catherine only
O R
I G
I N
A L
A R
T I C
L E
Diabetes, Obesity and Metabolism 2016. © 2016 John Wiley & Sons Ltdoriginal article
Efficacy and safety of sodium-glucose co-transporter-2 inhibitors in type 2 diabetes mellitus: systematic review and network meta-analysis F. Zaccardi1,2, D. R. Webb1,2, Z. Z. Htike1,2, D. Youssef1,2, K. Khunti1,2 & M. J. Davies1,2 1 Diabetes Research Centre, University of Leicester, Leicester, UK 2 Diabetes Research Center, Leicester Diabetes Centre, UHL NHS Trust, Leicester, UK
Aim: To assess the comparative efficacy and safety of sodium-glucose co-transporter-2 (SGLT2) inhibitors in adults with type 2 diabetes. Methods: We electronically searched randomized controlled trials (≥24 weeks) including canagliflozin, dapagliflozin or empagliflozin that were published up to 3 November 2015. Data were collected on cardiometabolic and safety outcomes and synthesized using network meta-analyses. Results: A total of 38 trials (23 997 participants) were included. Compared with placebo, all SGLT2 inhibitors reduced glycated haemoglobin (HbA1c), fasting plasma glucose (FPG), body weight and blood pressure, and slightly increased HDL cholesterol. Canagliflozin 300 mg reduced HbA1c, FPG and systolic blood pressure and increased LDL cholesterol to a greater extent compared with other inhibitors at any dose. At their highest doses, canagliflozin 300 mg reduced: HbA1c by 0.2% [95% confidence interval (CI) 0.1–0.3] versus both dapagliflozin 10 mg and empagliflozin 25 mg; FPG by 0.6 mmol/l (95% CI 0.3–0.9) and 0.5 mmol/l (95% CI 0.1–0.8) versus dapagliflozin and empagliflozin, respectively; and systolic blood pressure by 2 mmHg (95% CI 1.0–3.0) versus dapagliflozin; and increased LDL cholesterol by 0.13 mmol/l (95% CI 0.03–0.23) and 0.15 mmol/l (95% CI 0.06–0.23) versus dapagliflozin and empagliflozin, respectively. The highest doses of inhibitors had similar effects on body weight reduction. Canagliflozin 300 and 100 mg increased the risk of hypoglycaemia versus placebo, dapagliflozin 10 mg and empagliflozin 10 mg [odds ratios (ORs) 1.4–1.6]. Dapagliflozin 10 mg increased the risk of urinary tract infection versus placebo and empagliflozin 25 mg (ORs 1.4). All inhibitors similarly increased the risk of genital infection (ORs 4–6 versus placebo). Conclusions: Although they increase the risk of genital infection, SGLT2 inhibitors are effective in improving cardiometabolic markers in type 2 diabetes, with canagliflozin 300 mg performing better in this respect than other inhibitors. Further studies will clarify whether these differences are likely to translate into differing long-term outcomes. Keywords: canagliflozin, dapagliflozin, empagliflozin, meta-analysis, review, SGLT2 inhibitor, systematic
Date submitted 24 February 2016; date of first decision 15 March 2016; date of final acceptance 31 March 2016
Introduction Type 2 diabetes mellitus is a complex disorder charac- terized by hyperglycaemia and progressive dysregulation of insulin–glucose feedback mechanisms [1]. Multiple intervention studies have shown the importance of glu- cose control in the reduction of long-term microvascular and, to some extent, macrovascular complications of the disease [2].
A range of glucose-lowering treatments are currently avail- able and they exert their main effects by modulating peripheral insulin resistance or 𝛽-cell insulin secretion [3]. More recently, a new class of glucose-lowering agents, which act through the inhibition of renal glucose reabsorption in the kidney, has been introduced [4]. In physiological conditions, glycosuria arises when the tubular threshold for glucose reabsorption is exceeded. As sodium-glucose co-transporter-2 (SGLT2) is the major cotransporter involved in tubular glucose reuptake,
Correspondence to: F. Zaccardi, Diabetes Research Centre, Leicester General Hospital, LE5 4PW Leicester, UK. E-mail: frazac@fastwebnet.it
inhibitors of its activity have been developed with the aim of enhancing glycosuria and reducing blood glucose levels [5].
The efficacy and safety of SGLT2 inhibitors have been inves- tigated in several randomized controlled trials (RCTs), showing improved glucose control and a reduction of body weight and blood pressure with a low risk of hypoglycaemia [5]. These drugs are recommended by the American Diabetes Association and European Association for the Study of Diabetes as a treat- ment option in patients on metformin with or without another glucose-lowering treatment if the personalized glucose target is not achieved [3]; however, no direct comparisons between specific SGLT2 inhibitors are available to date, thus limiting the possibility of a comparative assessment of their efficacy and safety.
Network meta-analysis is considered the methodology of choice to allow estimation of the comparative effectiveness of multiple treatments when direct ‘head-to-head’ data are unavailable [6]. Further to this, its value in informing health- care decision-making is increasingly recognized, given the pos- sibility of ranking treatments according to efficacy and safety [7]. Within this context, we conducted a systematic review
original article DIABETES, OBESITY AND METABOLISM and network meta-analysis to assess the comparative efficacy and safety of SGLT2 inhibitors canagliflozin, dapagliflozin and empagliflozin.
Materials and Methods Data Sources and Searches
This study was conducted according to a prespecified pro- tocol and followed the standard guidelines for conducting and reporting systematic reviews and network meta-analysis (PRISMA checklist reported in the Appendix) [8–10]. We searched PubMed, ISI Web of Science and the Cochrane Library for RCTs published in any language from inception until 3 November 2015 and comparing licensed doses of canagliflozin (100 or 300 mg), dapagliflozin (5 or 10 mg), or empagliflozin (10 or 25 mg) with placebo or other glucose-lowering drugs in adults with type 2 diabetes.
Study Selection
All RCTs lasting at least 24 weeks and reporting data on one or more cardiometabolic or safety outcomes were included. We excluded RCTs that included only patients with chronic kidney disease at baseline. Cardiometabolic outcomes com- prised glycated haemoglobin (HbA1c), fasting plasma glucose (FPG), body weight (primary outcomes); systolic and diastolic blood pressure, total cholesterol, LDL and HDL cholesterol, and triglycerides. Safety outcomes included all hypoglycaemic events, urinary tract infection, genital infection, diabetic ketoacidosis and bone fractures. Reference lists of eligible stud- ies, as well as systematic reviews and meta-analyses of SGLT2 inhibitors, were manually scanned for additional relevant studies.
Data Extraction and Quality Assessment
Three authors independently performed the literature search. After the identification of eligible studies, the three authors independently extracted data, using standardized prede- fined forms, on: first author name; clinical trial registration number; year of journal article publication; background glucose-lowering therapy; SGLT2 inhibitor(s) and compara- tor(s); duration of follow-up; sample size; gender distribution; age; diabetes duration; baseline HbA1c; and outcome mea- sured. We extracted outcomes data as: arm-specific counts (i.e. number of participants, mean difference and standard error (or standard deviation) for continuous outcomes in patients with baseline and at least one post-baseline measurement); total number of participants and participants with event for dichotomous outcomes in patients who were randomized and received treatment; or contrast-based estimations (i.e. pairwise comparisons). When studies reported outcomes data for differ- ent durations of follow-up, the longest was used. We retrieved data from ClinicalTrials.gov when it was not possible to extract relevant information from the published report. In cases where the independent reviewers disagreed on the eligibility of an article or data extraction, consensus was reached by arbitra- tion. Study quality was assessed using the Cochrane risk of bias tool [11].
Data Synthesis and Analysis
We undertook a network meta-analysis within a frequentist model, an alternative to the Bayesian approach. Stata 14.1 (Stata Corp, College Station, TX, USA) was used for all analyses. Pair- wise random-effects meta-analyses were performed using the DerSimonian and Laird method [12]. Network meta-analyses were based on the method of multivariate meta-analysis, with the assumption that all treatment contrasts have the same het- erogeneity variance [13–15]. Results were reported with 95% confidence intervals (CIs); we considered p values <0.05 as sta- tistically significant.
In three-arm trials reporting contrasted-based estimates for continuous outcomes, pairwise comparisons were available only for two out of the three possible contrasts (i.e. A vs B and B vs C, but not A vs C, where A, B and C denote the three arms); in these cases, given the presence of corre- lations between the treatment differences, the standard error (𝜎) of the missing contrast was estimated using the formula: 𝜎2 AC = 𝜎2 AB + 𝜎2 BC − 2𝜌𝜎AB𝜎BC [16]. As 𝜌 values (correla- tions) were not reported, we used in the main analysis a value similar to those obtained from other similar studies included in this systematic review (𝜌 = 0.5), as previously advocated [16]. We performed sensitivity analyses assuming different values of 𝜌 (0.2, 0.7 and 0.9). Similar results were obtained and therefore we report only the results of the main analysis.
We combined linagliptin and sitagliptin in a single group [dipeptidyl peptidase-4 (DPP-4) inhibitors] and glimepiride and gliclazide in another (sulphonylureas). For the primary outcomes, we reported random-effect pairwise meta-analyses with heterogeneity across studies estimated using the I2 statis- tic. For dichotomous outcomes, we used odds ratio (OR) as effect measure and added 0.5 when studies reported zero events. For all outcomes, we summarized the evidence by using a net- work diagram [17]. We reported characteristics and summary data of included studies in tables. We presented results against a common comparator (placebo) in forest plots and showed comparisons across SGLT2 inhibitors in tables [18]; we also displayed graphically with radar plots the ranking probabil- ities (network rank command) for different cardiometabolic and safety outcomes and reported comparison-adjusted funnel plots (netfunnel command) to assess the association between study size and result. We assumed that participants of the included RCTs could be randomly allocated to any of the three treatments being compared (on average, the baseline charac- teristics of participants are similar as the treatments are tested for a wide range of patients). For each network, we assessed consistency between direct and indirect evidence using the ‘design-by-treatment’ interaction model [19].
Results Study Characteristics
From 2174 identified records we excluded non-human and observational studies, leaving 79 reports for full-text assess- ment. After further selection (Figure S1), 38 unique RCTs ful- filled the inclusion criteria (Table 1) [20–57]. RCTs published between 2012 and 2015 included 23 997 (range 136–2072)
2 Zaccardi et al. 2016
DIABETES, OBESITY AND METABOLISM original article Ta
bl e
1. B
as el
in e
ch ar
ac te
ri st
ic s
of th
e in
cl ud
ed st
ud ie
s.
Fi rs
ta ut
ho r
Ye ar
B ac
kg ro
un d
th er
ap y
SG LT
2 in
hi bi
to r
C om
pa ra
to r
(s )
St ud
y du
ra ti
on ,w
ee ks
To ta
ls am
pl e
si ze
,N M
al e,
% A
ge *,
ye ar
s D
ia be
te s
du ra
ti on
*, ye
ar s
H bA
1c *,
%
In ag
ak i[
33 ]
20 14
D ie
t+ PA
C an
a 10
0 m
g Pl
ac eb
o 24
18 3
65 .0
58 .3
5. 2
8. 0
B od
e [2
4] 20
15 D
ie t±
O A
D s±
in su
lin C
an a
10 0
m g,
C an
a 30
0 m
g Pl
ac eb
o 10
4 71
4 55
.5 63
.6 11
.7 7.
8 Fo
rs t[
30 ]
20 14
M et +
Pi o
C an
a 10
0 m
g, C
an a
30 0
m g
Pl ac
eb o
26 34
2 63
.2 57
.3 10
.5 8.
0 N
ea l[
45 ]
20 15
In su
lin ±
O A
D s
C an
a 10
0 m
g, C
an a
30 0
m g
Pl ac
eb o
52 20
72 66
.1 62
.7 16
.2 8.
3 St
en lö
f[ 53
] 20
13 D
ie t+
PA C
an a
10 0
m g,
C an
a 30
0 m
g Pl
ac eb
o 26
58 4
44 .2
55 .4
4. 3
8. 0
W ild
in g
[5 5]
20 13
M et +
SU C
an a
10 0
m g,
C an
a 30
0 m
g Pl
ac eb
o 52
46 9
51 .0
56 .8
9. 6
8. 1
La va
lle -G
on zá
le z
[3 9]
20 13
M et
C an
a 10
0 m
g, C
an a
30 0
m g
Pl ac
eb o,
D PP
-4 in
hi bi
to rs
26 12
84 47
.1 55
.4 6.
9 7.
9 Le
ite r
[4 1]
20 15
M et
C an
a 10
0 m
g, C
an a
30 0
m g
SU 10
4 14
50 52
.1 56
.2 6.
6 7.
8 Sc
he rn
th an
er [5
2] 20
13 M
et +
SU C
an a
30 0
m g
D PP
-4 in
hi bi
to rs
52 75
5 55
.9 56
.7 9.
6 8.
1 B
ai le
y [2
1] 20
12 D
ie t+
PA D
ap a
5 m
g Pl
ac eb
o 24
13 6
50 .7
52 .4
1. 3
7. 9
B ai
le y
[2 2]
20 13
M et
D ap
a 5
m g,
D ap
a 10
m g
Pl ac
eb o
10 2
40 9
54 .3
53 .6
6. 1
8. 1
Ji [3
5] 20
14 D
ie t+
PA D
ap a
5 m
g, D
ap a
10 m
g Pl
ac eb
o 24
39 3
65 .4
51 .3
1. 4
8. 3
K ak
u [3
7] 20
14 D
ie t+
PA D
ap a
5 m
g, D
ap a
10 m
g Pl
ac eb
o 24
26 1
59 .4
58 .8
4. 9
7. 5
R os
en st
oc k
[4 8]
20 12
Pi o
D ap
a 5
m g,
D ap
a 10
m g
Pl ac
eb o
24 42
0 49
.5 53
.5 5.
5 8.
4 St
ro je
k [5
4] 20
14 G
lim ep
ir id
e D
ap a
5 m
g, D
ap a
10 m
g Pl
ac eb
o 48
43 8
47 .5
59 .8
7. 3
8. 1
B ai
le y
[2 3]
20 15
D ie
t+ PA
D ap
a 5
m g,
D ap
a 10
m g
M et
10 2
20 9
45 .9
52 .0
1. 8
7. 9
H en
ry [3
2] 20
12 D
ie t+
PA D
ap a
5 m
g, D
ap a
10 m
g M
et 24
83 1
46 .8
52 .0
1. 8
9. 1
B ol
in de
r [2
5] 20
14 M
et D
ap a
10 m
g Pl
ac eb
o 10
2 18
0 55
.6 60
.7 5.
7 7.
2 C
ef al
u [2
6] 20
15 O
A D
s± In
su lin
D ap
a 10
m g
Pl ac
eb o
52 91
4 68
.3 62
.9 12
.4 8.
1 Ja
bb ou
r [3
4] 20
14 Si
ta ±
M et
D ap
a 10
m g
Pl ac
eb o
48 44
7 54
.8 54
.9 5.
7 8.
0 Le
ite r
[4 0]
20 14
O A
D s+
In su
lin D
ap a
10 m
g Pl
ac eb
o 52
96 2
66 .9
63 .7
13 .2
8. 0
M at
th ae
i[ 43
] 20
15 M
et +
SU D
ap a
10 m
g Pl
ac eb
o 52
21 8
48 .6
61 .0
9. 5
8. 2
M at
hi eu
[5 7]
20 15
Sa xa
+ M
et D
ap a
10 m
g Pl
ac eb
o 24
32 0
45 .6
55 .1
7. 6
8. 2
W ild
in g
[5 6]
20 14
In su
lin ±
O A
D s
D ap
a 10
m g
Pl ac
eb o
10 4
38 7
47 .0
59 .1
13 .9
8. 5
R os
en st
oc k
[5 1]
20 15
M et
D ap
a 10
m g
D PP
-4 in
hi bi
to rs
24 35
5 51
.5 54
.5 7.
8 8.
9 D
el Pr
at o
[2 8]
20 15
M et
D ap
a 10
m g
SU 20
8 80
1 55
.1 58
.4 6.
4 7.
7 H
ae ri
ng [3
1] 20
15 M
et +
SU Em
pa 10
m g,
Em pa
25 m
g Pl
ac eb
o 76
66 6
50 .9
57 .1
8. 5
8. 1
K ov
ac s
[3 8]
20 15
Pi o ±
M et
Em pa
10 m
g, Em
pa 25
m g
Pl ac
eb o
76 49
8 48
.4 54
.5 5.
5 8.
1 M
er ke
r [4
4] 20
15 M
et Em
pa 10
m g,
Em pa
25 m
g Pl
ac eb
o 76
63 7
71 .3
55 .7
6. 3
7. 9
R os
en st
oc k
[4 9]
20 14
In su
lin ±
M et
Em pa
10 m
g, Em
pa 25
m g
Pl ac
eb o
52 56
3 45
.5 56
.7 10
.5 8.
3 R
os en
st oc
k [5
0] 20
15 In
su lin
Em pa
10 m
g, Em
pa 25
m g
Pl ac
eb o
78 49
4 55
.9 58
.8 7.
0 8.
3 R
od en
[4 7]
20 13
D ie
t+ PA
Em pa
10 m
g, Em
pa 25
m g
Pl ac
eb o,
D PP
-4 in
hi bi
to rs
24 89
9 61
.3 55
.0 3.
3 7.
9 Le
w in
[4 2] †
20 15
D ie
t+ PA
Em pa
10 m
g, Em
pa 25
m g
D PP
-4 in
hi bi
to rs
52 40
5 53
.3 54
.6 3.
5 8.
0 D
eF ro
nz o
[2 7] †
20 15
D ie
t+ PA
+ M
et Em
pa 10
m g,
Em pa
25 m
g D
PP -4
in hi
bi to
rs 52
40 5
51 .1
55 .9
6. 5
8. 0
Fe rr
an ni
ni [2
9] ‡
20 13
D ie
t+ PA
,M et
Em pa
10 m
g, Em
pa 25
m g
M et
,D PP
-4 in
hi bi
to rs
90 65
9 50
.7 59
.5 4.
6 8.
0 A
ra ki
[2 0]
§ 20
15 A
G I,
B ig
ua ni
de ,D
PP -4
in hi
bi to
rs ,G
lin id
e, T
Z D
,S U
Em pa
10 m
g, Em
pa 25
m g
M et
52 11
60 72
.0 60
.3 6.
9 7.
9 K
ad ow
ak i[
36 ]
20 15
D ie
t+ PA
± O
A D
(n o
T Z
D )
Em pa
10 m
g, Em
pa 25
m g
— 52
53 2
74 .8
57 .6
N R
7. 9
R id
de rs
tr al
e [4
6] 20
14 D
ie t+
PA +
M et
Em pa
25 m
g SU
10 4
15 45
55 .2
55 .9
5. 6
7. 9
A G
I, 𝛼
-g lu
co si
da se
in hi
bi to
r; C
an a,
ca na
gl ifl
oz in
;D ap
a, da
pa gl
ifl oz
in ;D
PP -4
,d ip
ep ti
dy lp
ep ti
da se
-4 ;E
m pa
,e m
pa gl
ifl oz
in ;M
et ,m
et fo
rm in
;N R
,n ot
re po
rt ed
;O A
D s,
or al
gl uc
os e–
lo w
er in
g dr
ug s;
PA ,p
hy si
ca la
ct iv
ity ;P
io ,p
io gl
ita zo
ne ;S
ax a,
sa xa
gl ip
ti n;
Si ta
,s ita
gl ip
ti n;
SU ,s
ul ph
on yl
ur ea
;T Z
D ,t
hi az
ol id
in ed
io ne
. *W
he n
no tr
ep or
te d
fo r
th e
ov er
al lp
op ul
at io
n, va
lu es
ha ve
be en
es ti
m at
ed as
w ei
gh te
d m
ea ns
. †D
at a
fo r
pr im
ar y
ou tc
om es
(H bA
1c ,f
as ti
ng pl
as m
a gl
uc os
e, bo
dy w
ei gh
t) at
24 w
ee ks
. ‡C
om pa
ra to
rs w
er e
m et
fo rm
in (b
ac kg
ro un
d, D
ie t+
PA )o
r D
PP -4
in hi
bi to
rs .(
ba ck
gr ou
nd ,m
et fo
rm in
); da
ta fo
r bl
oo d
pr es
su re
at 78
w ee
ks ;a
ge re
po rt
ed as
m ed
ia n.
§D at
a re
po rt
ed by
ba ck
gr ou
nd th
er ap
y; m
et fo
rm in
as op
en -l
ab el
co m
pa ra
to r
fo r
su lp
ho ny
lu re
a ba
ck gr
ou nd
th er
ap y.
2016 doi:10.1111/dom.12670 3
original article DIABETES, OBESITY AND METABOLISM
Figure 1. Network maps for cardiometabolic and safety outcomes. Nodes represent the competing treatments and their size is proportional to the number of participants; edges represent the available direct comparisons between pairs of treatments and their width is proportional to the number of trials comparing every pair. Cana100, canagliflozin 100 mg; Cana300, canagliflozin 300 mg; Dapa5, dapagliflozin 5 mg; Dapa10, dapagliflozin 10 mg; DPP-4i, dipeptidyl peptidase-4 inhibitor; Empa10, empagliflozin 10 mg; Empa25, empagliflozin 25 mg; Met, metformin; SU, sulphonylurea.
participants with type 2 diabetes and 34 (89.5%) studies were multinational RCTs. Baseline HbA1c, age and disease duration weighted means were 8.1%, 58 and 8 years, respectively, 57% of the participants were male and follow-up duration ranged from 24 to 208 weeks. Other characteristics of the RCTs, such as study-, drug- and outcome-specific available data, are reported in Tables S1–S5.
Overall, the risk of bias for the domains included in the Cochrane risk assessment tool were judged to be low, high and unclear in 89.5, 1.8 and 8.7% of the cases, respectively; high or unclear domain-specific bias was lowest for blind- ing of outcome assessment (2.7%) and highest for random sequence generation (15.8%; Table S5 and Figure S2). The risk of bias was high or unclear in 1.8, 10.8 and 16.7% of canagliflozin, dapagliflozin and empagliflozin RCTs, respec- tively. Networks of evidence for all outcomes are graphically displayed in Figure 1.
Meta-analyses
Primary Outcomes: Glycated Haemoglobin, Fasting Plasma Glucose and Body Weight. Data on HbA1c were available from all RCTs. Direct pairwise random-effects meta-analyses showed significant reductions in HbA1c versus placebo, from −0.9% (95% CI −1.0 to −0.7) or −9.8 mmol/mol (95% CI −10.9 to −7.6) for canagliflozin 300 mg to −0.6% (95% CI −0.7 to −0.4) or −6.5 mmol/mol (95% CI −7.6 to −4.4) for
dapagliflozin 5 mg (Figure S3). When compared with other glucose-lowering drugs (sulphonylureas, DPP-4 inhibitors or metformin), pairwise differences ranged from a significant reduction of −0.3% (95% CI −0.5 to −0.1) or −3.3 mmol/mol (95% CI −5.4 to −1.1), comparing dapagliflozin 10 mg with DPP-4 inhibitors, to a non-significant increase of 0.1% (95% CI −0.1 to 0.2) or 1.1 mmol/mol (95% CI −1.1 to 2.2) for empagliflozin 10 mg versus metformin (Figure S3). The results of the network meta-analysis showed a mean HbA1c reduc- tion, compared with placebo, of: −0.9% (95% CI −1.0 to −0.8) for canagliflozin 300 mg [−9.4 mmol/mol (95% CI −10.5 to −8.3)]; −0.8% (95% CI −0.9 to −0.7) for canagliflozin 100 mg [−8.3 mmol/mol (95% CI −9.4 to −7.2)]; −0.7% (95% CI −0.8 to −0.6) for empagliflozin 25 mg [−7.2 mmol/mol (95% CI −8.3 to −6.1)]; −0.7% (95% CI −0.7 to −0.6) for dapagliflozin 10 mg [−7.2 mmol/mol (95% CI −8.1 to −6.3)]; −0.6% (95% CI −0.7 to 0.5) for empagliflozin 10 mg [−6.6 mmol/mol (95% CI −7.7 to 5.5)]; and −0.6% (95% CI −0.7 to −0.4) for dapagliflozin 5 mg [−6.1 mmol/mol (95% CI −7.3 to 4.8); Table 2 and Figure 2]. Comparisons among SGLT2 inhibitors showed greater HbA1c reductions with canagliflozin 300 mg compared with all other SGLT2 drugs [from −0.3% (−3.3 mmol/mol) vs dapagliflozin 5 mg to −0.1% (−1.1 mmol/mol) vs canagliflozin 100 mg] and no significant differences between dapagliflozin and empagliflozin at different doses (Table 2). Figures S4 and S5 show SGLT2 inhibitors according to the ranking probabilities and the mean rank, respectively.
4 Zaccardi et al. 2016
DIABETES, OBESITY AND METABOLISM original article Ta
bl e
2. C
om pa
ri so
ns of
so di
um -g
lu co
se co
-t ra
ns po
rt er
-2 in
hi bi
to rs
w ith
re ga
rd to
ca rd
io m
et ab
ol ic
ou tc
om es
.
H bA
1c ,%
H bA
1c ,m
m ol
/m ol
P la
ce bo
P la
ce bo
C an
a 30
0 m
g 0.
86 (0
.7 6,
0. 96
) C
an a
30 0
m g
9. 40
(8 .3
1, 10
.4 9)
C an
a 10
0 m
g −
0. 10
(− 0.
20 ,−
0. 00
) 0.
76 (0
.6 6,
0. 86
) C
an a
10 0
m g
− 1.
09 (−
2. 19
,0 .0
0) 8.
31 (7
.2 1,
9. 40
) D
ap a
10 m
g −
0. 10
(− 0.
23 ,0
.0 2)
− 0.
21 (−
0. 33
,− 0.
08 )
0. 66
(0 .5
8, 0.
74 )
D ap
a 10
m g
− 1.
09 (−
2. 51
,0 .2
2) −
2. 30
(− 3.
61 ,−
0. 87
) 7.
21 (6
.3 4,
8. 09
) D
ap a
5 m
g −
0. 10
(− 0.
21 ,0
.0 2)
− 0.
20 (−
0. 35
,− 0.
05 )
− 0.
30 (−
0. 45
,− 0.
15 )
0. 56
(0 .4
4, 0.
67 )
D ap
a 5
m g
− 1.
09 (−
2. 30
,0 .2
2) −
2. 19
(− 3.
83 ,−
0. 55
) −
3. 28
(− 4.
92 ,−
1. 64
) 6.
12 (4
.8 1,
7. 32
) E
m pa
25 m
g 0.
10 ( −
0. 05
,0 .2
4) 0.
00 (−
0. 12
,0 .1
2) −
0. 10
(− 0.
23 ,0
.0 3)
− 0.
20 (−
0. 33
,− 0.
08 )
0. 66
(0 .5
6, 0.
76 )
E m
pa 25
m g
1. 09
(− 0.
55 ,2
.6 2)
0. 00
(− 1.
31 ,1
.3 1)
− 1.
09 (−
2. 51
,0 .3
3) −
2. 19
(− 3.
61 ,−
0. 87
) 7.
21 (6
.1 2,
8. 31
) E
m pa
10 m
g −
0. 05
(− 0.
13 ,0
.0 2)
0. 04
(− 0.
10 ,0
.1 9)
− 0.
05 (−
0. 17
,0 .0
7) −
0. 16
(− 0.
29 ,−
0. 03
) −
0. 26
(− 0.
39 ,−
0. 13
) 0.
60 (0
.5 0,
0. 70
) E
m pa
10 m
g −
0. 55
(− 1.
42 ,0
.2 2)
0. 44
(− 1.
09 ,2
.0 8)
− 0.
55 (−
1. 86
,0 .7
7) −
1. 75
(− 3.
17 ,−
0. 33
) −
2. 84
(− 4.
26 , −
1. 42
) 6.
56 (5
.4 7,
7. 65
)
Fa st
in g
pl as
m a
gl uc
os e,
m m
ol /l
B od
y w
ei gh
t, kg
P la
ce bo
P la
ce bo
C an
a 30
0 m
g 1.
93 (1
.6 8,
2. 19
) C
an a
30 0
m g
2. 47
(2 .1
1, 2.
84 )
C an
a 10
0 m
g −
0. 33
(− 0.
59 ,−
0. 07
) 1.
60 (1
.3 5,
1. 85
) C
an a
10 0
m g
− 0.
61 (−
0. 99
,− 0.
23 )
1. 86
(1 .4
9, 2.
23 )
D ap
a 10
m g
− 0.
22 (−
0. 53
,0 .0
8) −
0. 55
(− 0.
86 ,−
0. 25
) 1.
38 (1
.1 9,
1. 56
) D
ap a
10 m
g 0.
32 (−
0. 13
,0 .7
7) −
0. 29
(− 0.
74 ,0
.1 5)
2. 18
(1 .9
0, 2.
47 )
D ap
a 5
m g
− 0.
26 (−
0. 51
,− 0.
01 )
− 0.
48 (−
0. 83
,− 0.
13 )
− 0.
81 (−
1. 16
,− 0.
46 )
1. 12
(0 .8
7, 1.
37 )
D ap
a 5
m g
− 0.
60 (−
1. 00
,− 0.
19 )
− 0.
28 (−
0. 82
,0 .2
6) −
0. 89
(− 1.
43 ,−
0. 35
) 1.
58 (1
.1 8,
1. 99
) E
m pa
25 m
g 0.
36 (0
.0 5,
0. 67
) 0.
10 (−
0. 16
,0 .3
7) −
0. 12
(− 0.
43 ,0
.1 9)
− 0.
45 (−
0. 76
,− 0.
14 )
1. 48
(1 .2
7, 1.
70 )
E m
pa 25
m g
0. 66
(0 .1
6, 1.
16 )
0. 06
(− 0.
35 ,0
.4 7)
0. 38
(− 0.
08 ,0
.8 5)
− 0.
23 (−
0. 69
,0 .2
2) 2.
24 (1
.9 1,
2. 58
) E
m pa
10 m
g −
0. 14
(− 0.
30 ,0
.0 2)
0. 22
(− 0.
09 ,0
.5 4)
− 0.
04 (−
0. 30
,0 .2
3) −
0. 26
(− 0.
57 ,0
.0 6)
− 0.
59 (−
0. 90
,− 0.
28 )
1. 34
(1 .1
2, 1.
56 )
E m
pa 10
m g
− 0.
11 (−
0. 38
,0 .1
5) 0.
55 (0
.0 4,
1. 05
) −
0. 05
(− 0.
47 ,0
.3 7)
0. 27
(− 0.
20 ,0
.7 4)
− 0.
34 (−
0. 81
,0 .1
2) 2.
13 (1
.7 9,
2. 47
)
Sy st
ol ic
bl oo
d pr
es su
re ,m
m H
g D
ia st
ol ic
bl oo
d pr
es su
re ,m
m H
g
P la
ce bo
P la
ce bo
C an
a 30
0 m
g 4.
87 (3
.8 7,
5. 87
) C
an a
30 0
m g
2. 03
(1 .4
4, 2.
63 )
C an
a 10
0 m
g −
0. 98
(− 1.
96 ,−
0. 00
) 3.
89 (2
.8 8,
4. 90
) C
an a
10 0
m g
− 0.
39 (−
0. 97
,0 .1
9) 1.
64 (1
.0 3,
2. 25
) D
ap a
10 m
g −
0. 88
(− 2.
16 ,0
.3 9)
− 1.
87 (−
3. 13
,− 0.
61 )
3. 01
(2 .1
3, 3.
89 )
D ap
a 10
m g
0. 09
(− 0.
91 ,1
.1 0)
− 0.
30 (−
1. 29
,0 .6
9) 1.
74 (0
.8 9,
2. 58
) D
ap a
5 m
g −
0. 17
(− 1.
47 ,1
.1 3)
− 1.
05 (−
2. 66
,0 .5
6) −
2. 04
(− 3.
64 ,−
0. 43
) 2.
84 (1
.5 4,
4. 14
) D
ap a
5 m
g −
0. 20
(− 1.
15 ,0
.7 5)
− 0.
11 (−
1. 21
,0 .9
9) −
0. 50
(− 1.
59 ,0
.5 9)
1. 53
(0 .5
9, 2.
48 )
E m
pa 25
m g
0. 82
(− 0.
69 ,2
.3 2)
0. 65
(− 0.
51 ,1
.8 1)
− 0.
24 ( −
1. 49
,1 .0
1) −
1. 22
(− 2.
44 ,0
.0 1)
3. 66
(2 .7
7, 4.
54 )
E m
pa 25
m g
0. 38
(− 0.
66 ,1
.4 2)
0. 17
(− 0.
77 ,1
.1 2)
0. 27
(− 0.
47 ,1
.0 0)
− 0.
12 (−
0. 84
,0 .5
9) 1.
91 (1
.3 8,
2. 44
) E
m pa
10 m
g −
0. 33
(− 1.
12 ,0
.4 6)
0. 49
(− 1.
04 ,2
.0 2)
0. 32
(− 0.
88 ,1
.5 2)
− 0.
57 (−
1. 85
,0 .7
2) −
1. 55
(− 2.
82 ,−
0. 28
) 3.
33 (2
.4 1,
4. 25
) E
m pa
10 m
g −
0. 26
(− 0.
73 ,0
.2 1)
0. 12
(− 0.
93 ,1
.1 7)
− 0.
08 (−
1. 03
,0 .8
7) 0.
01 (−
0. 75
,0 .7
7) −
0. 38
(− 1.
13 ,0
.3 6)
1. 65
(1 .1
0, 2.
20 )
2016 doi:10.1111/dom.12670 5
original article DIABETES, OBESITY AND METABOLISM
Ta bl
e 2.
co nt
in ue
d
To ta
lc ho
le st
er ol
,m m
ol /l
L ow
-d en
si ty
lip op
ro te
in ch
ol es
te ro
l, m
m ol
/l
P la
ce bo
P la
ce bo
C an
a 30
0 m
g —
C an
a 30
0 m
g −
0. 20
(− 0.
26 ,−
0. 13
) C
an a
10 0
m g
— —
C an
a 10
0 m
g 0.
10 (0
.0 4,
0. 17
) −
0. 09
(− 0.
16 ,−
0. 03
) D
ap a
10 m
g −
— −
0. 04
(− 0.
11 ,0
.0 3)
D ap
a 10
m g
0. 02
(− 0.
08 ,0
.1 3)
0. 13
(0 .0
3, 0.
23 )
− 0.
07 (−
0. 15
,0 .0
1) D
ap a
5 m
g −
0. 02
(− 0.
11 ,0
.0 7)
— —
− 0.
06 (−
0. 14
,0 .0
2) D
ap a
5 m
g 0.
03 (−
0. 09
,0 .1
5) 0.
05 (−
0. 08
,0 .1
9) 0.
16 (0
.0 2,
0. 29
) −
0. 04
(− 0.
16 ,0
.0 8)
E m
pa 25
m g
− 0.
03 (−
0. 13
,0 .0
7) −
0. 05
(− 0.
14 ,0
.0 4)
— —
− 0.
09 (−
0. 15
,− 0.
03 )
E m
pa 25
m g
− 0.
01 (−
0. 14
,0 .1
2) 0.
02 (−
0. 08
,0 .1
2) 0.
04 (−
0. 04
,0 .1
3) 0.
15 (0
.0 6,
0. 23
) −
0. 05
(− 0.
11 ,0
.0 1)
E m
pa 10
m g
0. 03
(− 0.
02 ,0
.0 7)
− 0.
00 (−
0. 10
,0 .0
9) −
0. 02
(− 0.
11 ,0
.0 6)
— —
− 0.
06 (−
0. 12
,− 0.
00 )
E m
pa 10
m g
0. 02
(− 0.
03 ,0
.0 6)
0. 01
(− 0.
12 ,0
.1 4)
0. 04
(− 0.
06 ,0
.1 4)
0. 06
(− 0.
02 ,0
.1 5)
0. 17
(0 .0
8, 0.
25 )
− 0.
03 (−
0. 09
,0 .0
3)
H ig
h- de
ns it
y lip
op ro
te in
ch ol
es te
ro l,
m m
ol /l
Tr ig
ly ce
ri de
s, m
m ol
/l
P la
ce bo
P la
ce bo
C an
a 30
0 m
g −
0. 07
(− 0.
10 ,−
0. 05
) C
an a
30 0
m g
0. 20
(0 .1
0, 0.
30 )
C an
a 10
0 m
g 0.
01 (−
0. 01
,0 .0
4) −
0. 06
(− 0.
09 ,−
0. 04
) C
an a
10 0
m g
− 0.
04 (−
0. 14
,0 .0
6) 0.
16 (0
.0 6,
0. 26
) D
ap a
10 m
g −
0. 01
(− 0.
05 ,0
.0 3)
0. 00
(− 0.
04 ,0
.0 5)
− 0.
07 (−
0. 10
,− 0.
04 )
D ap
a 10
m g
− 0.
06 (−
0. 20
,0 .0
7) −
0. 10
(− 0.
24 ,0
.0 3)
0. 09
(− 0.
01 ,0
.1 9)
D ap
a 5
m g
0. 01
(− 0.
04 ,0
.0 6)
0. 00
(− 0.
05 ,0
.0 6)
0. 02
(− 0.
04 ,0
.0 7)
− 0.
06 (−
0. 11
,− 0.
01 )
D ap
a 5
m g
0. 02
(− 0.
11 ,0
.1 5)
− 0.
04 (−
0. 20
,0 .1
2) −
0. 08
(− 0.
24 ,0
.0 8)
0. 12
( − 0.
01 ,0
.2 4)
E m
pa 25
m g
0. 02
(− 0.
04 ,0
.0 7)
0. 03
(− 0.
01 ,0
.0 7)
0. 02
(− 0.
01 ,0
.0 5)
0. 03
(− 0.
00 ,0
.0 6)
− 0.
04 (−
0. 07
,− 0.
02 )
E m
pa 25
m g
− 0.
14 (−
0. 29
,0 .0
2) −
0. 12
(− 0.
24 ,0
.0 1)
− 0.
18 (−
0. 31
,− 0.
05 )
− 0.
22 (−
0. 35
,− 0.
09 )
− 0.
02 (−
0. 11
,0 .0
7) E
m pa
10 m
g −
0. 01
(− 0.
02 ,0
.0 1)
0. 01
(− 0.
04 ,0
.0 6)
0. 02
(− 0.
02 ,0
.0 6)
0. 01
(− 0.
02 ,0
.0 5)
0. 03
(− 0.
01 ,0
.0 6)
− 0.
05 (−
0. 07
,− 0.
03 )
E m
pa 10
m g
0. 03
(− 0.
04 ,0
.0 9)
− 0.
11 (−
0. 27
,0 .0
5) −
0. 09
(− 0.
22 ,0
.0 4)
− 0.
15 (−
0. 28
,− 0.
02 )
− 0.
19 (−
0. 32
,− 0.
06 )
0. 01
(− 0.
09 ,0
.1 0)
D at
a ar
e re
po rt
ed as
m ea
n di
ffe re
nc e
(9 5%
co nfi
de nc
e in
te rv
al )a
nd in
di ca
te co
lu m
n- to
-r ow
di ffe
re nc
es (i
.e .c
om pa
re d
w ith
pl ac
eb o,
ca na
gl ifl
oz in
30 0
m g
re du
ce s
H bA
1c by
0. 86
% ).
St at
is tic
al ly
si gn
ifi ca
nt di
ffe re
nc es
ar e
in bo
ld .
C an
a, ca
na gl
ifl oz
in ;D
ap a,
da pa
gl ifl
oz in
;E m
pa ,e
m pa
gl ifl
oz in
;H bA
1c ,g
ly ca
te d
ha em
og lo
bi n.
6 Zaccardi et al. 2016
DIABETES, OBESITY AND METABOLISM original article
Figure 2. Differences vs placebo (dotted lines) in cardiometabolic outcomes for the drugs included in the network meta-analysis. Cana100, canagliflozin 100 mg; Cana300, canagliflozin 300 mg; Dapa5, dapagliflozin 5 mg; Dapa10, dapagliflozin 10 mg; Empa10, empagliflozin 10 mg; Empa25, empagliflozin 25 mg; DPP-4i, dipeptidyl peptidase-4 inhibitor; Met, metformin; SU, sulphonylurea.
Values of FPG were available from 37 RCTs. Pairwise random-effects meta-analyses evidenced significant reduc- tions in FPG versus placebo for all SGLT2 inhibitors, from −2.0 mmol/l (95% CI −2.4 to −1.6) for canagliflozin 300 mg to −1.1 mmol/l (95% CI −1.5 to −0.7) for dapagliflozin 5 mg (Figure S3). Comparing SGLT2 inhibitors with other glucose-lowering drugs, differences ranged from −1.2 mmol/l (95% CI −1.6 to −0.8) for canagliflozin 300 mg versus DPP-4 inhibitors to −0.3 mmol/l (95% CI −0.6 to −0.1) for empagliflozin 10 mg versus metformin (Figure S3). Net- work meta-analysis results similarly showed a reduction of FPG for all SGLT2 inhibitors compared with placebo: −1.9 mmol/l (95% CI −2.2 to −1.7) for canagliflozin 300 mg; −1.6 mmol/l (95% CI −1.9 to −1.4) for canagliflozin 100 mg; −1.5 mmol/l (95% CI −1.7 to −1.3) for empagliflozin 25 mg; −1.4 mmol/l (95% CI −1.6 to −1.2) for dapagliflozin 10 mg; −1.3 mmol/l (95% CI −1.6 to −1.1) for empagliflozin 10 mg; and −1.1 mmol/l (95% CI −1.4 to −0.9) for dapagliflozin 5 mg (Table 2; Figure 2). Among SGLT2 inhibitors, canagliflozin 300 mg reduced FPG to a greater extent compared with all other inhibitors (from −0.8 mmol/l vs dapagliflozin 5 mg to −0.3 mmol/l vs canagliflozin 100 mg; Table 2).
Data on body weight were available from 37 RCTs. Pairwise random-effects meta-analyses showed significant reductions in body weight versus placebo for all SGLT2 treatments, from −2.8 kg (95% CI −3.2 to −2.4) for canagliflozin 300 mg to −1.6 kg (95% CI −2.1 to −1.0) for dapagliflozin 5 mg (Figure S3). When compared with other glucose-lowering drugs, SGLT2 inhibition effects ranged from a −4.4 kg (95% CI −4.8 to
−4.1) reduction for empagliflozin 25 mg versus sulphonylurea to −1.2 (95% CI −1.9 to −0.6) for dapagliflozin 5 mg versus metformin (Figure S3). The results of the network analysis showed a reduction of body weight compared with placebo of −2.5 kg (95% CI −2.8 to −2.1) for canagliflozin 300 mg, −2.3 kg (95% CI −2.6 to −1.9) for empagliflozin 25 mg, −2.2 kg (95% CI −2.5 to 1.9) for dapagliflozin 10 mg, −2.1 kg (95% CI −2.5 to −1.8) for empagliflozin 10 mg, −1.9 kg (95% CI −2.2 to −1.5) for canagliflozin 100 mg and −1.6 kg (95% CI −2.0 to −1.2) for dapagliflozin 5 mg (Table 2 and Figure 2). There was a statistical inconsistency for the body weight network (Table S6). Funnel plots for primary outcomes are shown in Figure S6.
Similar results were found for HbA1c, FPG and body weight in analyses restricted to studies with a similar duration of follow-up (Table S7).
Secondary Cardiometabolic Outcomes. Data for other car- diometabolic outcomes ranged from 7828 participants (18 RCTs) for total cholesterol to 17 600 participants (33 RCTs) for systolic blood pressure. Compared with placebo, net- work meta-analysis results showed a reduction of systolic (from −4.9 mmHg with canagliflozin 300 mg to −2.8 mmHg with dapagliflozin 5 mg) and diastolic (from −2.0 mmHg with canagliflozin 300 mg to −1.5 mmHg with dapagliflozin 5 mg) blood pressure for all SGLT2 inhibitors (Table 2 and Figure 2). Canagliflozin 300 mg reduced systolic blood pres- sure to a greater extent than other SGLT2 inhibitors, while no differences were found among inhibitors for diastolic blood pressure. All SGLT2 inhibitors slightly increased HDL
2016 doi:10.1111/dom.12670 7
original article DIABETES, OBESITY AND METABOLISM cholesterol levels compared with placebo (highest increase, 0.07 mmol/l for canagliflozin 300 mg and dapagliflozin 10 mg), and no differences were found among SGLT2 inhibitors. Canagliflozin at all doses reduced triglyceride levels compared with placebo and empagliflozin, while canagliflozin 300 mg increased LDL cholesterol versus placebo and all other SGLT2 inhibitors. No differences were found among SGLT2 for total cholesterol, although data were not available for all treatments (Table 2). There was no statistical inconsistency for all out- comes, although p values were ‘borderline’ for triglycerides and systolic blood pressure (Table S6).
Safety Outcomes. Data on hypoglycaemic events were available from 37 RCTs, reporting a total of 4347 participants with a hypoglycaemic event. The results of the network meta-analysis showed an increased risk of hypoglycaemia compared with placebo for canagliflozin 300 and 100 mg, with respective ORs of 1.6 (95% CI 1.3 to 1.9) and 1.5 (95% CI 1.3 to 1.8; Table 3 and Figure S7). Among SGLT2 inhibitors, canagliflozin at both doses significantly increased the risk of hypoglycaemia com- pared with dapagliflozin 10 mg (ORs 1.5) and empagliflozin 10 mg (ORs 1.4; Table 3). In a sensitivity analysis excluding studies with insulin or sulphonylurea as background therapy, canagliflozin at both doses increased the risk of hypoglycaemia compared with dapagliflozin 10 mg (ORs 1.7 to 1.9), although no significant differences were found versus placebo for all SGLT2 inhibitors (Table S8).
Based on 1959 participants with a hypoglycaemic event from all RCTs, network meta-analysis showed an increased risk of urinary tract infection for dapagliflozin 10 mg versus placebo and empagliflozin 25 mg (ORs 1.4; Table 3).
Data on genital infection were available from 37 RCTs (1285 participants reporting event). Compared with placebo, there was an increased risk of infection for all SGLT2 inhibitors, with ORs ranging from 4.2 (95% CI 2.7–6.3) for empagliflozin 10 mg to 5.9 (95% CI 4.0–8.3) for canagliflozin 300 mg. No differences were found among SGLT2 inhibitors (Table 3).
Mean ranks and ranking probabilities are graphically dis- played in Figures S5 and S8, respectively. No inconsistency was found for all three safety outcome networks (Table S6).
Data on diabetic ketoacidosis and bone fractures were reported only in three and nine studies, respectively, limiting the possibility of performing a formal analysis.
Discussion Several randomized clinical trials have investigated the efficacy and safety of SGLT2 inhibitors compared with placebo or other glucose-lowering drugs (sulphonylureas, DPP-4 inhibitors or metformin); however, to date, no direct ‘head-to-head’ tri- als comparing SGLT2 inhibitors have been reported or are ongoing, thus limiting the possibility of a direct evaluation of their comparative clinical profiles. As network meta-analysis allows indirect assessment between treatments when direct evi- dence is unavailable, we used this approach to compare SGLT2 inhibitors for multiple outcomes.
While previous network meta-analysis assessed the efficacy and safety of a single SGLT2 inhibitor [58–61] or restricted the analyses only to efficacy outcomes in patients with type 2 Ta
bl e
3. C
om pa
ri so
ns of
so di
um -g
lu co
se co
-t ra
ns po
rt er
-2 in
hi bi
to rs
w ith
re ga
rd to
sa fe
ty ou
tc om
es .
H yp
og ly
ca em
ia U
ri na
ry tr
ac ti
nf ec
ti on
P la
ce bo
P la
ce bo
C an
a 30
0 m
g 0.
64 (0
.5 4,
0. 78
) C
an a
30 0
m g
0. 88
(0 .6
9, 1.
12 )
C an
a 10
0 m
g 1.
02 (0
.8 5,
1. 22
) 0.
66 (0
.5 5,
0. 78
) C
an a
10 0
m g
0. 93
(0 .7
4, 1.
17 )
0. 82
(0 .6
3, 1.
05 )
D ap
a 10
m g
1. 48
(1 .1
7, 1.
87 )
1. 51
(1 .1
8, 1.
91 )
0. 97
(0 .8
2, 1.
15 )
D ap
a 10
m g
0. 91
(0 .6
6, 1.
24 )
0. 84
(0 .6
2, 1.
15 )
0. 74
(0 .6
0, 0.
91 )
D ap
a 5
m g
0. 96
(0 .5
9, 1.
56 )
1. 41
(0 .8
4, 2.
38 )
1. 44
(0 .8
5, 2.
43 )
0. 93
(0 .5
7, 1.
52 )
D ap
a 5
m g
1. 22
(0 .8
6, 1.
73 )
1. 11
(0 .7
2, 1.
71 )
1. 03
(0 .6
7, 1.
58 )
0. 90
(0 .6
3, 1.
30 )
E m
pa 25
m g
0. 87
(0 .5
1, 1.
48 )
0. 83
(0 .6
4, 1.
08 )
1. 23
(0 .9
4, 1.
60 )
1. 25
(0 .9
6, 1.
64 )
0. 81
(0 .6
5, 1.
00 )
E m
pa 25
m g
1. 14
(0 .7
7, 1.
71 )
1. 39
(1 .0
7, 1.
81 )
1. 26
(0 .9
4, 1.
70 )
1. 18
(0 .8
8, 1.
57 )
1. 03
(0 .8
5, 1.
26 )
E m
pa 10
m g
1. 12
(0 .8
9, 1.
40 )
0. 97
(0 .5
6, 1.
67 )
0. 93
(0 .7
0, 1.
23 )
1. 37
(1 .0
3, 1.
82 )
1. 40
(1 .0
5, 1.
86 )
0. 90
(0 .7
1, 1.
14 )
E m
pa 10
m g
0. 89
(0 .7
4, 1.
07 )
1. 02
(0 .6
8, 1.
53 )
1. 24
(0 .9
4, 1.
63 )
1. 12
(0 .8
2, 1.
53 )
1. 04
(0 .7
7, 1.
41 )
0. 92
(0 .7
4, 1.
13 )
G en
it al
in fe
ct io
n
P la
ce bo
C an
a 30
0 m
g 0.
17 (0
.1 2,
0. 25
) C
an a
10 0
m g
1. 10
(0 .9
0, 1.
34 )
0. 19
(0 .1
4, 0.
27 )
D ap
a 10
m g
0. 94
(0 .5
9, 1.
48 )
1. 03
(0 .6
6, 1.
62 )
0. 18
(0 .1
3, 0.
25 )
D ap
a 5
m g
1. 30
(0 .9
1, 1.
87 )
1. 22
(0 .7
1, 2.
09 )
1. 34
(0 .7
9, 2.
29 )
0. 23
(0 .1
5, 0.
36 )
E m
pa 25
m g
1. 01
(0 .5
7, 1.
76 )
1. 31
(0 .8
1, 2.
12 )
1. 23
(0 .7
6, 1.
98 )
1. 35
(0 .8
4, 2.
16 )
0. 24
(0 .1
5, 0.
36 )
E m
pa 10
m g
1. 02
(0 .7
8, 1.
33 )
1. 03
(0 .5
8, 1.
83 )
1. 34
(0 .8
1, 2.
21 )
1. 26
(0 .7
6, 2.
07 )
1. 38
(0 .8
5, 2.
25 )
0. 24
(0 .1
6, 0.
37 )
D at
a ar
e re
po rt
ed as
od ds
ra ti
o (9
5% co
nfi de
nc e
in te
rv al
) an
d in
di ca
te co
lu m
n- to
-r ow
ra ti
os [i
.e .c
om pa
re d
w ith
ca na
gl ifl
oz in
30 0
m g,
pl ac
eb o
is as
so ci
at ed
w ith
an od
ds ra
ti o
of 0.
17 of
ge ni
ta li
nf ec
tio n,
or eq
ui va
le nt
ly ca
na gl
ifl oz
in 30
0 m
g in
cr ea
se s
th e
ri sk
by an
od ds
ra tio
of 5.
9 (=
1/ 0.
17 )]
.S ta
ti st
ic al
ly si
gn ifi
ca nt
di ffe
re nc
es ar
e in
bo ld
. C
an a,
ca na
gl ifl
oz in
;D ap
a, da
pa gl
ifl oz
in ;E
m pa
,e m
pa gl
ifl oz
in .
8 Zaccardi et al. 2016
DIABETES, OBESITY AND METABOLISM original article diabetes inadequately controlled with diet and exercise alone or metformin monotherapy [62], we collected data for inhibitors clinically available in most countries and for indicators usually considered when choosing glucose-lowering drugs as well as for other cardiometabolic and safety outcomes to provide a comprehensive picture of these inhibitors.
When compared with placebo, all SGLT2 inhibitors improved glucose control (0.6–0.9% decrease in HbA1c and 1.1–1.9 mmol/l decrease in FPG) and reduced body weight (1.6–2.5 kg), systolic (2.8–4.9 mmHg) and diastolic (1.5–2.0 mmHg) blood pressure. Further to this, all SGLT2 inhibitors modestly increased HDL cholesterol levels com- pared with placebo (greatest increase, 0.07 mmol/l). Available evidence also suggested a small increase in LDL cholesterol and a reduction in triglycerides with both doses of canagliflozin when compared with placebo. Given the limited data available for total cholesterol, however, the effects of SGLT2 inhibitors on the overall lipid profile should be further investigated. Overall change in these cardiometabolic biomarkers would suggest, at least theoretically, a potential microvascular and cardiovas- cular benefit. The recent EMPA-REG OUTCOME trial [63] has indeed demonstrated a reduction in cardiovascular events in patients with type 2 diabetes treated with empagliflozin. Ongoing RCTs [64,65] will confirm whether similar benefits could be extended to other drugs of the same class. Conversely, although SGLT2 inhibitors should not increase the risk of hypoglycaemia as they do not stimulate insulin secretion [5], the risk was ∼50% greater for both canagliflozin doses but not different for empagliflozin and dapagliflozin when compared with placebo. Notably, the increased canagliflozin risk was nominally lower than that of metformin and significantly lower than that of sulphonylureas (approximately ninefold). Moreover, when the analysis was restricted to studies without background sulphonylureas or insulin, the risk of hypogly- caemia for all SGLT2 inhibitors was similar to placebo. This would suggest an imbalance of insulin or sulphonylurea use across studies where SGLT2 inhibitors were compared with placebo or some heterogeneity possibly attributable to study design (insulin studies are more likely to be open label and treat-to-target with no stable doses during trial). As expected, the most relevant undesirable effect of SGLT2 inhibition is an increased risk of genitourinary infection as a direct effect of glycosuria. Infections of the upper urinary tract, interestingly, were not consistently increased by SGLT2 inhibitors versus placebo (the risk was increased only by dapagliflozin 10 mg), whereas all inhibitors significantly increased the risk of genital infection (balanitis, prosthetitis, vulvovaginitis; four- to sixfold versus placebo), with no difference across inhibitors.
Along with changes versus placebo, we also found some differences among SGLT2 inhibitors. The highest dose of canagliflozin reduced HbA1c, FPG and systolic blood pres- sure to a greater extent compared with dapagliflozin and empagliflozin at any dose. Conversely, the highest doses of SGLT2 inhibitors did not differ in the extent of body weight and diastolic blood pressure reduction or HDL cholesterol increase. Whilst incomplete data on total cholesterol limited a compar- ative and overall assessment, differences among inhibitors were found for LDL cholesterol and triglycerides (with the
highest dose of canagliflozin decreasing triglycerides versus empagliflozin at any dose and increasing LDL cholesterol versus all other SGLT2 inhibitors). Among SGLT2 inhibitors, the risk of urinary tract and genital infection was similar, while at their highest doses canagliflozin increased the risk of hypoglycaemia compared with dapagliflozin, also accounting for different background therapies.
The differences observed for some clinical outcomes could in part be attributed to outcome definition, study design and/or analysis, or intrinsic pharmacological properties of individual drugs. Indeed, in addition to SGLT2, the SGLT1 receptor has also been implicated in glucose regulation [66], and each inhibitor is known to have a different receptor selectivity profile (for SGLT2 over SGLT1, >2500-fold with empagliflozin; >1200-fold with dapagliflozin; and >250-fold with canagliflozin) [67]. Our findings of a better glucose control and of an increase in LDL cholesterol by canagliflozin would therefore underline the glicometabolic relevance of SGLT1 inhibition and support recent results on dual SGLT1/SGLT2 blockade [68,69].
We should acknowledge some limitations of the present study. First, we performed a study-level meta-analysis based only on data published in journal articles or available on ClinicalTrials.gov. This could introduce a bias as such studies are more likely to report ‘positive’ findings compared with unpublished reports; however, such risk of bias should be low for RCTs. Second, in some studies, outcomes were not reported or it was not possible to extract them in a suitable way. Informa- tion was retrieved from all 38 RCTs for HbA1c and urinary tract infection and from 37 RCTs for FPG, body weight, hypogly- caemia and genital infection. Recently, several cases of diabetic ketoacidosis without frank hyperglycaemia (‘euglycaemic diabetic ketoacidosis’) have been reported in association with SGLT2 inhibitors [70,71], along with an increased risk of bone fractures with canagliflozin [72,73]. Given the limited avail- ability of data on these outcomes, we could not perform formal assessments; future ad hoc analyses and studies will clarify how such complications are drug- or class-specific. Third, across RCTs, ethnicities of participants included, follow-up durations, or outcomes selection, definition and ascertainment could to some extent differ. Yet, the majority of trials used the same clas- sification system for urinary and genital tract infection (system organ class and preferred term, MedDRA). Further to this, anal- yses for the primary outcomes including studies with similar duration evidenced consistent results. Finally, we found a sig- nificant inconsistency for the body weight network (although not present for studies with similar follow-up) and caution is needed in interpreting these results. To our knowledge, this is the first attempt to summarize available data on SGLT2 inhibitors and to assess them comparatively for a wide range of outcomes.
In conclusion, SGLT2 inhibitors improved cardiometabolic markers in patients with type 2 diabetes, with canagliflozin 300 mg generally performing better than other inhibitors; however, they increased the risk of genital infection. RCTs with direct SGLT2 comparisons would further delineate their comparative efficacy and tolerability. Moreover, given their effects on blood pressure and lipoproteins, ongoing RCTs with
2016 doi:10.1111/dom.12670 9
original article DIABETES, OBESITY AND METABOLISM cardiovascular outcomes will clarify whether changes in inter- mediate biomarkers would also translate into a reduction in relevant vascular complications confirming early positive results of this class of glucose-lowering agents [63,74].
Acknowledgements We acknowledge the support from the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care – East Midlands, the Leices- ter Clinical Trials Unit and the NIHR Leicester-Loughborough Diet, Lifestyle and Physical Activity Biomedical Research Unit, which is a partnership between University Hospitals of Leices- ter NHS Trust, Loughborough University and the University of Leicester. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.
Conflict of Interest K. K. has acted as a consultant and speaker for Novartis, Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company and Merck Sharp & Dome, has received grants in support of investigator and investigator initiated trials from Novartis, Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Pfizer, Boehringer Ingelheim and Merck Sharp & Dome, has received funds for research, honoraria for speaking at meetings and has served on advisory boards for Eli Lilly and Company, Sanofi-Aventis, Merck Sharp & Dome and Novo Nordisk. M. J. D. has acted as consultant, advisory board member and speaker for Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Merck Sharp & Dome, Boehringer Ingelheim, AstraZeneca and Janssen and as a speaker for Mitsubishi Tanabe Pharma Corpo- ration, and has received grants in support of investigator and investigator initiated trials from Novo Nordisk, Sanofi-Aventis and Eli Lilly and Company. D. R.W. has received grant in support of investigator-initiated studies and honoraria from Sanofi-Aventis and Novo Nordisk. All other authors have no conflict of interests to disclose.
F. Z. is a Clinical Research Fellow funded with an unre- stricted Educational Grant from Sanofi-Aventis to the Univer- sity of Leicester. The funding source had no involvement in this study.
F.Z. and M.J.D. were responsible for the study concept and design. F. Z., Z. Z. H. and D. Y. conducted the literature search and data extraction. F. Z. performed the data analysis. F. Z., D. R. W., Z. Z. H., D. Y., K. K., M. J. D. were responsible for the critical revision of the study and the manuscript draft. All authors provided final approval of the version to publish. Statistical codes and datasets are available from the corresponding author. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Supporting Information Additional Supporting Information may be found in the online version of this article:
Appendix S1. PRISMA checklist.
Table S1. Number of participants, by cardiometabolic out- come and study.
Table S2. Number of arms and participants, by car- diometabolic outcome and drug.
Table S3. Number of total participants and participants with event, by safety outcome and study.
Table S4. Number of arms, total participants and participants with event, by safety outcome and drug.
Table S5. Assessment of risk of bias in individual studies. Table S6. Inconsistency of networks. Table S7. Characteristics of studies with 24–30 weeks of
follow-up and comparison of SGLT2 inhibitors for the primary outcomes.
Table S8. Comparisons of SGLT2 inhibitors for hypogly- caemia excluding studies with sulphonylurea or insulin as back- ground therapy.
Figure S1. PRISMA flow diagram. Figure S2. Overall risk of bias by drug. Figure S3. Pairwise random-effects meta-analyses for primary
outcomes. Figure S4. Ranking probabilities for the primary outcomes by
drug. Figure S5. Radar plots of mean ranks for selected car-
diometabolic and safety outcomes. Figure S6. Comparison-adjusted funnel plots for primary
outcomes. Figure S7. Differences versus placebo in safety outcomes for
the drugs included in the network meta-analysis. Figure S8. Ranking probabilities for safety outcomes by drug.
References 1. Defronzo RA. Banting lecture. From the triumvirate to the ominous octet: a
new paradigm for the treatment of type 2 diabetes mellitus. Diabetes 2009; 58: 773–795.
2. Holman RR, Sourij H, Califf RM. Cardiovascular outcome trials of glucose-lowering drugs or strategies in type 2 diabetes. Lancet 2014; 383: 2008–2017.
3. Inzucchi SE, Bergenstal RM, Buse JB et al. Management of hyperglycaemia in type 2 diabetes, 2015: a patient-centred approach. Update to a position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia 2015; 58: 429–442.
4. Abdul-Ghani MA, DeFronzo RA. Lowering plasma glucose concentration by inhibiting renal sodium-glucose cotransport. J Intern Med 2014; 276: 352–363.
5. Ferrannini E, Solini A. SGLT2 inhibition in diabetes mellitus: rationale and clinical prospects. Nat Rev Endocrinol 2012; 8: 495–502.
6. Sutton A, Ades AE, Cooper N, Abrams K. Use of indirect and mixed treat- ment comparisons for technology assessment. Pharmacoeconomics 2008; 26: 753–767.
7. Jansen JP, Trikalinos T, Cappelleri JC et al. Indirect treatment compari- son/network meta-analysis study questionnaire to assess relevance and credibility to inform health care decision making: an ISPOR-AMCP-NPC Good Practice Task Force report. Value Health 2014; 17: 157–173.
8. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol 2009; 62: 1006–1012.
9. Hoaglin DC, Hawkins N, Jansen JP et al. Conducting indirect-treatment- comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2. Value Health 2011; 14: 429–437.
10 Zaccardi et al. 2016
DIABETES, OBESITY AND METABOLISM original article 10. Hutton B, Salanti G, Caldwell DM et al. The PRISMA extension statement for
reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations. Ann Intern Med 2015; 162: 777–784.
11. Higgins JP, Altman DG, Gøtzsche PC et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011; 343: d5928.
12. Harris R, Bradburn M, Deeks J, Harbord R, Altman D, Sterne J. Metan: fixed- and random-effects meta-analysis. Stata J 2008; 8: 3–28.
13. Multiple-Treatments Meta-Analysis. Using STATA for network meta-analysis. Available from URL: http://www.mtm.uoi.gr/index.php/stata-routines-for- network-meta-analysis. Accessed 15 February 2016.
14. White IR. Multivariate random-effects meta-regression: updates to mvmeta. Stata J 2011; 11: 255–270.
15. White IR, Barrett JK, Jackson D, Higgins JP. Consistency and inconsistency in network meta-analysis: model estimation using multivariate meta-regression. Res Synth Methods 2012; 3: 111–125.
16. Dias S, Welton NJ, Sutton AJ, Ades AE. NICE Decision Support Unit technical support document 2: a generalised linear model framework for pairwise and network meta-analysis of randomised clinical trials. 2014. Available from URL: http://www.nicedsu.org.uk/TSD2%20General%20meta%20analysis%20 corrected%2015April2014.pdf. Accessed 15 February 2016.
17. Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network meta-analysis in STATA. PLoS One 2013; 8: e76654.
18. Salanti G, Ades AE, Ioannidis JP. Graphical methods and numerical summaries for presenting results from multiple-treatment meta-analysis: an overview and tutorial. J Clin Epidemiol 2011; 64: 163–171.
19. Jackson D, Barrett JK, Rice S, White IR, Higgins JP. A design-by-treatment inter- action model for network meta-analysis with random inconsistency effects. Stat Med 2014; 33: 3639–3654.
20. Araki E, Tanizawa Y, Tanaka Y et al. Long-term treatment with empagliflozin as add-on to oral antidiabetes therapy in Japanese patients with type 2 diabetes mellitus. Diabetes Obes Metab 2015; 17: 665–674.
21. Bailey CJ, Iqbal N, T’Joen C, List JF. Dapagliflozin monotherapy in drug-naive patients with diabetes: a randomized-controlled trial of low-dose range. Dia- betes Obes Metab 2012; 14: 951–959.
22. Bailey CJ, Gross JL, Hennicken D, Iqbal N, Mansfield TA, List JF. Dapagliflozin add-on to metformin in type 2 diabetes inadequately controlled with met- formin: a randomized, double-blind, placebo-controlled 102-week trial. BMC Med 2013; 11: 43.
23. Bailey CJ, Morales Villegas EC, Woo V et al. Efficacy and safety of dapagliflozin monotherapy in people with Type 2 diabetes: a randomized double-blind placebo-controlled 102-week trial. Diabet Med 2015; 32: 531–541.
24. Bode B, Stenlöf K, Harris S et al. Long-term efficacy and safety of canagliflozin over 104 weeks in patients aged 55–80 years with type 2 diabetes. Diabetes Obes Metab 2015; 17: 294–303.
25. Bolinder J, Ljunggren Ö, Johansson L et al. Dapagliflozin maintains glycaemic control while reducing weight and body fat mass over 2 years in patients with type 2 diabetes mellitus inadequately controlled on metformin. Diabetes Obes Metab 2014; 16: 159–169.
26. Cefalu WT, Leiter LA, de Bruin TW, Gause-Nilsson I, Sugg J, Parikh SJ. Dapagliflozin’s effects on glycemia and cardiovascular risk factors in high-risk patients with type 2 diabetes: a 24-week, multicenter, randomized, double-blind, placebo-controlled study with a 28-week extension. Diabetes Care 2015; 38: 1218–1227.
27. DeFronzo RA, Lewin A, Patel S et al. Combination of empagliflozin and linagliptin as second-line therapy in subjects with type 2 diabetes inadequately controlled on metformin. Diabetes Care 2015; 38: 384–393.
28. Del Prato S, Nauck M, Durán-Garcia S et al. Long-term glycaemic response and tolerability of dapagliflozin versus a sulphonylurea as add-on therapy to metformin in patients with type 2 diabetes: 4-year data. Diabetes Obes Metab 2015; 17: 581–590.
29. Ferrannini E, Berk A, Hantel S et al. Long-term safety and efficacy of empagliflozin, sitagliptin, and metformin: an active-controlled, parallel-group, randomized, 78-week open-label extension study in patients with type 2 diabetes. Diabetes Care 2013; 36: 4015–4021.
30. Forst T, Guthrie R, Goldenberg R et al. Efficacy and safety of canagliflozin over 52 weeks in patients with type 2 diabetes on background metformin and pioglitazone. Diabetes Obes Metab 2014; 16: 467–477.
31. Haering HU, Merker L, Christiansen AV et al. Empagliflozin as add-on to met- formin plus sulphonylurea in patients with type 2 diabetes. Diabetes Res Clin Pract 2015; 110: 82–90.
32. 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: 446–456.
33. Inagaki N, Kondo K, Yoshinari T, Takahashi N, Susuta Y, Kuki H. Efficacy and safety of canagliflozin monotherapy in Japanese patients with type 2 dia- betes inadequately controlled with diet and exercise: a 24-week, randomized, double-blind, placebo-controlled, phase III study. Expert Opin Pharmacother 2014; 15: 1501–1515.
34. Jabbour SA, Hardy E, Sugg J, Parikh S. Dapagliflozin is effective as add-on therapy to sitagliptin with or without metformin: a 24-week, multicenter, randomized, double-blind, placebo-controlled study. Diabetes Care 2014; 37: 740–750.
35. Ji L, Ma J, Li H et al. Dapagliflozin as monotherapy in drug-naive Asian patients with type 2 diabetes mellitus: a randomized, blinded, prospective phase III study. Clin Ther 2014; 36: 84–100.
36. Kadowaki T, Haneda M, Inagaki N et al. Efficacy and safety of empagliflozin monotherapy for 52 weeks in Japanese patients with type 2 diabetes: a randomized, double-blind, parallel-group study. Adv Ther 2015; 32: 306–318.
37. Kaku K, Kiyosue A, Inoue S et al. Efficacy and safety of dapagliflozin monother- apy in Japanese patients with type 2 diabetes inadequately controlled by diet and exercise. Diabetes Obes Metab 2014; 16: 1102–1110.
38. Kovacs CS, Seshiah V, Merker L et al. Empagliflozin as add-on therapy to pioglitazone with or without metformin in patients with type 2 diabetes mellitus. Clin Ther 2015; 37: 1773–1788.
39. Lavalle-González FJ, Januszewicz A, Davidson J et al. Efficacy and safety of canagliflozin compared with placebo and sitagliptin in patients with type 2 dia- betes on background metformin monotherapy: a randomised trial. Diabetologia 2013; 56: 2582–2592.
40. Leiter LA, Cefalu WT, de Bruin TW, Gause-Nilsson I, Sugg J, Parikh SJ. Dapagliflozin added to usual care in individuals with type 2 diabetes mellitus with preexisting cardiovascular disease: a 24-week, multicenter, randomized, double-blind, placebo-controlled study with a 28-week extension. J Am Geriatr Soc 2014; 62: 1252–1262.
41. Leiter LA, Yoon KH, Arias P et al. Canagliflozin provides durable glycemic improvements and body weight reduction over 104 weeks versus glimepiride in patients with type 2 diabetes on metformin: a randomized, double-blind, phase 3 study. Diabetes Care 2015; 38: 355–364.
42. Lewin A, DeFronzo RA, Patel S et al. Initial combination of empagliflozin and linagliptin in subjects with type 2 diabetes. Diabetes Care 2015; 38: 394–402.
43. Matthaei S, Bowering K, Rohwedder K, Sugg J, Parikh S, Johnsson E. Durability and tolerability of dapagliflozin over 52 weeks as add-on to metformin and sulphonylurea in type 2 diabetes. Diabetes Obes Metab 2015; 17: 1075–1084.
44. Merker L, Häring HU, Christiansen AV et al. Empagliflozin as add-on to met- formin in people with type 2 diabetes. Diabet Med 2015; 32: 1555–1567.
45. Neal B, Perkovic V, de Zeeuw D et al. Efficacy and safety of canagliflozin, an inhibitor of sodium-glucose cotransporter 2, when used in conjunction with insulin therapy in patients with type 2 diabetes. Diabetes Care 2015; 38: 403–411.
46. Ridderstrale M, Andersen KR, Zeller C, Kim G, Woerle HJ, Broedl UC. Comparison of empagliflozin and glimepiride as add-on to metformin in patients with type
2016 doi:10.1111/dom.12670 11
original article DIABETES, OBESITY AND METABOLISM 2 diabetes: a 104-week randomised, active-controlled, double-blind, phase 3 trial. Lancet Diabetes Endocrinol 2014; 2: 691–700.
47. Roden M, Weng J, Eilbracht J et al. Empagliflozin monotherapy with sitagliptin as an active comparator in patients with type 2 diabetes: a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet Diabetes Endocrinol 2013; 1: 208–219.
48. Rosenstock J, Vico M, Wei L, Salsali A, List JF. Effects of dapagliflozin, an SGLT2 inhibitor, on HbA(1c), body weight, and hypoglycaemia risk in patients with type 2 diabetes inadequately controlled on pioglitazone monotherapy. Diabetes Care 2012; 35: 1473–1478.
49. Rosenstock J, Jelaska A, Frappin G et al. Improved glucose control with weight loss, lower insulin doses, and no increased hypoglycaemia with empagliflozin added to titrated multiple daily injections of insulin in obese inadequately controlled type 2 diabetes. Diabetes Care 2014; 37: 1815–1823.
50. Rosenstock J, Jelaska A, Zeller C, Kim G, Broedl UC, Woerle HJ. Impact of empagliflozin added-on to basal insulin in type 2 diabetes inade- quately controlled on basal insulin: a 78-week randomized, double-blind, placebo-controlled trial. Diabetes Obes Metab 2015; 17: 936–948.
51. Rosenstock J, Hansen L, Zee P et al. Dual add-on therapy in type 2 diabetes poorly controlled with metformin monotherapy: a randomized double-blind trial of saxagliptin plus dapagliflozin addition versus single addition of saxagliptin or dapagliflozin to metformin. Diabetes Care 2015; 38: 376–383.
52. 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: 2508–2515.
53. Stenlöf K, Cefalu WT, Kim KA et al. Efficacy and safety of canagliflozin monotherapy in subjects with type 2 diabetes mellitus inadequately controlled with diet and exercise. Diabetes Obes Metab 2013; 15: 372–382.
54. Strojek K, Yoon KH, Hruba V, Sugg J, Langkilde AM, Parikh S. Dapagliflozin added to glimepiride in patients with type 2 diabetes mellitus sustains glycemic con- trol and weight loss over 48 weeks: a randomized, double-blind, parallel-group, placebo-controlled trial. Diabetes Ther 2014; 5: 267–283.
55. Wilding JP, Charpentier G, Hollander P et al. Efficacy and safety of canagliflozin in patients with type 2 diabetes mellitus inadequately controlled with met- formin and sulphonylurea: a randomised trial. Int J Clin Pract 2013; 67: 1267–1282.
56. Wilding JP, Woo V, Rohwedder K, Sugg J, Parikh S. Dapagliflozin in patients with type 2 diabetes receiving high doses of insulin: efficacy and safety over 2 years. Diabetes Obes Metab 2014; 16: 124–136.
57. Mathieu C, Ranetti AE, Li D et al. Randomized, double-blind, phase 3 trial of triple therapy with dapagliflozin add-on to saxagliptin plus metformin in type 2 diabetes. Diabetes Care 2015; 38: 2009–2017.
58. Goring S, Hawkins N, Wygant G et al. Dapagliflozin compared with other oral anti-diabetes treatments when added to metformin monotherapy: a systematic review and network meta-analysis. Diabetes Obes Metab 2014; 16: 433–442.
59. Taieb V, Pacou M, Schroeder M, Schubert A, Nielsen AT. Network meta-analysis (NMA) to assess relative efficacy measured as percentage of patients treated to HbA1c target with canagliflozin in patients with type 2 diabetes mellitus
(T2DM) inadequately controlled on metformin and sulphonylurea (MET + SU). Value Health 2015; 18: A598.
60. Taieb V, Pacou M, Schroeder M, Nielsen AT, Neslusan C, Schubert A. Bayesian network meta-analysis (NMA) to assess the relative efficacy of canagliflozin in patients with type 2 diabetes mellitus (T2DM) inadequately controlled with insulin. Value Health 2015; 18: A598.
61. Taieb V, Pacou M, Schroeder M, Nielsen AT, Schubert A, Neslusan C. A network meta-analysis (NMA) to assess the longer-term relative efficacy of canagliflozin in patients with type 2 diabetes inadequately controlled on metformin. Value Health 2015; 18: A598.
62. Shyangdan DS, Uthman OA, Waugh N. SGLT-2 receptor inhibitors for treating patients with type 2 diabetes mellitus: a systematic review and network meta-analysis. BMJ Open 2016; 6: e009417.
63. Zinman B, Wanner C, Lachin JM et al. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med 2015; 373: 2117–2128.
64. Neal B, Perkovic V, de Zeeuw D et al. Rationale, design, and baseline char- acteristics of the Canagliflozin Cardiovascular Assessment Study (CANVAS)–a randomized placebo-controlled trial. Am Heart J 2013; 166: 217–223.
65. ClinicalTrials.gov. Available from URL: https://clinicaltrials.gov/ct2/show/ NCT01730534. Accessed 15 February 2016.
66. Wright EM, Hirayama BA, Loo DF. Active sugar transport in health and disease. J Intern Med 2007; 261: 32–43.
67. Grempler R, Thomas L, Eckhardt M et al. Empagliflozin, a novel selective sodium glucose cotransporter-2 (SGLT-2) inhibitor: characterisation and comparison with other SGLT-2 inhibitors. Diabetes Obes Metab 2012; 14: 83–90.
68. Zambrowicz B, Freiman J, Brown PM et al. LX4211, a dual SGLT1/SGLT2 inhibitor, improved glycemic control in patients with type 2 diabetes in a randomized, placebo-controlled trial. Clin Pharmacol Ther 2012; 92: 158–169.
69. Rosenstock J, Cefalu WT, Lapuerta P et al. Greater dose-ranging effects on A1C levels than on glucosuria with LX4211, a dual inhibitor of SGLT1 and SGLT2, in patients with type 2 diabetes on metformin monotherapy. Diabetes Care 2015; 38: 431–438.
70. Peters AL, Buschur EO, Buse JB, Cohan P, Diner JC, Hirsch IB. Euglycemic dia- betic ketoacidosis: a potential complication of treatment with sodium-glucose cotransporter 2 inhibition. Diabetes Care 2015; 38: 1687–1693.
71. Erondu N, Desai M, Ways K, Meininger G. Diabetic ketoacidosis and related events in the canagliflozin type 2 diabetes clinical program. Diabetes Care 2015; 38: 1680–1686.
72. Food and Drug Administration. Available from URL: http://www.fda.gov/ downloads/Drugs/DrugSafety/UCM461790.pdf. Accessed 15 February 2016.
73. Watts NB, Bilezikian JP, Usiskin K et al. Effects of canagliflozin on fracture risk in patients with type 2 diabetes mellitus. J Clin Endocrinol Metab 2016; 101: 157–166.
74. Wu JHY, Foote C, Blomster J et al. Effects of sodium-glucose cotransporter-2 inhibitors on cardiovascular events, death, and major safety outcomes in adults with type 2 diabetes: a systematic review and meta-analysis. Lancet Diabetes Endocrinol 2016 [Epub before print]. DOI: 10.1016/S2213-8587(16)00052-8.
12 Zaccardi et al. 2016