Efficacy and safety of sodium glucose co-transport-2 inhibitors in type 2 diabetes: a meta-analysis of randomized clinical trials

Transcription

1 Diabetes, Obesity and Metabolism John Wiley & Sons Ltd Efficacy and safety of sodium glucose co-transport-2 inhibitors in type 2 diabetes: a meta-analysis of randomized clinical trials M. Monami 1, C. Nardini 1 & E. Mannucci 2 1 Geriatric Cardiology, Careggi Teaching Hospital, Florence, Italy 2 Diabetes Agency, Careggi Teaching Hospital, Florence, Italy original article Aims: Sodium glucose co-transport-2 (SGLT-2) inhibitors, a new class of glucose-lowering agents, reduce tubular glucose reabsorption, producing a reduction of blood glucose without stimulating insulin release. The aim of the present meta-analysis is the assessment of the overall efficacy and safety profile of these drugs. Methods: A meta-analysis was performed including all trials with a duration of at least 12 weeks, comparing a SGLT-2 inhibitor with a non-sglt-2 inhibitor agent in type 2 diabetes. The principal outcome of this analysis was the effect of SGLT-2 inhibitors on HbA1c at 12, 24 and 52 weeks. Hypoglycaemia, genital and urinary infections were retrieved and combined to calculate Mantel Haenszel odds ratio (MH-OR). Furthermore, data on body mass index (BMI), endpoint fasting plasma glucose, systolic and diastolic blood pressure, creatinine, hematocrit and lipid profile were collected. Results: Among placebo-controlled trials, HbA1c reduction at 12, 24 and 52 weeks was 0.5 [0.4; 0.6], 0.6 [0.6; 0.5] and 0.6 [0.7; 0.5]%. In placebo-controlled studies, 24-week reduction of HbA1c with SGLT-2 inhibitors was greater in trials enrolling patients with a lower mean age and duration of diabetes, and a higher baseline BMI, HbA1c and fasting glucose. In placebo-controlled trials, SGLT-2 inhibitors determined a weight loss during the first 24 weeks, which was maintained up to 52 weeks. Conclusions: SGLT-2 inhibitors are effective in the treatment of type 2 diabetes, providing additional benefits, such as weight loss, reduction of blood pressure and increase in high-density lipoprotein (HDL)-cholesterol. Apart from genital and urinary infections, rather frequent but usually mild, SGLT-2 inhibitors appear to be well tolerated. Keywords: meta-analysis, SGLT-2 inhibitor Date submitted 21 July 2013; date of first decision 24 September 2013; date of final acceptance 2 December 2013 Introduction Type 2 diabetes is a complex condition, which needs multiple drug treatments in many cases [1,2]. In addition, due to the clinical and pathophysiological heterogeneity of type 2 diabetes, different combinations of drugs are required in individual patients [2]. The availability of many classes of agents, acting through different mechanisms, can be of help in achieving optimal glycaemic control in a larger proportion of patients. As a consequence, pharmacological research is aiming at the development of further classes of therapeutic agents, which can be combined with some or all existing drugs. The last class which reached approval by regulatory authorities is that of sodium glucose co-transport-2 (SGLT-2) inhibitors, which reduce tubular glucose reabsorption, producing a reduction of blood glucose without stimulating insulin release [3]. Several SGLT-2 inhibitors have been developed so far. The aim of the present meta-analysis is the assessment of the overall Correspondence to: Edoardo Mannucci, MD, Diabetes Agency, Azienda Ospedaliero-Universitaria Careggi, Via delle Oblate 4, Florence, Italy. edoardo.mannucci@unifi.it efficacy and safety profile of these drugs, exploring potential differences across individual agents. Materials and Methods This meta-analysis is reported following the criteria of PRISMA statement [4]. Data Sources and Searches An extensive Medline, Embase and Cochrane database search for Sodium Glucose co-transport-2 inhibitor or SGLT-2 inhibitor or Canagliflozin or Dapagliflozin or Empagliflozin or Ipragliflozin or Remogliflozin or Sergliflozin or Tofogliflozin or ASP1941 or AVE2268 or BI or BMS or KGT-1681 or TA-7284 or TS-033 or YM543 was performed, collecting all randomized clinical trials on humans up to 21 May 2013, with no other filter. Furthermore, completed but yet unpublished studies with the drugs specified above were searched in the register. The identification of relevant abstracts, the selection of studies based on the criteria described

2 above, and the subsequent data extraction were performed independently by the two of the authors, and conflicts resolved by mutual discussion. Study Selection A meta-analysis was performed including all randomized clinical trials with a duration of at least 12 weeks, enrolling patients with type 2 diabetes, comparing SGLT-2 inhibitors with placebo or active drugs (oral hypoglycaemic agents and/or insulin) different from other SGLT-2 inhibitors. Trials enrolling non-diabetic, or type 1 diabetic, subjects were excluded. No review protocol was published elsewhere. Data Extraction and Quality Assessment Results of unpublished trials were retrieved, if available, on or as well as Food and Drug Administration (FDA, fuseaction=search.search_drug_name) and European Medicines Agency (EMEA, reviews of approved drugs. All those sources were also used to complete information on results of published trials, when not reported in publications (including the primary trial publications, and subsequent reviews and/or pooled analyses reporting data on individual trials). For all published trials, results reported in papers were used as the primary source of information, when available. When available in the primary or subsequent publications, data with adjudication of events were preferred. DIABETES, OBESITY AND METABOLISM The quality of trials was assessed using some of the parameters proposed by Jadad et al. [5]. The score was not used as a criterion for the selection of trials, whereas some items were used only for descriptive purposes. Data Synthesis and Analysis The principal outcome of this analysis was the effect of SGLT-2 inhibitors, compared either with placebo or active drugs, on HbA1c at 12, 24 and 52 weeks. Secondary outcomes included effects on body mass index (BMI), fasting plasma glucose (FPG), total, low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol, triglyceride and systolic and diastolic blood pressure, as well as the incidence of any or severe hypoglycaemia, urinary tract infections and genital infections. Data on the incidence of serious adverse events were also collected; among those, major cardiovascular events (MACE including cardiovascular death, non-fatal myocardial infarction and stroke and acute coronary syndromes and/or heart failure), malignancies, renal failure, bone fractures and all-cause mortality were considered separately. Subgroup analyses were performed for different classes of comparators and individual SGLT-2 inhibitors, whenever appropriate. Meta-regressions were performed for the exploration of possible moderators of efficacy, correlating placebo-subtracted 24-week HbA1c variation with age, duration of diabetes and baseline HbA1c, FPG and BMI. Heterogeneity (on 12-week HbA1c) was assessed by using I 2 statistics. To estimate possible publication/disclosure bias, we used funnel plots and the Begg adjusted rank correlation Figure 1. Trial flow summary. 2 Monami et al. 2013

3 DIABETES, OBESITY AND METABOLISM original article Study name Statistics for each study and 95% CI Stenlof (22) 0,900 0,088 0,008 1,073 0,727 10,221 0,000 Kaku (30) 0,700 0,106 0,011 0,908 0,492 6,599 0,000 Wilding (33) 0,700 0,186 0,035 1,065 0,335 3,764 0,000 Bode (21) 0,700 0,064 0,004 0,825 0,575 11,010 0,000 Strojek (32) 0,600 0,074 0,006 0,746 0,454 8,082 0,000 Bailey (25) 0,600 0,181 0,033 0,955 0,245 3,310 0,001 Bailey (26) 0,500 0,097 0,009 0,690 0,310 5,150 0,000 Yale (23) 0,400 0,116 0,014 0,628 0,172 3,439 0,001 Wilding (33) 0,400 0,066 0,004 0,528 0,272 6,101 0,000 Ferrannini (38) 0,400 0,124 0,015 0,643 0,157 3,232 0,001 Rosenstock (31) 0,400 0,104 0,011 0,603 0,197 3,857 0,000 Wilding (34) 0,300 0,093 0,009 0,483 0,117 3,210 0,001 Henry (36) 0,300 0,920 0,847 2,104 1,504 0,326 0,744 Fonseca (39) 0,300 0,077 0,006 0,450 0,150 3,908 0,000 Henry (36) 0,300 0,766 0,587 1,801 1,201 0,392 0,695 List (35) 0,200 0,101 0,010 0,399 0,001 1,974 0,048 Ferrannini (28) 0,200 0,081 0,007 0,359 0,041 2,467 0,014 Rosenstock (9) 0,100 0,091 0,008 0,278 0,078 1,103 0,270 Heerspink (29) 0,100 0,237 0,056 0,564 0,364 0,422 0,673 OVERALL 0,455 0,023 0,001 0,499 0,410 20,089 0,000 2,00 1,00 0,00 1,00 2,00 Favours SGLT-2i Favours Comparators HbA1c at 12 weeks Study name Statistics for each study and 95% CI Stenlof (22) 1,000 0,088 0,008 1,173 0,827 11,357 0,000 Strojek (32) 0,700 0,067 0,004 0,831 0,569 10,463 0,000 Bode (21) 0,700 0,064 0,004 0,825 0,575 11,010 0,000 Bailey (25) 0,600 0,181 0,033 0,955 0,245 3,310 0,001 Bailey (26) 0,500 0,097 0,009 0,690 0,310 5,150 0,000 Wilding (33) 0,500 0,066 0,004 0,628 0,372 7,627 0,000 Yale (23) 0,500 0,116 0,014 0,728 0,272 4,299 0,000 Rosenstock (31) 0,400 0,104 0,011 0,603 0,197 3,857 0,000 Ferrannini (28) 0,400 0,124 0,015 0,643 0,157 3,232 0,001 Bolinder (27) 0,300 0,067 0,005 0,432 0,168 4,469 0,000 Henry (36) 0,250 0,119 0,014 0,483 0,017 2,107 0,035 Henry (36) 0,250 0,102 0,010 0,450 0,050 2,455 0,014 OVERALL 0,512 0,066 0,004 0,641 0,384 7,805 0,000 2,00 1,00 0,00 1,00 2,00 Favours SGLT-2i Favours Comparators HbA1c at weeks Study name Statistics for each study and 95% CI Bailey (25) 0,700 0,091 0,008 0,879 0,521 7,658 0,000 Rosenstock (31) 0,600 0,104 0,011 0,803 0,397 5,786 0,000 Wilding (34) 0,500 0,066 0,004 0,628 0,372 7,627 0,000 OVERALL 0,575 0,047 0,002 0,668 0,482 12,127 0,000 2,00 1,00 0,00 1,00 2,00 Favours SGLT-2i Favours Comparators HbA1c at weeks Figure 2. with 95% confidence interval for 12-, 24- and 52-week HbA1c for each individual trial included in the meta-analysis. test [6,7]. However, as these tests have low statistical power when the number of trials is small, undetected bias may still be present. and Mantel Haenszel odds ratio (MH-OR) was calculated with 95% confidence interval, for continuous and categorical variables, respectively, on an intention-to-treat basis. For MH-OR, trials with zero events were excluded. All analyses were performed using Comprehensive Meta-analysis Version 2, Biostat (Englewood, NJ, USA). Results Retrieved Trials The trial flow summary is reported in Figure 1. A total of 25 trials fulfilling the inclusion criteria were identified. Of those, 21, 14 and 4 reported information on 12-, 24- and 52-week HbA1c, respectively (Figure 2). The characteristics of the retrieved trials (including parameters of trials quality) and the recorded outcomes are reported in Tables 1 3. Retrieved trials included 7524 and 3628 patients in SGLT-2 inhibitors and comparator groups, respectively; and the mean trial duration was 30 weeks. The mean age, duration of diabetes, baseline HbA1c and BMI of enrolled patients were 56.3 years, 6.1 years, 8.1% and 31.6 kg/m 2, respectively. Efficacy on HbA1c In the 21 trials reporting 12-week HbA1c, I 2 was 91.5 (p < 0.001). Funnel plot (Figure 3) and Begg adjusted rank correlation test (Kendall s τ : 0.1; p = 0.67) suggested no major publication bias. In placebo-controlled trials, SGLT- 2 inhibitors produced a significant reduction of HbA1c at all the time points considered. Among placebo-controlled 2013 doi: /dom

4 DIABETES, OBESITY AND METABOLISM Table 1. Baseline characteristics and quality assessment of trials included in the meta-analysis. First author Description of Trial duration (weeks) Randomization Allocation Blinding Dropouts ITT Mean Age (years) Diabetes duration (years) HbA1c (%) Bode [20] 26 A A A A Yes Stenlöf [21] 26 A NA A A Yes Yale [22] 24 A A NA A Yes NCT [23] 52 NA A A A Yes 68 NR NR NR NCT [23] 52 NA A A A Yes 57 NR NR NR Rosenstock [9] 12 NA NA NA A Yes Schernthaner [8] 52 A A A A Yes NCT [23] 52 NA A A A Yes 57 NR NR NR Bailey [24] 24 A A NA A Yes Bailey [25] 102 A A A A Yes Bolinder [26] 24 A A A A Yes Ferrannini [27] 24 NA NA NA A Yes Lambers Heerspink [28] 12 A A A A Yes NR Kaku [29] 12 A A A A Yes NR Rosenstock [30] 48 A NA NA A Yes NR Strojek [31] 26 A A A A Yes NR Wilding [32] 26 A NA NA A Yes Wilding [33] 48 A A A A Yes List [16] 12 NA NA NA A Yes 55 NR Henry [15] 24 A A A A Yes NR Henry [15] 24 A A A A Yes NR Nauck [10] 52 A A A A Yes Ferrannini [13] 12 A A OL A Yes 58 NR Fonseca [14] 12 A NA NA A Yes Wilding [34] 12 NA NA NA NA Yes A, adequately described; BMI, body mass index; ITT, intention-to-treat; NA, not adequately described; NR, not reported; OL, open label. BMI (kg/m 2 ) trials, HbA1c reduction at 12 weeks was 0.5 [0.4; 0.6]% (p < 0.001, N = 12 trials) with dapagliflozin, 0.7 [0.6; 0.8]% with canagliflozin (p < 0.001, N = 4) and 0.4 [0.3; 0.6]% (p < 0.001, N = 2) with ipragliflozin. The corresponding figures at 24 weeks were 0.6 [0.6; 0.5]% (p < 0.001, N = 12 trials) among placebo-controlled trials, and 0.6 [0.5; 0.7]% (p < 0.001, N = 9 trials) and 0.8 [0.7; 0.8]% (p < 0.001, N = 3) with dapagliflozin and canagliflozin, respectively. HbA1c reduction at 52 weeks was 0.6 [0.7; 0.5]% among placebo-controlled trials (p < 0.001; N = 3 trials). Results of active comparator trials on 12- and 24-week HbA1c are summarized in Table 4. SGLT-2 inhibitors had a similar efficacy as that of metformin in direct comparisons. In the two trials [8,9] comparing SGLT-2 inhibitors with sitagliptin, 12-week HbA1c was similar between the two treatments; in the only trial reporting results on HbA1c at 24 weeks [8], canagliflozin was superior to sitagliptin and this difference was maintained at 52 weeks [8]. There was only one available trial [10] comparing a SGLT-2 inhibitor (dapagliflozin) with a sulphonylurea (glipizide), showing a superiority of the latter on HbA1c both at 12 and 24 weeks, but not at 52 weeks (0.0 [ 0.1; 0.1]%). In placebo-controlled studies, 24-week reduction of HbA1c with SGLT-2 inhibitors was greater in trials enrolling patients with a lower mean age and duration of diabetes (Slope: 0.06 [0.05; 0.07] and 0.05 [0.04; 0.06], respectively, all p < 0.001), and a higher baseline BMI, HbA1c and fasting glucose ( 0.28 [ 0.37; 0.19], 0.37 [ 0.46; 0.27] and 0.34 [ 0.42; 0.27], respectively, all p < 0.001). Effects on Other Metabolic Parameters In placebo-controlled trials, SGLT-2 inhibitors determined a weight loss during the first 24 weeks, which was maintained up to 52 weeks (Figure 4). In the only available trial versus glipizide [10], dapagliflozin has a more favourable effect on BMI ( 0.6 [+0.1; 1.3], p = 0.11; 1.1 [ 0.4; 1.8] and 1.1 [ 0.4; 1.8] kg/m 2,p= at 12, 24 and 52 weeks, respectively). In a head-to-head comparison with sitagliptin [8], canagliflozin was associated with a significantly lower BMI at 12 weeks ( 1.0 [ 2.0; 0.0] kg/m 2,p= 0.049), 24 weeks ( 1.2 [ 2.2; 0.2] kg/m 2,p= 0.02) and 52 weeks ( 1.0 [ 2.0; 0.0] kg/m 2,p= 0.42). The other metabolic parameters and cardiovascular risk factors are summarized in Table 5. SGLT-2 inhibitors reduced FPG not only in placebo-controlled trials, but also in direct comparisons with metformin and sitagliptin ( 0.5 [ 0.9; 0.1] and 0.8 [ 1.3; 0.2] mmol/l, respectively). Only some of the trials reported results for lipid profile. On the basis of this information, SGLT-2 inhibitors determined a modest but statistically significant increase in HDL cholesterol, with no effect on triglyceride and on total and LDL cholesterol. Endpoint systolic and diastolic blood pressure was significantly reduced by SGLT-2 inhibitors in comparison with placebo. 4 Monami et al. 2013

5 DIABETES, OBESITY AND METABOLISM original article Table 2. Endpoint HbA1c and BMI for each trial included in the meta-analysis. First author SGLT-2i Patients (n) Comparator Patients (n) Endpoint HbA1c (%) 12 wks (ID/C) 24 wks (ID/C) 52 wks (ID/C) Endpoint body mass index (kg/m 2 ) 12 wks (ID/C) 24 wks (ID/C, #) 52 wks (ID/C, #) Bode [20] Canagliflozin 477 Placebo / /7.8 NR/NR 30./ /1 1/0 Stenlöf [21] Canagliflozin 393 Placebo / /8.1 NR/NR 30.4/31.7 0/0 0/0 Yale [22] Canagliflozin 179 Placebo / /8.0 NR/NR 31.9/33.1 0/1 0/1 NCT [23] Canagliflozin 179 Placebo 90 NR/ NR NR/NR NR/NR NR/NR 0/0 0/0 NCT [23] Canagliflozin 313 Placebo 156 NR/ NR NR/NR NR/NR NR/NR 0/0 0/0 Rosenstock [9] Canagliflozin 321 Placebo /7.5 NR/NR NR/NR 30.6//30.2 4/4 2/1 Canagliflozin 321 Sitagliptin /6.9 NR/NR NR/NR 30.6/31.4 NR/NR NR/NR Schernthaner [8] Canagliflozin 377 Sitagliptin / / / /31.7 0/0 0/0 NCT [23] Canagliflozin 227 Sitagliptin 115 NR/ NR NR/NR NR/NR NR/NR 0/0 0/0 Bailey [24] Dapagliflozin 214 Placebo / /7.8 NR/NR 30.9/32.1 0/0 0/0 Bailey [25] Dapagliflozin 409 Placebo / / / /32.1 NR/NR NR/NR Bolinder [26] Dapagliflozin 91 Placebo 91 NR/ NR 6.8/7.1 NR/NR 31.5/31.6 4/2 NR/NR Ferrannini [27] Dapagliflozin 410 Placebo / /7.6 NR/NR 32.0/31.6 0/0 0/0 Lambers Heerspink [28] Dapagliflozin 24 Placebo /7.1 NR/NR NR/NR NR/NR 0/1 0/1 Kaku [29] Dapagliflozin 225 Placebo /8.5 NR/NR NR/NR NR/NR 0/0 0/0 Rosenstock [30] Dapagliflozin 281 Placebo / / /7.9 NR/NR 0/0 0/0 Strojek [31] Dapagliflozin 447 Placebo / /8.0 NR/NR NR/NR 2/1 2/1 Wilding [32] Dapagliflozin 48 Placebo /8.5 NR/NR NR/NR 34.2/34.1 NR/NR NR/NR Wilding [33] Dapagliflozin 610 Placebo / / / /33.2 0/0 0/0 List [16] Dapagliflozin 279 Placebo /7.7 NR//NR NR/NR 30.7/31.6 0/0 0/0 Dapagliflozin 279 Metformin /6.9 NR/NR NR/NR 30.7/31.5 0/0 0/0 Henry [15]* Dapagliflozin 211 Placebo / /7.6 NR/NR NR/NR NR/NR NR/NR Dapagliflozin 219 Metformin / /7.6 NR/NR NR/NR 0/1 0/1 Henry [15]* Dapagliflozin 194 Placebo / /7.8 NR/NR NR/NR 0/1 0/1 Dapagliflozin 203 Metformin / /7.8 NR/NR NR/NR 0/0 0/0 Nauck [10] Dapagliflozin 400 Glipizide / / / /31.6 0/0 0/0 Ferrannini [13] Empagliflozin 244 Placebo /7.9 NR/NR NR/NR 27.6/28.6 1/0 NR Empagliflozin 244 Metformin /7.4 NR/NR NR/NR 27.6/28.1 0/0 0/0 Fonseca [14] Ipragliflozin 273 Placebo /8.1 NR/NR NR/NR 30.6/30.6 0/0 0/0 Ipragliflozin 273 Metformin /7.6 NR/NR NR/NR 30.6/29.6 0/0 0/0 Wilding [34] Ipragliflozin 276 Placebo /7.4 NR/NR NR/NR 30.9/31.8 NR/NR NR/NR BMI, body mass index; C, comparator; ID, interventional drug; NR, not reported; SGLT-2i, sodium glucose co-transport-2 inhibitors; wks, weeks. *This article reported the results of two different trials. In addition, SGLT-1 inhibitors were associated with a significant increase in hematocrit, with no significant effect on serum creatinine (Table 5). Serious Adverse Events The overall incidence of serious adverse events was not significantly different with SGLT-2 inhibitors, both in placebocontrolled or active comparator-controlled trials (Figure 5, panel A). In the 12 trials with at least one death, the effect of SGLT-2 inhibitors on mortality was (MH-OR) 0.84 [ ] (p = 0.69). The MH-OR for MACE was 0.67 [ ] (p = 0.30; N = 7 trials). The corresponding figures for malignancies, renal failure and bone fractures were 1.42 [ ] (p = 0.60, N = 5), 1.37 [ ] (p = 0.67, N = 4) and 0.84 [ ] (p = 0.72; N = 7). Other Adverse Events of Interest Treatment with SGLT-2 inhibitors was associated with a similar hypoglycaemic risk as that of metformin and DPP-4 inhibitors, whereas a significantly lower risk was found in the only one available comparison with a sulphonylurea [10]. In placebo-controlled trials, SGLT-2 inhibitors were associated with a significant increase in the incidence of hypoglycaemia (Figure 5, panel B); however, that risk was found in trials in which SGLT-2 inhibitors were combined with sulphonylureas and/or insulin (N = 7, MH-OR 1.41 [ ], p = 0.005), but not in studies in which they were used as monotherapy or in combination with other agents (N = 13, MH-OR 1.13 [ ], p = 0.55). Episodes of severe hypoglycaemia were observed only in five placebo-controlled trials, in four of which SGLT-2 inhibitors had been used as add-on to insulin and/or sulphonylureas; the overall MH-OR was 1.26 [ ] (p = 0.67). Information on urinary tract and genital infections was available for all trials. The incidence of urinary and genital infections (Figure 5, panels C and D) was significantly increased with SGLT-2 inhibitors (1.23 [ ], p = 0.02 and 3.90 [ ], p < 0.001). When trials with different comparators were analysed separately, the increase in the 2013 doi: /dom

6 DIABETES, OBESITY AND METABOLISM Table 3. Other metabolic outcomes for each trial included in the meta-analysis. First author* Total cholesterol (ID/C, mmol/l) HDL cholesterol (ID/C, mmol/l) LDL cholesterol (ID/C, mmol/l) Triglycerides (ID/C, mmol/l) FPG (ID/C, mmol/l) SBP (ID/C, mmhg) DBP (ID/C, mmhg) Creatinine (ID/C, μmol/l) Bode [20] 4.6/ / / / / /132 73/ /85.0 Stenlöf [21] 5.1/ / / / / /128 76/ /75.0 Yale [22] 4.4/ / / / / /132 71/72 NR/NR NCT [23] NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NCT [23] NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR Rosenstock [9] 4.9/NR 1.2/NR 2.8/NR 1.8/NR 7.7/ /NR 76/NR NR/NR 4.9/ / / / / /128 76/79 NR/NR Schernthaner [8] NR/NR NR/NR NR/NR NR/NR 7.0/8.0 NR/NR NR/NR NR/NR NCT [23] NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR Bailey [24] 5.0/ / / / / /130 77/ /75.5 Bailey [25] 4.9/ / / / / /129 79/ /77.0 Bolinder [26] NR/NR NR/NR NR/NR NR/NR 7.3/ /133 80/ /74.6 Ferrannini [27] NR/NR NR/NR NR/NR NR/NR 7.7/8.6 NR/NR NR/NR NR/NR Lambers Heerspink [28] NR/NR NR/NR NR/NR NR/NR 7.4/ /128 77/80 NR/NR Kaku [29] 5.5/ / / /1.7 NR/NR NR/NR NR/NR NR/NR Rosenstock [30] NR/NR NR/NR NR/NR NR/NR 7.7/8.2 NR/NR NR/NR NR/NR Strojek [31] 5.0/ / / / / /132 77/78 NR/NR Wilding [32] NR/NR NR/NR NR/NR NR/NR 8.6/10.1 NR/NR NR/NR NR/NR Wilding [33] NR/NR NR/NR NR/NR NR/NR NR/NR 134/134 77/ /80.5 List [16] NR/NR NR/NR NR/NR NR/NR 6.0/ /128 77/ /74.0 NR/NR NR/NR NR/NR NR/NR 6.0/ /126 77/ /74.0 Henry [15] NR/NR NR/NR NR/NR NR/NR 7.3/ /129 78/ /74.1 NR/NR NR/NR NR/NR NR/NR 8.2/ /129 78/ /74.1 Henry [15] NR/NR NR/NR NR/NR NR/NR 7.3/ /126 77/ /71.2 NR/NR NR/NR NR/NR NR/NR 8.3/ /126 77/ /71.2 Nauck [10] 4.8/ / / /1.9 NR/NR 129/135 79/80 74/76 Ferrannini [13] NR/NR NR/NR 2.3/ / /9.0 NR/NR NR/NR NR/NR NR/NR NR/NR 2.3/ / /8.1 NR/NR NR/NR NR/NR Fonseca [14] NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR NR/NR Wilding [34] NR/NR NR/NR NR/NR NR/NR 7.6/8.5 NR/NR NR/NR NR/NR bpm, beat per minute; DBP, diastolic blood pressure; Delta HR, differences between endpoint and baseline heart rate; FGP, fasting plasma glucose; HDL, high-density lipoprotein; ID/C, interventional drug/comparator; LDL, low-density lipoprotein; NR, not reported; SBP, systolic blood pressure. Figure 3. Funnel plot for HbA1c at 12 weeks. 6 Monami et al. 2013

7 DIABETES, OBESITY AND METABOLISM original article Table 4. s in HbA1c between sodium glucose cotransport-2 inhibitors and active comparators at 12 and 24 weeks in trials included in the meta-analysis. Sodium glucose co-transport-2 inhibitorsversus HbA1c (%) Glipizide Metformin Sitagliptin N arms (1) (5) (2) 12 weeks 0.3 [0.2; 0.4]* 0.0 [ 0.1; 0.1] 0.0 [ 0.3; 0.3] N arms (1) (5) (1) 24 weeks 0.3 [0.2; 0.5]* 0.1 [ 0.1; 0.2] 0.2 [ 0.4; 0.1]* *p < 0.05; **p < incidence of urinary tract infections did not quite reach statistical significance for any of the groups. On the other hand, the risk of genital infections was increased both in placebo- and active comparator-controlled trials. Discussion SGLT-2 inhibitors produce a significant reduction of HbA1c in comparison with placebo. This result, which is not surprising for drugs developed for the treatment of type 2 diabetes, confirms findings of previous meta-analyses performed on a smaller number of trials [11,12]. In fact, the amount of available information from phase II and phase III studies on SGLT-2 inhibitors is rapidly growing, as new molecules beside dapagliflozin approach the final phase of their registrative process. Given the larger number of trials with disclosed results, it is possible to estimate the therapeutic efficacy of SGLT-2 inhibitors at different time points, showing that the reduction in HbA1c in comparison with placebo reaches its maximum at approximately 6 months and is maintained up to 1 year. In placebo-controlled trials, canagliflozin appears to produce a slightly greater reduction of HbA1c than dapagliflozin, although the difference is small and its clinical relevance is questionable. This result needs to be considered with great caution, because the difference could be determined by diversities in case mix, entry criteria or other characteristics of study protocol, rather than by real differences in the pharmacodynamics action of the two molecules. The available number of active comparator-controlled trials remains very small; in most instances, those are studies Study name Statistics for each study and 95% CI Bailey (25) 1,540 0,508 0,259 2,537 0,543 3,029 0,002 Stenlof (22) 1,280 0,552 0,305 2,362 0,198 2,319 0,020 Bailey (26) 1,200 0,757 0,574 2,684 0,284 1,584 0,113 Bode (21) 1,180 0,366 0,134 1,897 0,463 3,227 0,001 Yale (23) 1,170 0,797 0,636 2,733 0,393 1,467 0,142 Wilding (40) 0,940 0,680 0,462 2,273 0,393 1,382 0,167 List (35) 0,920 0,547 0,299 1,992 0,152 1,682 0,092 Wilding (34) 0,460 0,447 0,200 1,336 0,416 1,030 0,303 Bolinder (27) 0,160 0,578 0,334 1,293 0,973 0,277 0,782 Fonseca (39) 0,050 0,603 0,363 1,231 1,131 0,083 0,934 Wilding (33) 0,080 1,065 1,135 2,008 2,168 0,075 0,940 Rosenstock (9) 0,360 0,508 0,258 0,636 1,356 0,708 0,479 Ferrannini (28) 0,420 0,680 0,463 0,913 1,753 0,618 0,537 OVERALL 0,695 0,156 0,024 1,000 0,389 4,456 0,000 4,00 2,00 0,00 2,00 4,00 Favours SGLT-2i Favours Comparator BMI at 12 weeks Study name Statistics for each study and 95% CI Bailey (26) 1,900 0,508 0,259 2,897 0,903 3,737 0,000 Bode (21) 1,430 0,366 0,134 2,147 0,713 3,910 0,000 Stenlof (22) 1,390 0,552 0,305 2,472 0,308 2,519 0,012 Bailey (25) 1,380 0,757 0,574 2,864 0,104 1,822 0,068 Yale (23) 1,280 0,797 0,636 2,843 0,283 1,605 0,108 Wilding (34) 0,560 0,447 0,200 1,436 0,316 1,253 0,210 Bolinder (27) 0,340 0,578 0,334 1,473 0,793 0,588 0,556 Ferrannini (28) 0,040 0,680 0,463 1,293 1,373 0,059 0,953 OVERALL 1,094 0,189 0,036 1,464 0,725 5,802 0,000 BMI at weeks 4,00 2,00 0,00 2,00 4,00 Favours SGLT-2i Favours Comparator Study name Statistics for each study and 95% CI Bailey (26) 1,300 0,508 0,259 2,297 0,303 2,557 0,011 Wilding (34) 0,740 0,447 0,200 1,616 0,136 1,656 0,098 OVERALL 0,984 0,336 0,113 1,642 0,326 2,932 0,003 BMI at weeks 4,00 2,00 0,00 2,00 4,00 Favours SGLT-2i Favours Comparator Figure 4. with 95% confidence interval for 12-, 24- and 52-week HbA1c for each individual trial included in the meta-analysis doi: /dom

8 Table 5. s in fasting plasma glucose, lipid profile, creatinine, hematocrit and blood pressure at the endpoint, between sodium glucose co-transport-2 and active comparators/placebo in trials included in the meta-analysis. Sodium glucose co-transport-2 inhibitors versus s All comparators Only placebo Fasting plasma glucose 1.0 [ 0.9; 1.2]** 1.2 [ 1.0; 1.4]** (mmol/l) Total cholesterol (mmol/l) 0.1 [ 0.1; 0.2] 0.0 [ 0.1; 0.1] HDL cholesterol (mmol/l) 0.1 [0.0;0.1]* 0.0 [ 0.1; 0.1] LDL cholesterol (mmol/l) 0.1 [ 0.1; 0.2] 0.0 [ 0.1; 0.1] Triglycerides (mmol/l) 0.1 [ 0.3; 0.0] 0.1 [ 0.3; 0.1] Creatinine (μmol/l) 0.7 [ 1.7; 0.3] Hematocrit (%) 1.4 [0.2; 2.7]* Systolic blood pressure 1.2 [ 1.4; 1.0]** (mmhg) Diastolic blood pressure (mmhg) 1.9 [ 2.6; 1.1]** HDL, high-density lipoprotein; LDL, low-density lipoprotein. *p < 0.05; **p < exploring the effect of early combination therapy with SGLT- 2 inhibitors and metformin versus the two monotherapies [13 16], which can provide useful information for the comparison of the new drugs with metformin. Despite an efficacy very similar to that of metformin in monotherapy, the first SGLT-2 inhibitor (dapagliflozin) has been approved only for use in combination with other agents, with monotherapy allowed only in metformin-intolerant subjects. Two trials [8,17] compared a SGLT-2 inhibitor (canagliflozin) with a DPP4 inhibitor (sitagliptin). Combining the two studies, the effects of canagliflozin on 12-week HbA1c were not different from those of sitagliptin. However, in the only available trial with a longer follow-up, the SGLT-2 inhibitor was superior to the DIABETES, OBESITY AND METABOLISM DPP4 inhibitor at 24 and 52 weeks [8]. Considering that the difference in HbA1c between the two groups was small, further head-to-head comparisons are needed before the superiority of SGLT-2 inhibitors can be firmly established. There is also one trial comparing dapagliflozin with glipizide, both combined with metformin [10]; the two drugs have a similar effect on endpoint HbA1c, although the sulphonylurea shows a more rapid onset of action. Interestingly, SGLT-2 inhibitors appeared to have a greater efficacy on FPG in comparison with both metformin and sitagliptin. Unfortunately, information on the effects of these drugs on post-prandial glucose peaks, which were not among the predefined endpoints of this analysis, is still scarce. The favourable effect of SGLT-2 inhibitors on hyperglycaemia is not associated with an increased hypoglycaemic risk, unless the drugs are associated with sulphonylureas or insulin. Meta-regression analyses should always be considered with caution, because of the possibility of ecological fallacy [18]. Despite this ation, the analysis can be used as hypothesis generating. The direct correlation between placebo-subtracted HbA1c reduction and baseline HbA1c is quite obvious. In older patients, the smaller effect of SGLT-2 inhibitors could be the consequence of a longer duration of diabetes, or of impaired renal function; in fact, patients with renal insufficiency could have a lower response to SGLT-2 inhibitors in clinical trials [19]. Interestingly, the efficacy of SGLT-2 inhibitors appeared to increase as a function of baseline BMI. It is possible that weight loss induced by SGLT-2 inhibitors plays a role in the attainment and maintenance of glycaemic control in obese patients with type 2 diabetes. One of the interesting features of SGLT-2 inhibitors is their effect on body weight, which, in placebo-controlled studies, reaches a nadir after 6 months from the initiation of therapy; weight loss is then maintained up to 1 year. A significantly lower endpoint BMI was also observed in direct comparisons with glipizide [10] and sitagliptin [8]. PANEL A B C D , Placebo Metformin Glipizide Sitagliptin Serious adverse events Total hypoglycaemia Urinary infections Genital infections Trials MH-OR, 95% CI p Placebo 22 0,81[0.64;1.01] Metformin 5 0,97[0.50;1.89] 0.93 Glipizide [0.47;1.18] 0.20 Sitagliptin [0.63;1.81] 0.81 N trials MH-OR, 95% CI p N trials MH-OR, 95% CI p N trials MH-OR, 95% CI p 20 1,34[1.09;1.65] < [0.99;1.52] [2.40;4.59] < ,73[0.29;1.84] [0.83;2.07] [1.98;6.82] < [0.03;0.09] < [1.07;2.96] [2.60;9.88] < [0.42;1.30] [0.50;1.39] [3.01;12.11] <0.001 Figure 5. Mantel Haenzel odds ratio with 95% confidence interval (MH-OR, 95% CI) for serious adverse events (panel A), total hypoglycaemia (panel B), urinary infections (panel C) and genital infections (panel D) in placebo- and active comparator-controlled trials. 8 Monami et al. 2013

9 DIABETES, OBESITY AND METABOLISM The previously described beneficial effect of SGLT-2 inhibitors on blood pressure [12] was confirmed by the present meta-analysis. In addition, we also observed a small increase in HDL cholesterol in placebo-controlled trials. These two factors could contribute to cardiovascular protection, although the number of major cardiovascular events reported in clinical trials is far too small for now to draw any conclusion. Similarly, any consideration on incidence of cancer, bone fractures or renal failure would be totally speculative at this stage. SGLT- 2 inhibitors have been reported to determine a small but detectable decrease in glomerular filtration [19]; however, in clinical trials no significant effect could be detected on serum creatinine in comparisons with placebo. The most relevant adverse event observed with this new class of drugs is represented by genital infections, which are increased about fourfold, both in placebo- and active comparator-controlled trials. Detectable concentrations of glucose in urines can facilitate the onset of mycotic infections, as observed in patients who experience severe hyperglycaemia with relevant glycosuria. The same mechanisms could also promote infections of lower urinary tract, which are increased in a significant manner, although to a lower extent. In conclusion, SGLT-2 inhibitors, orally administered once daily, are effective in the treatment of type 2 diabetes, providing additional benefits beyond glucose lowering, such as weight loss, reduction of blood pressure and increase in HDL cholesterol. Apart from genital and, to a lesser extent, urinary tract infections, which are rather frequent but usually mild, SGLT-2 inhibitors appear to be well tolerated. Further studies are granted to confirm their long-term safety. Acknowledgement This research was performed independently of any funding, as part of the institutional activity of the investigators. Conflict of Interest M. M. has received speaking fees from Bristol Myers Squibb, Eli- Lilly, Merck, Novonordisk, Merck and Takeda, and research grants from Bristol Myers Squibb. C. N. has no conflict of interests. E. M. has received consultancy fees from Merck and Novartis, speaking fees from Astra Zeneca, Bristol Myers Squibb, Merck and Novartis, and research grants from Merck, Novartis and Takeda. M. M. and E. M. designed this study and performed the data collection, analysis and writing of the manuscript. C. N. performed data collection and manuscript revision. All the authors approved the final version of this manuscript. References 1. Turner RC, Cull CA, Frighi V, Holman RR. Glycemic control with diet, sulfonylurea, metformin, or insulin in patients with type 2 diabetes mellitus: progressive requirement for multiple therapies (UKPDS 49). UK Prospective Diabetes Study (UKPDS) Group. JAMA 1999; 281: Inzucchi SE, Bergenstal RM, Buse JB et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach: position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2012; 35: original article 3. Abdul-Ghani MA, Norton L, Defronzo RA. Role of sodium-glucose cotransporter 2 (SGLT 2) inhibitors in the treatment of type 2 diabetes. Endocr Rev 2011; 32: Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009; 151: Jadad AR, Moore RA, Carroll D et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17: Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994; 50: Egger M, Davey SG, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: Schernthaner G, Gross JL, Rosenstock J et al. Canagliflozin compared with sitagliptin for patients with type 2 diabetes who do not have adequate glycemic control with metformin plus sulfonylurea: a 52-week randomized trial. Diabetes Care 2013; 36: Rosenstock J, Aggarwal N, Polidori D et al. Dose-ranging effects of canagliflozin, a sodium-glucose cotransporter 2 inhibitor, as add-on to metformin in subjects with type 2 diabetes. Diabetes Care 2012; 35: Nauck MA, Del Prato S, Meier JJ et al. Dapagliflozin versus glipizide as add-on therapy in patients with type 2 diabetes who have inadequate glycemic control with metformin: a randomized, 52-week, double-blind, active-controlled noninferiority trial. Diabetes Care 2011; 34: Clar C, Gill JA, Court R, Waugh N. Systematic review of SGLT2 receptor inhibitors in dual or triple therapy in type 2 diabetes. BMJ Open 2012; 2: e Musso G, Gambino R, Cassader M, Pagano G. A novel approach to control hyperglycemia in type 2 diabetes: sodium glucose co-transport (SGLT) inhibitors: systematic review and meta-analysis of randomized trials. Ann Med 2012; 44: Ferrannini E, Seman L, Seewaldt-Becker E, Hantel S, Pinnetti S, Woerle HJ. A Phase IIb, randomized, placebo-controlled study of the SGLT2 inhibitor empagliflozin in patients with type 2 diabetes. Diabetes Obes Metab 2013; 15: Fonseca VA, Ferrannini E, Wilding JP et al. Active- and placebo-controlled dose-finding study to assess the efficacy, safety, and tolerability of multiple doses of ipragliflozin in patients with type 2 diabetes mellitus. J Diabetes Complications 2013; 27: Henry RR, Murray AV, Marmolejo MH, Hennicken D, Ptaszynska A, List JF. Dapagliflozin, metformin XR, or both: initial pharmacotherapy for type 2 diabetes, a randomised controlled trial. Int J Clin Pract 2012; 66: List JF, Woo V, Morales E, Tang W, Fiedorek FT. Sodium-glucose cotransport inhibition with dapagliflozin in type 2 diabetes. Diabetes Care 2009; 32: NCT The CANTATA-MP Trial (CANagliflozin Treatment and Trial Analysis Metformin and Pioglitazone). Available from URL: Accessed 21 May Robinson WS. Ecological correlations and the behavior of individuals. Am Sociol Rev 1950; 15: Sha S, Devineni D, Ghosh A et al. Canagliflozin, a novel inhibitor of sodium glucose co-transporter 2, dose dependently reduces calculated renal threshold for glucose excretion and increases urinary glucose excretion in healthy subjects. Diabetes Obes Metab 2011; 13: Bode B, Stenlöf K, Sullivan D, Fung A, Usiskin K. Efficacy and safety of canagliflozin treatment in older subjects with type 2 diabetes mellitus: a randomized trial. Hosp Pract 2013; 41: Stenlöf K, Cefalu WT, Kim KA et al. Efficacy and safety of canagliflozin monotherapy in subjects with type 2 diabetes mellitus inadequately 2013 doi: /dom

10 controlled with diet and exercise. Diabetes Obes Metab 2013; 15: Yale JF, Bakris G, Cariou B et al. Efficacy and safety of canagliflozin in subjects with type 2 diabetes and chronic kidney disease. Diabetes Obes Metab 2013; 15: NCT Available from URL: Accessed 21 April Bailey CJ, Iqbal N, T joen C, List JF. Dapagliflozin monotherapy in drug-naïve patients with diabetes: a randomized-controlled trial of low-dose range. Diabetes Obes Metab 2012; 14: Bailey CJ, Gross JL, Hennicken D, Iqbal N, Mansfield TA, List JF. Dapagliflozin add-on to metformin in type 2 diabetes inadequately controlled with metformin: a randomized, double-blind, placebo-controlled 102-week trial. BMC Med 2013; 11: Bolinder J, Ljunggren Ö, Kullberg J et al. Effects of dapagliflozin on body weight, total fat mass, and regional adipose tissue distribution in patients with type 2 diabetes mellitus with inadequate glycemic control on metformin. J Clin Endocrinol Metab 2012; 97: Ferrannini E, Ramos SJ, Salsali A, Tang W, List JF. Dapagliflozin monotherapy in type 2 diabetic patients with inadequate glycemic control by diet and exercise: a randomized, double-blind, placebo-controlled, phase 3 trial. Diabetes Care 2010; 33: Lambers Heerspink HJ, de Zeeuw D, Wie L, Leslie B, List J. Dapagliflozin a glucose-regulating drug with diuretic properties in subjects with type 2 diabetes. Diabetes Obes Metab 2013; 5: DIABETES, OBESITY AND METABOLISM 29. Kaku K, Inoue S, Matsuoka O et al. Efficacy and safety of dapagliflozin as a monotherapy for type 2 diabetes mellitus in Japanese patients with inadequate glycaemic control: a phase II multicentre, randomized, double-blind, placebo-controlled trial. Diabetes Obes Metab 2013; 15: Rosenstock J, Vico M, Wei L, Salsali A, List JF. Effects of dapagliflozin, an SGLT2 inhibitor, on HbA(1c), body weight, and hypoglycemia risk in patients with type 2 diabetes inadequately controlled on pioglitazone monotherapy. Diabetes Care 2012; 35: Strojek K, Yoon KH, Hruba V, Elze M, Langkilde AM, Parikh S. Effect of dapagliflozin in patients with type 2 diabetes who have inadequate glycaemic control with glimepiride: a randomized, 24-week, double-blind, placebo-controlled trial. Diabetes Obes Metab 2011; 13: Wilding JP, Norwood P, T joen C, Bastien A, List JF, Fiedorek FT. A study of dapagliflozin in patients with type 2 diabetes receiving high doses of insulin plus insulin sensitizers: applicability of a novel insulin-independent treatment. Diabetes Care 2009; 32: Wilding JP, Woo V, Soler NG et al. Dapagliflozin 006 Study Group. Longterm efficacy of dapagliflozin in patients with type 2 diabetes mellitus receiving high doses of insulin: a randomized trial. Ann Intern Med 2012; 156: Wilding JP, Ferrannini E, Fonseca VA, Wilpshaar W, Dhanjal P, Houzer A. Efficacy and safety of ipragliflozin in patients with type 2 diabetes inadequately controlled on metformin: a dose-finding study. Diabetes Obes Metab 2013; 15: Monami et al. 2013