diabetes research and clinical practice 1 3 0 (2 0 17 ) 5 3–60

Contents available at ScienceDirect

Diabetes Research
and Clinical Practice
journal homepage: www.elsevier.com/locat e/dia bre s

Effects of SGLT-2 inhibitors on diabetic ketoacidosis:

A meta-analysis of randomised controlled trials

Matteo Monami a,*, Besmir Nreu b, Stefania Zannoni a, Carlotta Lualdi a,

Edoardo Mannucci a
a
Diabetology, University of Florence and Careggi Teaching Hospital, Florence, Italy
b
Geriatric Medicine, University of Florence and Careggi Teaching Hospital, Florence, Italy

A R T I C L E I N F O A B S T R A C T

Article history: Aims: Diabetic ketoacidosis (DKA) associated with SGLT-2 inhibitors (SGLT-2i) is a possible
Received 28 February 2017 adverse event. In fact, SGLT-2i are capable of stimulating the release of glucagon and ketone
Received in revised form re-absorption in the renal tubuli, thus increasing the concentration of ketone bodies.
7 April 2017 Methods: A Medline search for SGLT2i (dapagliflozin, empagliflozin, canagliflozin, ipragliflo-
Accepted 19 April 2017 zin, ertugliflozin, luseogliflozin) was performed, collecting all randomized trials with a
Available online 18 May 2017 duration of treatment  12 weeks, enrolling patients with type 2 diabetes, and comparing
a SGLT2i with placebo or other comparators. The principal outcome was the effect of
SGLT2i on ketoacidosis as serious adverse event.
Keywords:
Results: Out of 72 trials reporting information on DKA, 9 reported at least one event of
Type 2 diabetes
ketoacidosis; those eight trials enrolled 10,157 and 5396 in SGLT-2 inhibitors and compara-
Meta-analysis
tor groups, respectively. No signal of increased risk for ketoacidosis was observed for SGLT2
SGLT2 inhibitor
inhibitors as a class (MH-OR [95% CI] 1.14 [0.45–2.88], p = 0.78) or as individual molecule. The
Ketoacidosis
sensitivity analysis with continuity correction (inputing one event each in drug and com-
Serious adverse events
parator arms of each trial with zero events) suggested a reduced incidence of ketoacidosis
Randomized clinical trials
in patients treated with SGLT-2 inhibitors (MH-OR 0.65 [0.47–0.90]; p = 0.01).
Conclusions: The results of clinical trials summarized in the present meta-analysis reassure
us that, when the drug is properly prescribed, the risk of DKA is negligible.
Ó 2017 Elsevier B.V. All rights reserved.

1. Introduction available cardiovascular outcome study with a drug of this

class, empagliflozin, revealed a significant reduction in the
Sodium-Glucose Trasporter-2 (SGLT-2) inhibitors have been incidence of major cardiovascular events, cardiovascular
recently introduced in the treatment of type 2 diabetes as and all-cause mortality, and hospitalization for heart failure
glucose-lowering agents. In patients with preserved renal [3].
function, they are able to reduce blood glucose and HbA1c A correct assessment of the safety profile is of paramount
without inducing hypoglycaemia, unless combined with insu- importance for any novel drug. The main side effect of SGLT-2
lin or sulfonylureas. In addition, SGLT-2 inhibitors determine inhibitors is an increase in the risk of genital infections,
weight loss and a reduction of blood pressure [1,2]. The first which are generally mild, and more frequent in women with

* Corresponding author at: Diabetology, Azienda Ospedaliero-Universitaria Careggi, Via delle Oblate 4, 50141 Florence, Italy. Fax: +39
055 7949660.
E-mail address: [email protected] (M. Monami).
http://dx.doi.org/10.1016/j.diabres.2017.04.017
0168-8227/Ó 2017 Elsevier B.V. All rights reserved.
54 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0

a history of chronic or recurrent genital infections [1]. A mod- trial. For unpublished information, data were retrieved (in this
estly increased risk of urinary tract infections has also been hierarchical order) from FDA Medical Reviews, EMA public
observed in clinical trials [4], together with relatively rare assessment report, www.clinicaltrials.gov study results, and
events of volume depletion [4]. a company proprietor website (www.astellasclinicalstudyre-
Some further concerns were generated by reports of ketoaci- sults.com/hcp/Clinicalstudyresult.aspx). Retrieved data
dosis associated with SGLT-2 inhibitors [5,6]. These drugs stim- included ketoacidosis reported as serious adverse event, and
ulate the release of glucagon, thus increasing the production of baseline characteristics of enrolled patients (age, BMI, dura-
ketone bodies [7]. Furthermore, the inhibition of SGLT-2 stimu- tion of diabetes, HbA1c). The information on ketoacidosis
lates ketone re-absorption in the renal tubuli [8]. As a conse- was considered missing when serious adverse events were
quence, treatment with SGLT-2 inhibitors could be associated, not individually listed in any of the sources cited above. The
in some instances, with increased ketonemia, leading to acido- quality of trials was assessed using the Cochrane Collabora-
sis. Cases of ketosis without severe hyperglycemia have been tion’s Tool for Assessing Risk of Bias in randomized controlled
reported during treatment with SGLT-2 inhibitors [5], inducing trials; quality was not used as a criterion for the selection of
regulatory authorities to issue a warning [6]. trials, but only for descriptive purposes.
Relatively rare adverse events are difficult to discriminate
in individual trials, when the statistical power is generally 2.4. Data synthesis and analysis
insufficient. In these cases, meta-analyses of trials, capturing
all the available information, can add relevant insight. The The outcome of this meta-analysis was the effect of SGLT2
present study is aimed at providing a comprehensive and inhibitors on ketoacidosis. Heterogeneity was assessed by
updated summary of all evidence on this issue available from using I2 statistics. In order to estimate possible publication/
randomized clinical trials. disclosure bias we used funnel plots, including published
and unpublished, but disclosed, trials. Considering the differ-
2. Subjects, materials and methods ences across trials in molecules, treatment schedules, inclu-
sion criteria, and length of follow-up, a random-effects
This meta-analysis is part of a wider review on SGLT-2 inhibi- model was applied, calculating Mantel-Haenszel odds ratio
tors, the protocol of which (CRD42016046623) was published with 95% Confidence Interval (MH-OR) on an intention-to-
on the http://www.crd.york.ac.uk/PROSPERO website. treat basis, excluding trials with zero events. A sensitivity
analysis was performed with continuity correction, in order
2.1. Data sources and searches to avoid distortions due to the exclusion of trials with zero
events, inputing one event each in the investigational drug
A Medline search for SGLT2 inhibitors (dapagliflozin, empagli- and comparator arms of each trial with zero events. A further
flozin, canagliflozin, ipragliflozin, ertugliflozin, luseogliflozin, sensitivity analysis was performed using Peto’s Odds Ratio.
and tofogliflozin) was performed, collecting all randomized Subgroup analyses were performed for all endpoints for dif-
trials up to August 10th, 2016. The identification of relevant ferent drugs of the class, different classes of comparators, tri-
abstracts, the selection of studies based on the criteria als with cardiovascular and non-cardiovascular endpoints. All
described below, and the subsequent data extraction were analyses were performed using Comprehensive Meta-
performed independently by two of the authors (N.B., S.Z.), analysis Version 2, Biostat, (Englewood, NJ, USA). The meta-
and conflicts resolved by a third investigator (E.M.). Com- analysis was reported following the PRISMA checklist [9].
pleted but still unpublished trials were identified through a
search of www.clinicaltrials.gov website, using the same key- 3. Results
words. In addition, Medical Reviews were retrieved from the
Food and Drug Administration (FDA) website, and the Sum- Out of 181, 278, and 155 items identified through MEDLINE,
mary of Product Characteristics from the European Medicines www.clinicaltrials.gov, and FDA/EMA websites, respectively,
Agency (EMA) website, for the identification of further unpub- 80 trials fulfilled inclusion criteria. Of those, seventy-two,
lished and otherwise undisclosed trials. enrolling 27,455 and 15,867 patients in SGLT2 inhibitor and
comparator arms, respectively, with a mean duration of treat-
2.2. Study selection ment of 36.6 weeks, reported information on ketoacidosis
(Fig. 1). The characteristics of those 80 trials, are reported in
The present meta-analysis was performed including all trials Table 1 and Table 2 (Supplementary material). The search of
with a treatment duration >12 weeks, enrolling patients with www.clinicaltrials.gov website allowed the identification of 6
type 2 diabetes, and comparing a SGLT2 inhibitor with pla- unpublished and undisclosed, although completed, trials
cebo or active comparators, considering only molecules and (Table 3 of supplementary material).
doses approved by either FDA, EMA, or Japanese Pharmaceu- Out of 72 trials, 9 [3,10–16] reported at least one event of
ticals and Medical Devices Agency. ketoacidosis; those nine trials enrolled 10,157 and 5,396 in
SGLT-2 inhibitors and comparator groups, respectively. The
2.3. Data extraction and quality assessment number of events of diabetic ketoacidosis was 16 for SGLT2
inhibitors and 6 for comparators. Funnel plot (Fig. 2) did not
Results of trials were retrieved from the primary publication suggest any relevant disclosure bias. I2 was 0.00, with
and, if needed, from other publications referring to the same p = 0.67, suggesting no heterogeneity.
 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0 55

Pubmed FDA/EMA www.clinical trials.gov

(n=181) (n=148) (n=278)

Not RCT Not RCT Already published

(n= 16) (n= 0) (n= 87)

Not human Published Not RCT

(n= 0) (n= 50) (n= 19)

Not external Not type 2 diabetes Not external

comparison (n= 78) comparison
(n= 3) (n= 12)
Short duraon
Not type 2 diabetes (n= 20) Not type 2 diabetes
(n= 30) (n= 125)

Short duraon Trials included Short duraon

(n= 26) n= 0 (n= 22)

Duplicate Undisclosed
(n= 32) (n= 6)

Trials included TOTAL Trials included

n= 74 n= 81 n= 7

Fig. 1 – Trial flow summary.

No signal of increased risk for ketoacidosis was observed new drugs could miss some relevant clinical issues for either
for SGLT2 inhibitors as a class (Fig. 1); similar results were insufficient sample size or relatively short duration of studies.
obtained with Peto’s Odds Ratio (1.33 [0.50–3.58]; p = 0.57). In the case of drugs indicated for diabetes, FDA requires pre-
The sensitivity analysis with continuity correction suggested approval or post-marketing randomized trials for the assess-
a reduced incidence of ketoacidosis in patients treated with ment of cardiovascular safety [17]; these studies are a pre-
SGLT-2 inhibitors (MH-OR 0.65 [0.47–0.90]; p = 0.01). cious source of additional information also for non-
When trials with non-cardiovascular endpoints were anal- cardiovascular adverse events, because of the high number
ysed separately, the MH-OR was 0.68 [0.21–2.21]; p = 0.52, of patients randomized in these trials and the longer follow-
whereas MH-OR for the two trials with cardiovascular end- up.
point was 2.67 [0.59–12.17], p = 0.20. The mechanism of action of SGLT-2 inhibitors, which
Eight trials with at least one event of ketoacidosis com- reduce available glucose for glycolysis, stimulates gluconeo-
pared SGLT-2 inhibitors with placebo, whereas two used genesis, increase glucagon levels and lipolysis, and inhibit
DPP4 inhibitors as comparators. In subgroup analyses, differ- ketone bodies urinary elimination, is compatible with an
ences from individual comparators did not reach statistical increased risk of diabetic ketoacidosis (DKA) [18]. The warning
significance (Fig. 3). issued by FDA and EMA on the risk of ketoacidosis with SGLT-
A separate analysis was also performed for each SGLT-2 2 inhibitors is based on case reports [6] and mechanistic con-
inhibitor, versus any comparator (Fig. 4). None of the mole- siderations, rather than on results of clinical trials. A previous
cules, separately considered, was associated with a signifi- meta-analysis on a smaller number of trials failed to detect
cant increase in the risk of ketoacidosis; conversely, any significant increase in risk [4]. More recently, another
empagliflozin (MH-OR 0.48 [0.25;0.94], p = 0.032), but not cana- meta-analysis suggested a possible increase in risk (with an
gliflozin (MH-OR: 0.63 [0.32;1.26], p = 0.19), was associated estimated odds ratio of 1.71), although difference from com-
with a significant reduction of risk in the analysis with conti- parators did not reach statistical significance [19]. This latter
nuity correction, when excluding cardiovascular outcome analysis did not include some recently disclosed studies
studies. This latter analysis was performed because cardio- (including interim results from CANVAS [20]). In addition, cur-
vascular outcome trials, which enrol patients with different rently available meta-analyses are affected by a possible dis-
characteristics, are not available for all the drugs of the class, tortion determined by a large number of trials with zero
thus producing a marked heterogeneity in patient samples events; since the number of patients allocated to SGLT-2 inhi-
across drugs of the class. bitors in those studies is considerably greater than that of
patients in comparator groups, the overall analysis may pro-
4. Discussion duce a bias against the investigational drugs.
We confirm here the same result on a larger sample,
The occurrence of unexpected adverse events is a possibility including some newer publications and unpublished, but
for any novel drug, particularly when those events are rare, otherwise disclosed, trial results. Estimates obtained with dif-
or when they develop after prolonged drug exposure. In fact, ferent statistical approaches (i.e., Mantel-Haenzel and Peto’s
standard programs of clinical trials used for registration of Odds Ratios) provided very similar results. Notably, this result
is confirmed when the analysis is restricted to trials with FDA/
56 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0

EMA-approved drugs and doses, i.e. excluding treatment small, with a majority of studies reporting zero events. In
arms with low drug doses in phase II trials, in which the prob- these conditions, the apparent protective effect of some
ability of toxic effects of investigational drugs is modest. The SGLT-2 inhibitors with respect to DKA in the sensitivity anal-
overall number of events of DKA in controlled trials is very ysis can easily be interpreted as an artefact, considering that

Table 1 – Baseline characteristics and principal outcome of trials included in the meta-analysis. For references see
Supplementary Materials.

First author (year)a Comparator Dose SGLT2i Trial duration Number of patients Diabetic ketoacidosis
Drug mg/day weeks SGLT2i Comp. SGLT2i Comp.

Canagliflozin
Rosenstock 20121 Placebo 50–600 12 321 65 0 0
Sitagliptin 321 65 0 0
Inagaki 20132 Placebo 50–300 12 307 75 0 0
Sha 20143 Placebo 300 12 18 18 0 0
Ji 20155 Placebo 100–300 18 445 224 0 0
NCT010326296 Placebo 100–300 18 1382 690 7 1
Inagaki 20147 Placebo 100–300 24 179 93 0 0
Forst 20148 Placebo 100–300 26 227 115 0 0
Stenlof 20149 Placebo 100–300 26 392 192 0 0
Wilding 201310 Placebo 100–300 52 313 156 1 0
Yale 201311 Placebo 100–300 52 179 90 0 0
Lavalle-Gonzalez 201312 Placebo 100–300 52 735 183 0 0
Sitagliptin 735 366 0 1
Bode 201513 Placebo 100–300 104 477 237 1 0
Schernthaner 201314 Sitagliptin 300 52 377 378 0 0
Cefalu 201315 Glimepiride 100–300 52 952 473 0 0
Inagaki 201616 Placebo 100 16 186 93 0 0
Rodbard 201617 Placebo 100 16 76 70 0 0
Rosenstock 201618 Placebo 100–300 26 474 237 0 0
Metformin 475 237 0 0
Dapagliflozin
Kaku 201318 Placebo 1–10 12 225 54 0 0
List 200919 Placebo 2.5–50 12 279 54 NR NR
Wilding 200920 Placebo 10–20 12 48 23 0 0
Heerspink 201321 Placebo 10 12 24 25 0 0
Weber 201522 Placebo 10 12 225 224 0 0
Mudaliar 201423 Placebo 5 12 23 21 0 0
NCT0113747424 Placebo 2.5–10 12 302 311 0 0
Schumm-Draeger 201525 Placebo 5–10 16 299 101 0 0
Henry 2012 (Study 1)26 Placebo 5 24 194 201 NR NR
Metformin 203 201 NR NR
Henry 2012 (Study 2)26 Placebo 10 24 211 208 NR NR
Metformin 219 208 NR NR
Rosenstock 201527 Placebo 10 24 179 176 0 0
Saxagliptin 179 176 0 0
Bailey 201228 Placebo 2.5–10 24 214 68 0 0
Jabbour 201429 Placebo 10 24 223 224 0 0
Mathieu 201530 Placebo 10 24 160 160 0 0
Kaku 201431 Placebo 5–10 24 174 87 0 0
Strojek 201132 Placebo 2,5–10 24 447 145 0 0
Cefalu 201533 Placebo 10 24 455 459 0 0
Ji 201434 Placebo 5–10 24 261 132 NR NR
Yang 201535 Placebo 5–10 24 299 145 0 0
Rosenstock 201236 Placebo 5–10 48 281 139 0 0
Matthaei 201537 Placebo 10 52 109 109 0 0
Leiter 201438 Placebo 10 52 480 482 0 0
Bolinder 201439 Placebo 10 102 89 41 0 0
Bailey 201540 Placebo 2.5–10 102 199 75 0 0
Bailey 201341 Placebo 5–10 102 272 137 0 0
Kohan 201442 Placebo 5–10 104 168 84 NR NR
Wilding 201443 Placebo 2,5–10 104 607 193 0 0
List 200944 Metformin 2.5–50 12 279 56 NR NR
Nauck 201445 Glipizide 10 104 400 401 0 0
Araki 201646 Placebo 5 26 122 60 0 0
 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0 57

Table 1 – (Continued).
First author (year)a Comparator Dose SGLT2i Trial duration Number of patients Diabetic ketoacidosis

Drug mg/day weeks SGLT2i Comp. SGLT2i Comp.

Empagliflozin
Rosenstock 201347 Placebo 1–50 12 353 71 0 0
Sitagliptin 353 71 0 0
Ferrannini 201348 Placebo 5–25 12 244 82 0 0
Metformin 244 80 0 0
Kadowaki 201549 Placebo 5–50 12 438 109 0 0
Tikkanen 201550 Placebo 10–25 12 552 271 0 1
NCT0164929751 Placebo 10–25 16 858 107 0 0
Häring 201352 Placebo 10–25 24 441 225 0 0
Roden 201553 Placebo 10–25 76 448 228 2 0
Sitagliptin 448 223 2 0
Barnett 201454 Placebo 10–25 24 419 319 0 1
Rosenstock 201455 Placebo 10–25 52 375 188 1 1
DeFronzo 201556 Placebo 10–25 52 273 132 0 0
Metformin 281 132 0 0
Lewin 201557 Placebo 10–25 52 273 135 0 0
Linagliptin 269 135 0 0
Kovacs 201458 Placebo 10–25 76 333 165 0 0
Merker 201559 Placebo 10–25 76 431 206 0 0
Rosenstock 201560 Placebo 10–25 78 324 170 0 0
Zinman 201561 Placebo 10–25 161 4687 2333 4 1
Araki 201562 Metformin 10–25 52 273 63 0 0
Ridderstråle 201463 Glimepiride 25 104 769 780 0 0
Hadjadi 201664 Placebo 10–25 24 680 341 0 0
Metformin 10–25 339 341 0 0
Ipragliflozin
Wilding 201365 Placebo 12.5–300 12 276 66 0 0
Fonseca 201366 Placebo 12.5–300 12 273 69 NR NR
Metformin 273 69 NR NR
Kashiwagi 201567 Placebo 50 24 112 56 0 0
Kashiwagi 201568 Placebo 50 24 118 46 NR NR
Kashiwagi 201469 Placebo 12.5–100 24 291 69 NR NR
NCT0105762870 Placebo 50 16 67 62 0 0
NCT0122508171 Placebo 50 24 98 54 0 0
NCT0124221572 Placebo 50 24 166 77 0 0
NCT0151483873 Acarbose 50 24 27 19 0 0
Lu 201674 Placebo 50 24 87 83 0 0
Ishiara 201675 Placebo 50 16 175 87 0 0
Luseogliflozin
Haneda 201676 Placebo 2.5 24 95 50 NR NR
Seino 201477 Placebo 0.5–5 12 122 56 0 0
Seino 201478 Placebo 2.5 24 76 72 0 0
Tofogliflozin
Ikeda 201579 Placebo 20 12 66 15 NR NR
Kaku 201480 Placebo 20 24 56 27 NR NR
a
Data for myocardial infarction and stroke available only at 24 weeks and ** 26 weeks. SGLT2i: Sodium Glucose Transporter-2 inhibitors;
Comp.: Comparator; NR: Not reported.

the number of patients randomized to SGLT-2 inhibitors is patients included in clinical trials are not representative of
considerably greater than that allocated to control groups. those actually using the drug in routine clinical practice,
DKA, if detected, is very unlikely to be classified as ‘‘non- who may include some populations at higher risk of DKA.
serious”; therefore, the possibility of missing some cases in Notably, in at least some of the reported cases, patients with
trials listing serious adverse events is remote. On the other SGLT-2 inhibitor-associated DKA were affected by type 1 dia-
hand, it is possible that DKA during SGLT-2 inhibitor treat- betes [7]. Other conditions, which are less likely to occur in
ment was misclassified in some instances. In fact, the typical clinical trials, and which are risk factors for ketoacidosis
feature of this particular form of DKA is a relevant acidosis (e.g., intercurrent illnesses, malnutrition, severe restriction
and ketosis with only mild hyperglycemia [7], which may rep- of dietary carbohydrates, alcohol abuse, etc.) could amplify
resent an obstacle for a correct diagnosis. In addition, the effect of SGLT-2 inhibitors on the risk of DKA. Therefore,
58 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0

Fig. 2 – Funnel plot of standard error by MH log odds ratio for diabetic ketro-acidois.

Fig. 3 – Subgroup analysis comparing SGLT2 inhibitors (SGLT2i) considered as a class with individual comparators.

these results do not allow to overlook the due attention to versus 6 cases in 10,351 patient*years in comparator groups.
condition of risk for DKA when prescribing SGLT-2 inhibitors. Therefore, we can conclude that, on the basis of the results
Although the difference from placebo is not statistically of clinical trials summarized here, the risk of DKA associated
significant, and despite the fact that the estimated odds ratio with SGLT-2 inhibitors is negligible when those drugs are
is not very far from 1.00 (1.22 in the principal analysis), results properly prescribed in patients with type 2 diabetes [21].
of clinical trials do not exclude an increase in risk of ketoaci-
dosis with SGLT-2 inhibitors. In fact, the upper confidence Conflicts of interest
limit is 2.88, meaning that available data are compatible with
an almost 3-fold increase in risk. This latter figure may seem Edoardo Mannucci has received consultancy fees from Astra-
impressive; however, the effect of such a risk increase may be Zeneca, Boehringer, Eli Lilly, Janssen, Merck, Novartis, Novo
very small when the baseline incidence of the event is extre- Nordisk, Sanofi, Takeda; speaking fees from AstraZeneca,
mely low. In fact, in all available trials with SGLT-2 inhibitors, Boehringer, Eli Lilly, Janssen, Merck, Novartis, Novo Nordisk,
only 16 cases of DKA were reported in 20,015 patient*years, Sanofi, Takeda; research grants from AstraZeneca, Eli Lilly,
 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0 59

A B
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0.0 1.0 2.0 3.0 4.0 5.0

Canagliflozin

Empagliflozin

Dapagliflozin

Ipragliflozin

Luseogliflozin

Favours Favours Favours Favours

SGLT2i comparators SGLT2i comparators

N trials MH-OR [95%, CI] p N trials MH-OR [95%, CI] p

Canagliflozin: 4 1.55[0.39-6.21] 0.53 18 0.75[0.39-1.44] 0.38
Empagliflozin 5 0.88[0.25-3.07] 0.84 19 0.54[0.29-1.03] 0.12
Dapagliflozin - - 24 0.68[0·.38-1.19] 0.19
Ipragliflozin - - 12 0.68[0.30-1.52] 0.34
Luseogliflozin - - 2 0.66[0.09-4.72] 0.67

Fig. 4 – Subgroup analysis comparing individual SGLT2 inhibitors (SGLT2i) with any comparators (Panel A: without continuity
correction; Panel B: with continuity correction).

Janssen, Merck, Novartis, Novo Nordisk, Sanofi. Stefania Zan- All the authors approved the final version of this
noni, Carlotta Lualdi and Besmir Nreu have no conflicts of manuscript.
interest to declare. Matteo Monami has received speaking
fees from BMS, Eli Lilly, Merck, Novo Nordisk, Sanofi Takeda,
Acknowledgements
and consultancy fees from Boehringer.

This research was performed independently of any funding,

Contributor statements as part of the institutional activity of the investigators.

Matteo Monami was involved in each of the following points: Appendix A. Supplementary material

1. Design Supplementary data associated with this article can be found,

2. Data Collection in the online version, at http://dx.doi.org/10.1016/j.diabres.
3. Analysis 2017.04.017.
4. Writing manuscript

Besmir Nreu was involved in each of the following points: R E F E R E N C E S

1. Data Collection
2. Manuscript revision [1] Hemmingsen B, Krogh J, Metzendorf MI, Richter B. Sodium-
glucose cotransporter (SGLT) 2 inhibitors for prevention or
delay of type 2 diabetes mellitus and its associated
Stafania Zannoni was involved in each of the following
complications in people at risk for the development of type 2
points:
diabetes mellitus. Cochrane Database Syst Rev 2016. Epub-
ahead of print.
1. Data Collection [2] Scheen AJ. Pharmacodynamics, efficacy and safety of
2. Manuscript revision sodium-glucose co-transporter type 2 (SGLT2) inhibitors for
the treatment of type 2 diabetes mellitus. Drugs
2015;75:33–59.
Edoardo Mannucci was involved in each of the following
[3] Zinman B, Wanner C, Lachin JM, Fitchett D, Bluhmki E, Hantel
points:
S, et al. Empagliflozin, cardiovascular outcomes, and
mortality in type 2 diabetes. N Engl J Med 2015;373:2117–28.
1. Design [4] Wu JH, Foote C, Blomster J, Toyama T, Perkovic V, Sundström
2. Data Collection J, et al. Effects of sodium-glucose cotransporter-2 inhibitors
3. Analysis on cardiovascular events, death, and major safety outcomes
4. Writing manuscript in adults with type 2 diabetes: a systematic review and meta-
analysis. Lancet Diabetes Endocrinol 2016;4:411–9.
60 diabetes research and clinical practice 1 3 0 ( 2 0 1 7 ) 5 3 –6 0

[5] Burke KR et al. SGLT2 inhibitors: a systematic review of diabetes receiving high doses of insulin: efficacy and safety
diabetic ketoacidosis and related risk factors in the primary over 2 years. Diabetes Obes Metab 2014;16:124–36.
literature. Pharmacotherapy 2017;37:187–94. [14] Barnett AH, Mithal A, Manassie J, Jones R, Rattunde H, Woerle
[6] Food and Drug Administration. Safety Alert on Canagliflozin, HJ, et al. Efficacy and safety of empagliflozin added to
issued on Oct 10th, 2015. http://www.fda.gov/Safety/ existing antidiabetes treatment in patients with type 2
MedWatch/SafetyInformation/ diabetes and chronic kidney disease: a randomised, double-
SafetyAlertsforHumanMedicalProducts/ucm461876.htm. Last blind, placebo-controlled trial. Lancet Diabetes Endocrinol
accessed on 3rd August, 2016. 2014;2:369–84.
[7] Peters AL, Buschur EO, Buse JB, Cohan P, Diner JC, Hirsch IB. [15] Rosenstock J, Jelaska A, Frappin G, Salsali A, Kim G, Woerle
Euglycemic diabetic ketoacidosis: a potential complication of HJ, et al. Improved glucose control with weight loss, lower
treatment with sodium-glucose cotransporter 2 inhibition. insulin doses, and no increased hypoglycemia with
Diabetes Care 2015;38:1687–93. empagliflozin added to titrated multiple daily injections of
[8] Qiu H, Novikov A, Vallon V. Ketosis and diabetic ketoacidosis insulin in obese inadequately controlled type 2 diabetes.
in response to SGLT2 inhibitors: basic mechanisms and Diabetes Care 2014;37:1815–23.
therapeutic perspectives. Diabetes Metab Res Rev 2017 [Epub [16] Tikkanen I, Narko K, Zeller C, Green A, Salsali A, Broedl UC,
ahead of print]. et al. Empagliflozin reduces blood pressure in patients with
[9] Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting type 2 diabetes and hypertension. Diabetes Care
items for systematic reviews and meta-analyses: the PRISMA 2015;38:420–8.
statement. Ann Intern Med 2009;151:264–9. [17] http://www.fda.gov/downloads/drugs/
[10] Bode B, Stenlöf K, Sullivan D, Fung A, Usiskin K. Efficacy and guidancecomplianceregulatoryinformation/guidances/
safety of canagliflozin treatment in older subjects with type 2 ucm071627.pdf. Last accessed on 3rd August, 2016.
diabetes mellitus: a randomized trial. Hosp Pract 1995;2013 [18] Taylor SI, Blau JE, Rother KI. SGLT2 inhibitors may predispose
(41):72–84. to ketoacidosis. J Clin Endocrinol Metab 2015;100:2849–52.
[11] https://clinicaltrials.gov/ct2/show/NCT01032629?term= [19] Tang H, Li D, Wang T, Zhai S, Song Y. Effect of sodium-glucose
nct01032629&rank=1. Last accessed on 3rd August, 2016. cotransporter 2 inhibitors on diabetic ketoacidosis among
[12] Lavalle-González FJ, Eliaschewitz FG, Cerdas S, Chacon Mdel patients with type 2 diabetes: a meta-analysis of randomized
P, Tong C, Alba M. Efficacy and safety of canagliflozin in controlled trials. Diabetes Care 2016;39:e123–4.
patients with type 2 diabetes mellitus from Latin America. [20] http://www.fda.gov/Drugs/DrugSafety/ucm446845.htm. Last
Curr Med Res Opin 2016;32:427–39. accessed on 3rd August, 2016.
[13] Wilding JP, Woo V, Rohwedder K, Sugg J, Parikh SDapagliflozin [21] Rosenstock G, Ferrannini E. Euglycemic diabetic ketoacidosis:
006 Study Group. Dapagliflozin in patients with type 2 a predictable, detectable, and preventable safety concern
with SGLT2 inhibitors. Diabetes Care 2015;38:1638–42.