate analyses were performed for trials on the basis of (Fig. 8). the main inclusion criteria. The effect of SGLT2 on mor- tality was significant only in trials in which the maximum Myocardial infarction HbA1c was over 10 % (86 mmol/mol), the maximum BMI greater than 45 kg/m [2], and the maximum age lower than Information on myocardial infarction was reported for 64 81 years. However, those results were again driven by the trials, 36 of which with at least one event. SGLT-2 inhi- EMPAREG OUTCOME study [1]; when trials with CV bitors produced a significant reduction in the incidence of endpoints were excluded, differences based on inclusion myocardial infarction (MH-OR [95 % CI] 0.77 criteria were no longer detectable (Fig. 6). [0.63–0.94], p \ 0.01); data for canagliflozin (N = 6 tri- als), dapagliflozin (N = 14), and empagliflozin (N = 15) Secondary endpoints were 0.57 [0.26–1.24] (p = 0.15), 0.48 [0.26–0.87] (p = 0.017), and 0.83 [0.67–1.04] (p = 0.10), respectively. Cardiovascular mortality The reduction in risk of myocardial infarction was not statistically significant in the only one available trial with a Only a fraction of available trials reported information on cardiovascular endpoint, the EMPAREG OUTCOME causes of death. Cardiovascular mortality was available for study [1]; conversely, in trials with non-cardiovascular 61 trials, 23 of which with at least one event. SGLT-2 endpoints, the effect of SGLT-2 inhibitors was significant, inhibitors significantly reduced cardiovascular mortality with no apparent difference across molecules (Fig. 7). The (MH-OR [95 % CI] 0.43 [0.36–0.53], p \ 0.001). MH-OR number and size of available trials versus active com- for canagliflozin (N = 7 trials), dapagliflozin (N = 9), and parators was too small to obtain reliable results for empagliflozin (N = 7) were 0.50 [0.27–0.95] (p = 0.034), myocardial infarction; the exclusion of those trials for the 0.46 [0.20–1.07] (p = 0.071), and 0.42 [0.34–0.52] analysis did not produce major changes in results (Fig. 8).

123 Acta Diabetol

Table 1 Baseline characteristics and quality assessment of trials included in the meta-analysis Study ID code Description of Mean Randomiz. Allocat. Blind. Attrit. Detect. Age Diab. dur. HbA1c BMI bias bias (years) (years) (%) (kg/m2)

Canagliflozin Rosenstock [32] NCT00642278 NA NA NA NO NO 52 6 7.7 31.3 Inagaki [33] NCT01022112 A A A NO NO 57 NR 8.1 25.7 Sha [34] NCT01483781 A NA NA NO NO 63 8 7.6 29.8 Fulcher [35] NCT01032629 A A A NO NO 65 10 8.4 29.9 Ji [36] NCT01381900 A A A NO NO 56 7 8.0 25.8 Neal [37] NCT01032629 A A A NO NO 63 16 8.3 33.1 Inagaki [38] NCT01413204 A A A NO NO 58 NR NR NR Forst [39] NCT01106690 A A A NO NO 57 10 7.9 32.5 Stenlof [40] NCT01081834 A NA A NO NO 55 4 8.0 31.6 Wilding [41] NCT01106625 A A A NO NO 57 10 8.1 33.1 Yale [42] NCT01064414 A A NA NO NO 68 16 8.0 33.0 Lavalle-Gonzalez NCT01106677 A A A NO NO 55 7 7.9 31.8 [43] CANVAS [9] NCT01032629 A A A NO NO 62 13 8.2 32.1 Bode [44] NCT01106651 A A A NO NO 64 12 7.7 31.6 Schernthaner [45] NCT01137812 A A A NO NO 56 10 8.1 31.6 Cefalu [46] NCT00968812 A A A NO NO 56 6 7.8 31.0 NCT01340664 [47] NCT01340664 NA NA A NO NO 57 NR NR NR Dapagliflozin Kaku [48] NCT00972244 A A NA NO NO 57 5 8.0 NR List [49] NCT00263276 NA NA NA NO NO 55 NR 7.7 31.7 Wilding [50] NCT00357370 A NA NA NO NO 56 12 8.4 35.5 Heerspink [51] NCT00976495 A A A NO NO 56 6 7.6 NR Weber [52] NCT01195662 A A A NO NO 56 7 8.0 nr Mudaliar [53] NR NA NA A NO NO 54 7 7.5 NR NCT01137474 [54] NCT01137474 NA NA A YES NO NR NR NR NR Schumm-Draeger NCT01217892 A A A NO NO 58 5 7.8 32.7 [55] Henry 2012 (Study NCT00643851 A A A NO NO 52 2 9.2 NR 1) [56] Henry 2012 (Study NCT00859898 A A A NO NO 52 2 9.1 NR 2) [56] Rosenstock [57] NCT01606007 A A A NO NO NR 8 8.9 31.7 Bailey [58] NCT00736879 A A NA NO NO 53 1 7.9 31.8 Jabbour [59] NCT00984867 A NA NA NO NO 55 6 7.9 NR Mathieu [60] NCT01646320 A A A NO NO 55 NR 8.2 31.7 Kaku [61] NCT01294423 NA NA NA NO NO 59 5 7.5 25.3 Strojek [62] NCT00680745 A A A NO NO 60 7 8.1 NR Cefalu [63] NCT01031680 A A A NO NO 63 NR 8.1 NR Ji [64] NCT01095653 A A A NO NO 51 1 8.2 25.4 Yang [65] NCT01095666 A NA OL NO NO 54 5 8.1 26.1 Rosenstock [66] NCT00683878 A NA NA NO NO 53 6 8.4 NR Matthaei [67] NCT01392677 NA NA A NO NO 61 NR 8.2 32.0 Leiter [68] NCT01042977 A NA NA NO NO 64 13 8.0 32.8 Bolinder [69] NCT00855166 A A A NO NO 61 6 7.2 31.9 Bailey [70] NCT00528372 A A A NO NO 52 2 7.9 32.6

123 Acta Diabetol

Table 1 continued Study ID code Description of Mean Randomiz. Allocat. Blind. Attrit. Detect. Age Diab. dur. HbA1c BMI bias bias (years) (years) (%) (kg/m2)

Bailey [71] NCT00528879 A A A NO NO 54 6 8.0 31.5 Kohan [72] NCT00663260 A NA NA NO NO 67 17 8.3 NR Wilding [73] NCT00673231 A A A NO NO 59 13 8.5 33.1 List [49] NCT00263276 NA NA NA NO NO 55 NR 7.7 31.7 Nauck [74] NCT00660907 A A A NO NO 58 6 7.7 31.4 Empagliflozin Ferrannini 2013 NCT00881530 A A OL NO NO 58 NR 8.0 28.5 (Study 1) [75] Ferrannini 2013 NCT00881530 A A OL NO NO 58 NR 7.9 30.1 (Study 2) [75] Rosenstock [76] NCT00749190 A A A NO NO 59 NR 8.0 31.3 Ferrannini [75] NCT00789035 A A A NO NO 58 NR 7.9 28.5 Kadowaki [77] NCT01193218 A A A NO NO 57 NR 7.9 25.4 Tikkanen [78] NCT01370005 A A A NO NO 60 NR 7.9 32.6 NCT01649297 [79] NCT01649297 NA NA NA NO NO 58 NR NR NR Ha¨ring [80] NCT01159600 A A A NO NO 57 NR 8.1 28.2 Roden [81] NCT01177813 A A A NO NO 55 NR 7.9 28.4 Barnett [40] NCT01164501 A A A NO NO 64 NR 7.9 28.4 Rosenstock [83] NCT01306214 A A A NO NO 57 NR 8.3 34.8 DeFronzo [84] NCT01422876 A A A NO NO 55 NR NR NR Lewin [85] NCT01422876 A A A NO NO 55 NR 8.0 31.5 Kovacs [86] NCT01289990 A A A NO NO 54 NR 8.1 29.2 Merker [87] NCT01289990 and A A A NO NO 56 NR 7.9 29.2 NCT01159600 Rosenstock [88] NCT01011868 A A A NO NO 59 NR 8.2 32.2 Zinman [1] NCT01131676 A A A NO NO 63 NR 8.1 30.6 Araki [89] NCT01368081 A A OL NO NO 61 NR 8.0 25.0 Ridderstra˚le [90] NCT01167881 A A A NO NO 56 NR 8.0 31.6 Ertugliflozin Amin [91] NCT01059825 A A A A NO 54 6 8.1 30.6 Ipragliflozin Wilding [92] NCT01117584 NA NA NA NO NO 57 6 7.7 31.7 Fonseca [93] NCT01071850 A NA NA NO NO 53 4 7.8 31.2 Kashiwagi [94] NCT01135433 NA NA NA NO NO 57 8 8.3 25.6 Kashiwagi [95] NCT01316094 A NA NA NO NO 64 10 7.5 25.4 Kashiwagi [96] NCT00621868 NA NA NA NO NO 56 7 8.3 25.7 Randomiz. randomization, Allocat. allocation, Blind. blinding, Attrit. attrition, Detect. detection, Diab. dur. diabetes duration, A adequately described, NA not adequately described, OL open label, NR not reported

Stroke risk of stroke (MH-OR) associated with SGLT2 inhibitor therapy was 1.36 [0.54–3.40] for canagliflozin (N = 8 Informationonstrokewasreportedby63studies,29of trials; p = 0.51), 0.68 [0.31–1.46] for dapagliflozin which with no events. The analysis was therefore per- (N = 11; p = 0.32), and 1.13 [0.87–1.46] for empagli- formed on 34 trials with at least one event. The inci- flozin (N = 15, p = 0.37), respectively. No significant dence of stroke with SGLT-2 inhibitors was not differences between SGLT-2 inhibitors, either as indi- significantly different from that of comparator groups vidual molecules or as a class, were detected after the (MH-OR [95 % CI] 1.09 [0.86–1.38], p = 0.50). The exclusion of trials with cardiovascular endpoints (Fig. 7),

123 caDiabetol Acta Table 2 Baseline characteristics and quality assessment of trials included in the meta-analysis Study Comparator Dose SGLT2i (mg/day) Trial duration (weeks) Number of All-cause death Cardiovascular Myocardial Stroke Drug patients death infarction SGLT2i Comp. SGLT2i Comp. SGLT2i Comp. SGLT2i Comp. SGLT2i Comp.

Canagliflozin Rosenstock [32] Placebo 50–600 12 321 65 0 0 0 0 0 0 0 0 Sitagliptin 321 65 0 0 0 0 0 0 0 0 Inagaki [33] Placebo 50–300 12 307 75 0 0 0 0 0 0 0 0 Sha [34] Placebo 300 12 18 18 0 0 0 0 0 0 0 0 Fulcher [35] Placebo 100–300 18 82 45 0 0 0 0 0 0 0 0 Ji [36] Placebo 100–300 18 445 224 0 0 0 0 0 0 0 1 Neal [37] Placebo 100–300 18 1382 690 20 13 14 11 NR NR NR NR Inagaki [38] Placebo 100–300 24 179 93 NR NR NR NR 0 0 0 0 Forst [39] Placebo 100–300 26 227 115 0 0 0 0 0 0 0 0 Stenlof [40] Placebo 100–300 26 392 192 1 1 0 0 0 0 0 1 Wildinga [41] Placebo 100–300 52 313 156 0 0 0 0 0 1 1 0 Yale [42] Placebo 100–300 52 179 90 2 1 0 0 3 3 2 0 Inagaki [33] Placebo 100–300 52 735 183 1 1 1 0 1 0 2 0 Sitagliptin 735 366 1 1 1 1 1 1 2 1 CANVASa [9] Placebo 100–300 52 2886 1441 NR NR NR NR NR NR NR NR Bode [44] Placebo 100–300 104 477 237 2 0 1 0 5 3 4 0 Schernthaner [45] Sitagliptin 300 52 377 378 2 1 2 1 0 2 1 0 Cefalu [46] Glimepiride 100–300 52 952 473 3 2 0 1 4 3 7 2 NCT01340664 [47] Placebo 50–150 18 186 93 NR NR NR NR 0 0 0 0 Dapagliflozin Kaku [48] Placebo 1–10 12 225 54 1 0 0 0 1 0 0 0 List [49] Placebo 2.5–50 12 279 54 0 0 0 0 1 0 0 0 Wilding [50] Placebo 10–20 12 48 23 0 0 0 0 0 0 0 0 Heerspink [51] Placebo 10 12 24 25 0 0 0 0 0 0 0 0 Weber [52] Placebo 10 12 225 224 0 0 0 0 0 0 0 0 Mudaliar [53] Placebo 5 12 23 21 0 0 0 0 0 0 0 0 NCT01137474 [54] Placebo 2.5–10 12 302 311 NR NR NR NR 0 0 0 1 Schumm-Draeger [55] Placebo 5–10 16 299 101 0 0 0 0 0 0 0 0 Henry (Study 1) [56] Placebo 5 24 194 201 0 0 0 0 1 0 0 1 Metformin 203 201 1 0 1 0 0 0 0 0 123 Henry (Study 2) [56] Placebo 10 24 211 208 0 1 0 1 1 1 0 0 Metformin 219 208 0 1 0 1 0 1 0 0 Rosenstock [57] Placebo 10 24 179 176 0 0 0 0 1 0 0 0 123 Table 2 continued Study Comparator Dose SGLT2i (mg/day) Trial duration (weeks) Number of All-cause death Cardiovascular Myocardial Stroke Drug patients death infarction SGLT2i Comp. SGLT2i Comp. SGLT2i Comp. SGLT2i Comp. SGLT2i Comp.

Saxagliptin 179 176 0 0 0 0 0 0 0 0 Bailey [58] Placebo 2.5–10 24 214 68 0 0 0 0 0 0 0 0 Jabbour [59] Placebo 10 24 223 224 0 1 0 0 0 0 0 0 Mathieu [60] Placebo 10 24 160 160 0 0 0 0 0 0 0 0 Kaku [61] Placebo 5–10 24 174 87 0 0 0 0 0 0 0 1 Strojek [62] Placebo 2.5–10 24 447 145 2 0 0 0 1 1 2 0 Cefalu [63] Placebo 10 24 455 459 2 1 2 1 1 1 2 2 Ji [64] Placebo 5–10 24 261 132 0 0 0 0 NR NR NR NR Yang [65] Placebo 5–10 24 299 145 0 0 0 0 NR NR NR NR Rosenstock [66] Placebo 5–10 48 281 139 1 0 0 0 0 0 0 0 Matthaeia [67] Placebo 10 52 109 109 0 0 0 0 0 0 0 0 Leiterb [68] Placebo 10 52 480 482 5 4 2 1 3 2 1 3 Bolinder [69] Placebo 10 102 89 41 0 0 0 0 0 1 0 0 Bailey [70] Placebo 2.5–10 102 199 75 1 0 0 0 2 0 1 0 Bailey [71] Placebo 2.5–10 104 409 137 2 1 2 0 2 3 1 1 Kohan [72] Placebo 5–10 104 168 84 5 5 1 1 2 5 2 0 Wilding [73] Placebo 2,5–10 104 607 193 3 0 3 0 2 3 3 2 List [49] Metformin 2.5–50 12 279 56 0 0 0 0 0 0 0 0 Nauck [74] Glipizide 10 104 400 401 0 4 0 2 3 5 3 1 Empagliflozin Ferrannini (Study 1) [75] Metformin 10–25 78 215 56 0 1 NR NR 1 0 1 0 Ferrannini (Study 2) [75] Sitagliptin 10–25 78 332 56 0 0 NR NR 1 0 1 3 Rosenstock [76] Placebo 1–50 12 353 71 0 0 0 0 1 1 0 0 Sitagliptin 353 71 0 0 0 0 1 0 0 0 Ferrannini [75] Placebo 5–25 12 244 82 0 0 0 0 2 0 0 0 Metformin 244 80 0 0 0 0 2 0 0 0 Kadowaki [77] Placebo 5–50 12 438 109 0 0 0 0 0 2 1 0 Tikkanen [78] Placebo 10–25 12 552 271 1 0 1 0 0 0 1 0 NCT01649297 [79] Placebo 10–25 16 858 107 NR NR NR NR 0 0 1 1 Ha¨ring [80] Placebo 10–25 24 441 225 1 0 1 0 1 1 1 0 caDiabetol Acta Roden [81] Placebo 10–25 24 448 228 0 1 0 1 1 1 1 1 Sitagliptin 448 223 0 1 0 1 0 1 0 1 Barnett [82] Placebo 10–25 24 419 319 1 3 NR NR 0 2 4 1 Rosenstock [83] Placebo 10–25 52 375 188 1 0 0 0 1 0 1 1 caDiabetol Acta Table 2 continued Study Comparator Dose SGLT2i (mg/day) Trial duration (weeks) Number of All-cause death Cardiovascular Myocardial Stroke Drug patients death infarction SGLT2i Comp. SGLT2i Comp. SGLT2i Comp. SGLT2i Comp. SGLT2i Comp.

DeFronzo [84] Placebo 10–25 52 273 132 1 0 0 0 0 0 2 1 Metformin 281 132 1 0 0 0 2 0 0 1 Lewin [85] Placebo 10–25 52 273 135 1 0 1 0 0 0 NR NR Linagliptin 269 135 3 0 0 0 0 0 NR NR Kovacsa [86] Placebo 10–25 76 333 165 4 1 2 1 0 0 1 0 Merker [87] Placebo 10–25 76 431 206 0 0 0 0 2 2 2 2 Rosenstock [88] Placebo 10–25 78 324 170 0 1 NR NR 3 0 3 1 Zinman [1] Placebo 10–25 161 4687 2333 269 194 172 137 213 121 164 69 Araki [89] Metformin 10–25 52 273 63 0 0 0 0 1 0 0 0 Ridderstra˚le [90] Glimepiride 25 104 769 780 5 5 1 2 1 7 7 3 Ertugliflozin Amin [91] Placebo 1–25 12 219 55 0 0 0 0 2 0 0 0 Sitagliptin 219 54 0 0 0 0 2 0 0 0 Ipragliflozin Wilding [92] Placebo 12.5–300 12 276 66 0 0 0 0 0 0 0 0 Fonseca [93] Placebo 12.5–300 12 273 69 0 0 0 0 NR NR NR NR Metformin 273 69 0 0 0 0 NR NR NR NR Kashiwagi [94] Placebo 50 24 112 56 0 0 0 0 0 0 0 0 Kashiwagi [95] Placebo 50 24 118 46 NR NR NR NR NR NR NR NR Kashiwagi [96] Placebo 12.5–100 24 291 69 0 0 0 0 NR NR NR NR SGLT2i sodium-glucose transporter 2 inhibitors, Comp. comparator, NR not reported Data for myocardial infarction and stroke available only at a 24 weeks and b 26 weeks 123 Acta Diabetol

Table 3 Principal characteristics of unpublished and undisclosed trials NCT code Other ID SGLT2i Comparator Enrollment (n. patients) Trial duration (weeks) Sponsor

NCT02025907 CR103477 Canagliflozin Placebo 218 28 Jansenn NCT01809327 MB102-045 Canagliflozin Placebo 44 12 Astra Zeneca NCT00831779 1275.9 Dapagliflozin Placebo 607 24 Boehringer NCT01734785 1941-CL-2004 Empagliflozin Placebo 171 24 Astellas NCT01719003 CR100034 Empagliflozin Metformin 1185 26 Jansenn NCT01505426 1276.1 Ipragliflozin Metformin 1418 24 Boehringer SGLT2i sodium-glucose transporter 2 inhibitors, n. number

Fig. 2 Funnel plot of standard error by MH log odds ratio for all-cause mortality

or after the exclusion of trials versus active comparators and doses approved by EMA and/or FDA) support the (Fig. 8). robustness of this effect. However, the result is largely driven by the EMPAREG OUTCOME study [1], which accounts for 79 % of all reported deaths. When the trials Discussion with cardiovascular endpoints (i.e., the EMPAREG OUT- COME study [1] and the CANVAS trial interim analysis The reduction in cardiovascular and all-cause mortality [21]) were excluded, the reduction in cardiovascular mor- observed in the EMPAREG OUTCOME study was much tality was no longer significant. Smaller, non-cardiovas- greater than that of major cardiovascular events, and it was cular trials were performed in populations different from already evident after a few weeks of treatment [1]. Such a those enrolled in cardiovascular outcome studies, with a result cannot be attributed to the prevention of lower mortality rate. The small number of events could atherosclerosis, rather suggesting alternative and yet have prevented the detection of relevant differences unknown protective mechanisms [6–8, 20]. The lack of between groups. In addition, the result was affected by the information on mechanisms underlying the reduction in exclusion of a relatively high number of trials with no mortality does not allow the prediction of the effects of events. In many of those latter studies, the proportion of SGLT-2 inhibitors other than empagliflozin. In other patients allocated to SGLT-2 inhibitors was higher than words, we do not know whether the beneficial actions of comparators; this explains the greater reduction in mor- empagliflozin are specific of the molecule or characteristic tality in the analysis with continuity correction, which of the whole class. reached statistical significance. In both analyses (with and The present meta-analysis shows that SGLT-2 inhibi- without continuity correction), after excluding trials with tors, as a class, are associated with a reduction in all-cause cardiovascular endpoints, the estimated effects on mortality mortality. The concordant results of sensitivity analyses were similar for all SGLT-2 inhibitors. These results do not (with fixed effect model, or with continuity correction, or support the hypothesis of differences in the effects on all- including only placebo-controlled trials, or only molecules cause mortality across the molecules of the class.

123 Acta Diabetol

Fig. 3 MH-OR for all-cause mortality with 95 % CI. SGLT2i sodium-glucose transporter 2 inhibitors

Data on cardiovascular mortality were not available for not for dapagliflozin [24, 25] or canagliflozin [12, 21]. all trials. In fact, only a fraction of studies with non-car- Unfortunately, such endpoint could not be used in this diovascular endpoints reported causes of death. In addition, meta-analysis, because non-cardiovascular trials usually the classification of causes of death can be problematic, report single events, but not composite endpoints. particularly in those trials in which a formal adjudication is Therefore, we chose to explore the endpoints individu- not provided. Despite these limitations, the meta-analysis ally: cardiovascular mortality (see above), myocardial clearly shows that SGLT-2 inhibitors markedly reduced infarction (including both fatal and non-fatal), and cardiovascular mortality, either versus placebo or versus all stroke. comparators. This result retained statistical significance For myocardial infarction, SGLT-2 inhibitors as a class even after excluding cardiovascular outcome trials. Again, were associated with a significant reduction in risk, which no relevant differences were apparent across molecules of was confirmed when the analysis was restricted to placebo- the class. controlled trials only. A similar result had been recently The typical endpoint for cardiovascular outcome trials is reported by a meta-analysis of published trials, on a smaller major cardiovascular events (MACE), i.e., a composite of number of studies, which did not use regulatory submis- non-fatal myocardial infarction, non-fatal stroke, and car- sions as a source for identification of trials, and which diovascular mortality, sometimes with the addition of included only larger studied, while posing no minimum unstable angina. Both the EMPAREG OUTCOME [1] and duration limit [26]. However, in that meta-analysis, the CANVAS [9] studies had three-point MACE as their interpretation of results was complicated by the detection principal endpoint. For MACE, pooled analyses of patient- of a potential publication bias, which was not present in level data submitted to regulatory authorities had shown a this meta-analysis. The difference in results can be attrib- significant risk reduction with empagliflozin [22, 23], but uted to the diversity in data sources and in trial selection

123 Acta Diabetol

Fig. 4 Subgroup analyses for all-cause mortality considering trials molecules and doses approved by FDA and/or EMA). SGLT2i with and without cardiovascular endpoints (a all randomized clinical sodium-glucose transporter 2 inhibitors trials; b sensitivity analysis with continuity correction; c only

Fig. 5 Subgroup analyses of different molecules for all-cause mortality in placebo-controlled trials. SGLT2i sodium-glucose transporter 2 inhibitors

criteria, which are a key issue for meta-analyses. Interest- with non-cardiovascular endpoints, but not in the ingly, despite a smaller number of events, the reduction in EMPAREG OUTCOME study [1]. When excluding trials myocardial infarction was statistically significant in trials with cardiovascular endpoints, the protective effect was

123 Acta Diabetol

Fig. 6 Subgroup analyses for all-cause mortality on the basis of the main inclusion criteria. a all randomized clinical trials; b excluding trials with cardiovascular endpoints. SGLT2i sodium-glucose transporter 2 inhibitors significant only for dapagliflozin, but the estimates of risk unproved, is intriguing, and it deserves further with the other molecules of the class were similar (-50, investigation. -48, and -59 % for dapagliflozin, empagliflozin, and The effect of empagliflozin on stroke was a potential canagliflozin, respectively). Overall, these results suggest a weakness in the results of the EMPAREG OUTCOME protective effect of SGLT-2 inhibitors as a class with study [1]. In fact, stroke (either total or non-fatal) was the respect to the risk of myocardial infarction. The fact that a only cardiovascular endpoint for which the estimated HR therapeutic action is more evident in a population at lower was greater than 1.0, although differences from placebo did risk, and with a lower number of overall events, may seem not reach statistical significance. In addition, initial results surprising. However, similar findings have been reported from the CANVAS trial, showing a small excess of cases in for other classes of glucose-lowering drugs: DPP4 inhibi- the canagliflozin arm, had evoked concerns in some tors are associated with a reduction in cardiovascular risk members of a FDA advisory committee [30]. In fact, a in smaller trials with metabolic endpoints [27–29], but not recent meta-analysis reported a significant increase in the in cardiovascular outcome trials [2–4]. These discrepancies risk of stroke with SGLT-2 inhibitors [13]. Conversely, the could be the consequences of differences in trial design, results of the present meta-analysis are reassuring, failing with particular reference to the length of treatment and the to show any major adverse effect of SGLT-2 inhibitors on assessment of outcomes or, more probably, to the diversi- the incidence of stroke. In comparison with the previous ties in the characteristics of enrolled patients. The meta-analysis, which provided different results, the present hypothesis that patients at earlier stages of the natural analysis collects data from a larger number of trials. history of diabetes, with a lower number of concomitant However, it should be noted that the number of events, for cardioprotective therapies and a smaller burden of comor- this endpoint, is relatively small and confidence intervals bid conditions, could obtain a greater benefit from SGLT-2 are therefore wide; at present, we can only exclude an inhibitors and other treatments for diabetes, although increase in risk greater than 38 %.

123 Acta Diabetol

Fig. 7 Subgroup analyses for cardiovascular mortality (a), myocardial infarction (b), and stroke (c) considering all randomized clinical trials with and without cardiovascular endpoints. SGLT2i sodium-glucose transporter 2 inhibitors

Fig. 8 Subgroup analyses for cardiovascular mortality (a), myocardial infarction (b), and stroke (c) considering placebo-controlled trials with and without cardiovascular endpoints. SGLT2i sodium-glucose transporter 2 inhibitors

123 Acta Diabetol

Some limitations on the present meta-analysis should be Human and animal rights This article does not contain any studies recognized. The overall number and size of trials, although with animals performed by any of the authors. remarkable, is still insufficient to detect minor differences Informed consent For this type of study formal consent is not between treatment arms. The small number of trials versus required. other glucose-lowering agents does not allow any reliable comparative estimate of risk. In addition, although the number of completed, but unpublished and undisclosed References trials was rather small, the possibility of a selective dis- closure, particularly for some endpoints (e.g., cardiovas- 1. Zinman B, Wanner C, Lachin JM et al (2015) Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl cular mortality) cannot be ruled out. Conversely, the J Med 373:2117–2128 extensive search of unpublished studies, and the retrieval of 2. Scirica BM, Bhatt DL, Braunwald E et al (2013) Saxagliptin and data from different sources, is a strength of this meta- cardiovascular outcomes in patients with type 2 diabetes mellitus. analysis, which allowed the inclusion of information, N Engl J Med 369:1317–1326 3. White WB, Cannon CP, Heller SR et al (2013) Alogliptin after which had been disclosed without ever being reported on acute coronary syndrome in patients with type 2 diabetes. N Engl scientific journals. Another typical limitation of meta- J Med 369:1327–1335 analyses of cardiovascular events recorded in trials with 4. Green JB, Bethel MA, Armstrong PW et al (2015) Effect of non-cardiovascular endpoints is that, in many trials, those sitagliptin on cardiovascular outcomes in type 2 diabetes. N Engl J Med 373:232–242 events are not adjudicated [31]. However, in the case of 5. Pfeffer MA, Claggett B, Diaz R et al (2015) Lixisenatide in SGLT-2 inhibitors, the majority of non-cardiovascular tri- patients with type 2 diabetes and acute coronary syndrome. als (all those submitted to regulatory authorities), following N Engl J Med 373:2247–2257 a FDA guideline, had undergone adjudication for the 6. Ceriello A, Genovese S, Mannucci E, Gronda E (2015) Under- standing EMPA-REG OUTCOME. Lancet Diabetes Endocrinol inclusion in patient-level pooled analyses of MACE 3:929–930 [12, 23, 25]. 7. Gilbert RE, Connelly KA (2015) Understanding EMPA-REG In conclusion, available data suggest that the beneficial OUTCOME. Lancet Diabetes Endocrinol 3:930–931 action observed with empagliflozin on all-cause and car- 8. Muskiet MH, van Raalte DH, van Bommel E, Smits MM, Ton- neijck L (2015) Understanding EMPA-REG OUTCOME. Lancet diovascular mortality in the EMPAREG OUTCOME study Diabetes Endocrinol 3:928–929 [1] is a class effect. The results of randomized trials do not 9. CANVAS—CANagliflozin cardioVascular Assessment Study. highlight any specific risk of stroke associated with SGLT2 https://www.clinicaltrial.gov/ct2/show/NCT01032629?term=can inhibitor therapy. On the other hand, a reduction in the vas&rank=1. Last accessed 25 Jan 2016 10. Multicenter trial to evaluate the effect of dapagliflozin on the incidence of myocardial infarction is evident, particularly in incidence of cardiovascular events (DECLARE-TIMI58). https:// early trials enrolling patients at lower cardiovascular risk. www.clinicaltrial.gov/ct2/show/NCT01730534?term=declare? dapagliflozin&rank=1. Last accessed 25 Jan 2016 Acknowledgments This research was performed independently of 11. Cardiovascular outcomes following treatment with ertugliflozin any funding, as part of the institutional activity of the investigators. in participants with type 2 diabetes mellitus and established vascular disease (MK-8835-004). https://www.clinicaltrial.gov/ Authors contribution Matteo Monami designed the study, collected ct2/show/NCT01986881?term=ertugliflozin?cardiovascular&rank the data, performed the analysis, and wrote the manuscript. Ilaria =1. Last accessed 25 Jan 2016 Dicembrini collected the data and revised the manuscript. Edoardo 12. http://www.accessdata.fda.gov/drugsatfda_docs/nda/2013/204042 Mannucci designed the study, collected the data, performed the Orig1s000MedR.pdf. Last accessed 25 Jan 2016 analysis, and wrote the manuscript. All the authors approved the final 13. Wu JH, Foote C, Blomster J et al (2016) Effects of sodium- version of this manuscript. glucose cotransporter-2 inhibitors on cardiovascular events, death, and major safety outcomes in adults with type 2 diabetes: a Compliance with ethical standards systematic review and meta-analysis. Lancet Diabetes Endocrinol 4:411–419 Conflict of interest Matteo Monami has received speaking fees from 14. http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID= Bristol Myers Squibb, Eli-Lilly, Merck, Novonordisk, Merck, and CRD42015029573. Last accessed 25 Jan 2016 Takeda; and research grants from Bristol Myers Squibb. Ilaria 15. www.fda.gov. Last accessed on 25 Jan 2016 Dicembrini has no conflicts of interest. Edoardo Mannucci has 16. www.ema.europa.eu. Last accessed on 25 Jan 2016 received consultancy fees from Merck and Novartis; speaking fees 17. Begg CB, Mazumdar M (1994) Operating characteristics of a from Astra Zeneca, Bristol Myers Squibb, Merck, and Novartis; and rank correlation test for publication bias. Biometrics research grants from Merck, Novartis, and Takeda. 50:1088–1101 18. Egger M, Davey SG, Schneider M, Minder C (1997) Bias in Ethical standard All procedures performed in studies involving meta-analysis detected by a simple, graphical test. BMJ human participants were in accordance with the ethical standards of 315:629–634 the institutional and/or national research committee and with the 1964 19. Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Preferred Helsinki declaration and its later amendments or comparable ethical reporting items for systematic reviews and meta-analyses: the standards. PRISMA statement. Ann Intern Med 151:264–269

123 Acta Diabetol

20. Solini A (2016) Role of SGLT-2 inhibitors in the treatment of with diet and exercise: a 24-week, randomized, double-blind, type 2 diabetes mellitus. Acta Diabetol. (Epub ahead of print) placebo-controlled, Phase III study. Expert Opin Pharmacother 21. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR 15:1501–1515 _-_Public_assessment_report/human/002649/WC500156457.pdf. 39. Forst T, Guthrie R, Goldenberg R et al (2014) Efficacy and safety Last accessed on 25 Jan 2016 of canagliflozin over 52 weeks in patients with type 2 diabetes on 22. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_ background metformin and pioglitazone. Diabetes Obes Metab Product_Information/human/002677/WC500168592.pdf. Last 16:467–477 accessed 25 Jan 2016 40. Stenlo¨f K, Cefalu WT, Kim KA et al (2014) Long-term efficacy 23. http://www.accessdata.fda.gov/drugsatfda_docs/nda/2014/204629 and safety of canagliflozin monotherapy in patients with type 2 Orig1s000MedR.pdf. Last accessed 25 Jan 2016 diabetes inadequately controlled with diet and exercise: findings 24. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_ from the 52-week CANTATA-M study. Curr Med Res Opin Product_Information/human/002322/WC500136026.pdf.Last 30:163–175 accessed 25 Jan 2016 41. Wilding JP, Charpentier G, Hollander P et al (2013) Efficacy and 25. http://www.accessdata.fda.gov/drugsatfda_docs/nda/2014/202293 safety of canagliflozin in patients with type 2 diabetes mellitus Orig1s000MedR.pdf. Last accessed 25 Jan 2016 inadequately controlled with metformin and sulphonylurea: a 26. Savarese G, D’Amore C, Federici M et al (2016) Effects of randomised trial. Int J Clin Pract 67:1267–1282 dipeptidyl peptidase 4 inhibitors and sodium-glucose linked 42. Weir MR, Kline I, Xie J, Edwards R, Usiskin K (2014) Effect of cotransporter-2 inhibitors on cardiovascular events in patients canagliflozin on serum electrolytes in patients with type 2 dia- with type 2 diabetes mellitus: a meta-analysis. Int J Cardiol betes in relation to estimated glomerular filtration rate (eGFR). 220:595–601 Curr Med Res Opin 30:1759–1768 27. Monami M, Genovese S, Mannucci E (2013) Cardiovascular 43. Lavalle-Gonza´lez FJ, Januszewicz A, Davidson J et al (2013) safety of sulfonylureas: a meta-analysis of randomized clinical Efficacy and safety of canagliflozin compared with placebo and trials. Diabetes Obes Metab 15:938–953 sitagliptin in patients with type 2 diabetes on background met- 28. Monami M, Ragghianti B, Zannoni S, Vitale V, Nreu B, Man- formin monotherapy: a randomised trial. Diabetologia nucci E (2016) Identification of predictors of response to basal 56:2582–2592 insulin and DPP4 inhibitors in patients with type 2 diabetes 44. Bode B, Stenlo¨f K, Sullivan D, Fung A, Usiskin K (1995) Effi- failing to other therapies. Acta Diabetol 53:35–40 cacy and safety of canagliflozin treatment in older subjects with 29. Kim SC, Glynn RJ, Liu J, Everett BM, Goldfine AB (2014) type 2 diabetes mellitus: a randomized trial. Hosp Pract Dipeptidyl peptidase-4 inhibitors do not increase the risk of 2013(41):72–84 cardiovascular events in type 2 diabetes: a cohort study. Acta 45. Schernthaner G, Gross JL, Rosenstock J et al (2013) Canagliflozin Diabetol 51:1015–1023 compared with sitagliptin for patients with type 2 diabetes who do 30. http://www.fda.gov/downloads/AdvisoryCommittees/Committees not have adequate glycemic control with metformin plus sulfony- MeetingMaterials/Drugs/EndocrinologicandMetabolicDrugsAdv- lurea: a 52-week randomized trial. Diabetes Care 36:2508–2515 isoryCommittee/UCM339293.pdf. Last accessed 25 Jan 2016 46. Cefalu WT, Leiter LA, Yoon KH et al (2013) Efficacy and safety 31. Mahaffey KW, Roe MT, Dyke CK et al (2002) Misreporting of of canagliflozin versus glimepiride in patients with type 2 dia- myocardial infarction end points: results of adjudication by a betes inadequately controlled with metformin (CANTATA-SU): central clinical events committee in the PARAGON-B trial. 52 week results from a randomised, double-blind, phase 3 non- Second Platelet IIb/IIIa Antagonist for the Reduction of Acute inferiority trial. Lancet 382:941–950 Coronary Syndrome Events in a Global Organization Network 47. .https://clinicaltrials.gov/ct2/show/NCT01340664?term=NCT013 Trial. Am Heart J 143:242–248 40664&rank=1 32. Rosenstock J, Aggarwal N, Polidori D et al (2012) Dose-ranging 48. Kaku K, Inoue S, Matsuoka O et al (2013) Efficacy and safety of effects of canagliflozin, a sodium-glucose cotransporter 2 inhi- dapagliflozin as a monotherapy for type 2 diabetes mellitus in bitor, as add-on to metformin in subjects with type 2 diabetes. Japanese patients with inadequate glycaemic control: a phase II Diabetes Care 35:1232–1238 multicentre, randomized, double-blind, placebo-controlled trial. 33. Inagaki N, Kondo K, Yoshinari T, Maruyama N, Susuta Y, Kuki Diabetes Obes Metab 15:432–440 H (2013) Efficacy and safety of canagliflozin in Japanese patients 49. List JF, Woo V, Morales E, Tang W, Fiedorek FT (2009) with type 2 diabetes: a randomized, double-blind, placebo-con- Sodium-glucose cotransport inhibition with dapagliflozin in type trolled, 12-week study. Diabetes Obes Metab 15:1136–1145 2 diabetes. Diabetes Care 32:650–657 34. Sha S, Polidori D, Heise T et al (2014) Effect of the sodium 50. Wilding JP, Woo V, Soler NG et al (2012) Long-term efficacy of glucose co-transporter 2 inhibitor canagliflozin on plasma volume dapagliflozin in patients with type 2 diabetes mellitus receiving in patients with type 2 diabetes mellitus. Diabetes Obes Metab high doses of insulin: a randomized trial. Ann Intern Med 16:1087–1095 156:405–415 35. Fulcher G, Matthews DR, Perkovic V et al (2015) Efficacy and 51. Lambers HHJ, de Zeeuw D, Wie L, Leslie B, List J (2013) safety of canagliflozin used in conjunction with sulfonylurea in Dapagliflozin a glucose-regulating drug with diuretic properties patients with type 2 diabetes mellitus: a randomized, controlled in subjects with type 2 diabetes. Diabetes Obes Metab trial. Diabetes Ther 6:289–302 15:853–862 36. Ji L, Han P, Liu Y et al (2015) Canagliflozin in Asian patients 52. Weber MA, Mansfield TA, Cain VA, Iqbal N, Parikh S, with type 2 diabetes on metformin alone or metformin in com- Ptaszynska A (2015) Blood pressure and glycaemic effects of bination with sulphonylurea. Diabetes Obes Metab 17:23–31 dapagliflozin versus placebo in patients with type 2 diabetes on 37. Neal B, Perkovic V, de Zeeuw D et al (2015) Efficacy and safety combination antihypertensive therapy: a randomised, double- of canagliflozin, an inhibitor of sodium-glucose cotransporter 2, blind, placebo-controlled, phase 3 study. Lancet Diabetes Endo- when used in conjunction with insulin therapy in patients with crinol. (Epub ahead of print) type 2 diabetes. Diabetes Care 38:403–411 53. Mudaliar S, Henry RR, Boden G et al (2014) List J.AChanges in 38. Inagaki N, Kondo K, Yoshinari T, Takahashi N, Susuta Y, Kuki insulin sensitivity and insulin secretion with the sodium glucose H (2014) Efficacy and safety of canagliflozin monotherapy in cotransporter 2 inhibitor dapagliflozin. Diabetes Technol Ther Japanese patients with type 2 diabetes inadequately controlled 16:137–144

123 Acta Diabetol

54. https://clinicaltrials.gov/ct2/show/NCT01137474?term=NCT011 placebo-controlled study with a 28-week extension. J Am Geriatr 37474&rank=1 Soc 62:1252–1262 55. Schumm-Draeger PM, Burgess L, Kora´nyi L, Hruba V, Hamer- 69. Bolinder J, Ljunggren O¨ , Johansson L et al (2014) Dapagliflozin Maansson JE, de Bruin TW (2015) Twice-daily dapagliflozin co- maintains glycaemic control while reducing weight and body fat administered with metformin in type 2 diabetes: a 16-week ran- mass over 2 years in patients with type 2 diabetes mellitus domized, placebo-controlled clinical trial. Diabetes Obes Metab inadequately controlled on metformin. Diabetes Obes Metab 17:42–51 16:159–169 56. Henry RR, Murray AV, Marmolejo MH, Hennicken D, 70. Bailey CJ, Morales Villegas EC, Woo V, Tang W, Ptaszynska A, Ptaszynska A, List JF (2012) Dapagliflozin, metformin XR, or List JF (2015) Efficacy and safety of dapagliflozin monotherapy both: initial pharmacotherapy for type 2 diabetes, a randomised in people with Type 2 diabetes: a randomized double-blind pla- controlled trial. Int J Clin Pract 66:446–456 cebo-controlled 102-week trial. Diabet Med 32:531–541 57. Rosenstock J, Hansen L, Zee P et al (2015) Dual add-on therapy 71. Bailey CJ, Gross JL, Pieters A, Bastien A, List JF (2010) Effect in type 2 diabetes poorly controlled with metformin monother- of dapagliflozin in patients with type 2 diabetes who have inad- apy: a randomized double-blind trial of saxagliptin plus dapa- equate glycaemic control with metformin: a randomised, double- gliflozin addition versus single addition of saxagliptin or blind, placebo-controlled trial. Lancet 375:2223–2233 dapagliflozin to metformin. Diabetes Care 38:376–383 72. Kohan DE, Fioretto P, Tang W, List JF (2014) Long-term study 58. Bailey CJ, Iqbal N, T’joen C, List JF (2012) Dapagliflozin of patients with type 2 diabetes and moderate renal impairment monotherapy in drug-naı¨ve patients with diabetes: a randomized- shows that dapagliflozin reduces weight and blood pressure but controlled trial of low-dose range. Diabetes Obes Metab does not improve glycemic control. Kidney Int 85:962–971 14:951–959 73. Wilding JP, Woo V, Rohwedder K, Sugg J, Parikh S, Dapagli- 59. Jabbour SA, Hardy E, Sugg J, Parikh S, Study 10 Group (2014) flozin 006 Study Group (2014) Dapagliflozin in patients with type Dapagliflozin is effective as add-on therapy to sitagliptin with or 2 diabetes receiving high doses of insulin: efficacy and safety without metformin: a 24-week, multicenter, randomized, double- over 2 years. Diabetes Obes Metab 16:124–136 blind, placebo-controlled study. Diabetes Care 37:740–750 74. Nauck MA, Del Prato S, Dura´n-Garcı´a S et al (2014) Durability 60. Mathieu C, Ranetti AE, Li D et al (2015) Randomized, double- of glycaemic efficacy over 2 years with dapagliflozin versus blind, Phase 3 Trial of Triple Therapy with Dapagliflozin Add-on glipizide as add-on therapies in patients whose type 2 diabetes to Saxagliptin Plus Metformin in Type 2 Diabetes. Diabetes Care mellitus is inadequately controlled with metformin. Diabetes 38:2009–2017 Obes Metab 16:1111–1120 61. Kaku K, Kiyosue A, Inoue S et al (2014) Efficacy and safety of 75. Ferrannini E, Berk A, Hantel S et al (2013) Long-term safety and dapagliflozin monotherapy in Japanese patients with type 2 dia- efficacy of empagliflozin, sitagliptin, and metformin: an active- betes inadequately controlled by diet and exercise. Diabetes Obes controlled, parallel-group, randomized, 78-week open-label Metab 16:1102–1110 extension study in patients with type 2 diabetes. Diabetes Care 62. Strojek K, Yoon KH, Hruba V, Elze M, Langkilde AM, Parikh S 36:4015–4021 (2011) Effect of dapagliflozin in patients with type 2 diabetes 76. Rosenstock J, Seman LJ, Jelaska A et al (2013) Efficacy and who have inadequate glycaemic control with glimepiride: a ran- safety of empagliflozin, a sodium glucose cotransporter 2 domized, 24-week, double-blind, placebo-controlled trial. Dia- (SGLT2) inhibitor, as add-on to metformin in type 2 diabetes betes Obes Metab 13:928–938 with mild hyperglycaemia. Diabetes Obes Metab 15:1154–1160 63. Cefalu WT, Leiter LA, de Bruin TW, Gause-Nilsson I, Sugg J, 77. Kadowaki T, Haneda M, Inagaki N et al (2015) Efficacy and Parikh SJ (2015) Dapagliflozin’s effects on glycemia and car- safety of empagliflozin monotherapy for 52 weeks in Japanese diovascular risk factors in high-risk patients with type 2 diabetes: patients with type 2 diabetes: a randomized, double-blind, par- a 24-week, multicenter, randomized, double-blind, placebo-con- allel-group study. Adv Ther 32:306–318 trolled study with a 28-week extension. Diabetes Care 78. Tikkanen I, Narko K, Zeller C et al (2015) Empagliflozin reduces 38:1218–1227 blood pressure in patients with type 2 diabetes and hypertension. 64. Ji L, Ma J, Li H, Mansfield TA et al (2014) Dapagliflozin as Diabetes Care 38:420–428 monotherapy in drug-naive Asian patients with type 2 diabetes 79. https://clinicaltrials.gov/ct2/show/NCT01649297?term=NCT01 mellitus: a randomized, blinded, prospective phase III study. Clin 649297&rank=1 Ther 36:84–100 80. Ha¨ring HU, Merker L, Seewaldt-Becker E et al (2013) Empa- 65. Yang W, Han P, Min KW et al (2015) Efficacy and safety of gliflozin as add-on to metformin plus sulfonylurea in patients dapagliflozin in Asian patients with type 2 diabetes after met- with type 2 diabetes: a 24-week, randomized, double-blind, pla- formin failure: a randomized controlled trial. J Diabetes. (Epub cebo-controlled trial. Diabetes Care 36:3396–3404 ahead of print) 81. Roden M, Weng J, Eilbracht J et al (2013) Empagliflozin 66. Rosenstock J, Vico M, Wei L, Salsali A, List JF (2012) Effects of monotherapy with sitagliptin as an active comparator in dapagliflozin, an SGLT2 inhibitor, on HbA(1c), body weight, and patients with type 2 diabetes: a randomised, double-blind, hypoglycemia risk in patients with type 2 diabetes inadequately placebo-controlled, phase 3 trial. Lancet Diabetes Endocrinol controlled on pioglitazone monotherapy. Diabetes Care 1:208–219 35:1473–1478 82. Barnett AH, Mithal A, Manassie J et al (2014) Efficacy and safety 67. Matthaei S, Bowering K, Rohwedder K, Grohl A, Parikh S, Study of empagliflozin added to existing antidiabetes treatment in 05 Group (2015) Dapagliflozin improves glycemic control and patients with type 2 diabetes and chronic kidney disease: a ran- reduces body weight as add-on therapy to metformin plus sul- domised, double-blind, placebo-controlled trial. Lancet Diabetes fonylurea: a 24-week randomized, double-blind clinical trial. Endocrinol 2:369–384 Diabetes Care 38:365–372 83. Rosenstock J, Jelaska A, Frappin G et al (2014) Improved glucose 68. Leiter LA, Cefalu WT, de Bruin TW, Gause-Nilsson I, Sugg J, control with weight loss, lower insulin doses, and no increased Parikh SJ (2014) Dapagliflozin added to usual care in individuals hypoglycemia with empagliflozin added to titrated multiple daily with type 2 diabetes mellitus with preexisting cardiovascular injections of insulin in obese inadequately controlled type 2 disease: a 24-week, multicenter, randomized, double-blind, diabetes. Diabetes Care 37:1815–1823

123 Acta Diabetol

84. DeFronzo RA, Lewin A, Patel S et al (2015) Combination of transporter-2 inhibitor ertugliflozin, assessed via ambulatory empagliflozin and linagliptin as second-line therapy in subjects blood pressure monitoring in patients with type 2 diabetes and with type 2 diabetes inadequately controlled on metformin. hypertension. Diabetes Obes Metab 17:805–808 Diabetes Care 38:384–393 92. Wilding JP, Ferrannini E, Fonseca VA, Wilpshaar W, Dhanjal P, 85. Lewin A, DeFronzo RA, Patel S et al (2015) Initial combination Houzer A (2013) Efficacy and safety of ipragliflozin in patients of empagliflozin and linagliptin in subjects with type 2 diabetes. with type 2 diabetes inadequately controlled on metformin: a Diabetes Care 3:394–402 dose-finding study. Diabetes Obes Metab 15:403–409 86. Kovacs CS, Seshiah V, Swallow R et al (2014) Empagliflozin 93. Fonseca VA, Ferrannini E, Wilding JP et al (2013) Active- and improves glycaemic and weight control as add-on therapy to placebo-controlled dose-finding study to assess the efficacy, pioglitazone or pioglitazone plus metformin in patients with type safety, and tolerability of multiple doses of ipragliflozin in 2 diabetes: a 24-week, randomized, placebo-controlled trial. patients with type 2 diabetes mellitus. J Diabetes Complic Diabetes Obes Metab 16:147–158 27:268–273 87. Merker L, Ha¨ring HU, Christiansen AV et al (2015) Empagli- 94. Kashiwagi A, Kazuta K, Goto K, Yoshida S, Ueyama E, Utsuno flozin as add-on to metformin in people with Type 2 diabetes. A (2015) Ipragliflozin in combination with metformin for the Diabet Med 32:1555–1567 treatment of Japanese patients with type 2 diabetes: ILLUMI- 88. Rosenstock J, Jelaska A, Zeller C et al (2015) Impact of empa- NATE, a randomized, double-blind, placebo-controlled study. gliflozin added on to basal insulin in type 2 diabetes inadequately Diabetes Obes Metab 17:304–308 controlled on basal insulin: a 78-week randomized, double-blind, 95. Kashiwagi A, Takahashi H, Ishikawa H et al (2015) A random- placebo-controlled trial. Diabetes Obes Metab 17:936–948 ized, double-blind, placebo-controlled study on long-term effi- 89. Araki E, Tanizawa Y, Tanaka Y et al (2015) Long-term treatment cacy and safety of ipragliflozin treatment in patients with type 2 with empagliflozin as add-on to oral antidiabetes therapy in diabetes mellitus and renal impairment: results of the long-term Japanese patients with type 2 diabetes mellitus. Diabetes Obes ASP1941 safety evaluation in patients with type 2 diabetes with Metab 17:665–674 renal impairment (LANTERN) study. Diabetes Obes Metab 90. Ridderstra˚le M, Andersen KR et al (2014) Comparison of 17:152–160 empagliflozin and glimepiride as add-on to metformin in patients 96. Kashiwagi A, Kazuta K, Yoshida S, Nagase I (2014) Random- with type 2 diabetes: a 104-week randomised, active-controlled, ized, placebo-controlled, double-blind glycemic control trial of double-blind, phase 3 trial. Lancet Diabetes Endocrinol novel sodium-dependent glucose cotransporter 2 inhibitor ipra- 2:691–700 gliflozin in Japanese patients with type 2 diabetes mellitus. J Di- 91. Amin NB, Wang X, Mitchell JR, Lee DS, Nucci G, Rusnak JM abetes Investig 5:382–391 (2015) Blood pressure-lowering effect of the sodium glucose co-

123