Articles

Blood pressure and glycaemic effects of dapagliflozin versus

placebo in patients with type 2 diabetes on combination
antihypertensive therapy: a randomised, double-blind,
placebo-controlled, phase 3 study
Michael A Weber, Traci A Mansfield, Valerie A Cain, Nayyar Iqbal, Shamik Parikh, Agata Ptaszynska

Summary
Background Hypertension is a common comorbidity in patients with type 2 diabetes mellitus and a major risk factor Lancet Diabetes Endocrinol 2015
for microvascular and macrovascular disease. Although the blood pressure-lowering effects of sodium–glucose Published Online
cotransporter 2 (SGLT2) inhibitors are already established, guidance is needed on how to use these drugs in patients November 24, 2015
http://dx.doi.org/10.1016/
already receiving antihypertensive therapy. We aimed to compare blood pressure and glycaemic effects of the SGLT2
S2213-8587(15)00417-9
inhibitor dapagliflozin with placebo in patients with inadequately controlled type 2 diabetes mellitus and hypertension.
This online publication has
been corrected.
Methods In this double-blind, placebo-controlled, phase 3 study we enrolled patients from 311 centres in 16 countries The corrected version first
across five continents. Patients had uncontrolled type 2 diabetes (HbA1c 7·0%–10·5%; 53–91 mmol/mol) and appeared at thelancet.com/
diabetes-endocrinology on
hypertension (systolic 140–165 mm Hg and diastolic 85–105 mm Hg at both enrolment and randomisation, and a mean
November 27, 2015
24 h blood pressure of ≥130/80 mm Hg by ambulatory monitoring within 1 week of randomisation) and were receiving
See Online/Comment
oral antihyperglycaemic drugs, insulin, or both, plus a renin–angiotensin system blocker and an additional http://dx.doi.org/10.1016/
antihypertensive drug. Using an interactive voice-response system, we randomly assigned (1:1) patients to S2213-8587(15)00457-X
dapagliflozin 10 mg once a day or to placebo, with randomisation stratified by additional antihypertensive drug use SUNY Downstate College of
and insulin use at baseline, in a block size of two. The co-primary endpoints were changes in seated systolic blood Medicine, Brooklyn, NY, USA
pressure and HbA1c measured in the full analysis set, which included all patients who received at least one dose of (M A Weber MD); AstraZeneca,
Fort Washington, PA, USA
study drug and had both a baseline and at least one post-baseline measurement of efficacy. This trial is registered with (T A Mansfield PhD);
ClinicalTrials.gov, number NCT01195662. AstraZeneca, Wilmington, DE,
USA (V A Cain MSc);
Findings Between Oct 29, 2010, and Oct 4, 2012, we randomly assigned 225 patients to dapagliflozin and 224 to placebo. AstraZeneca, Gaithersburg,
MD, USA (N Iqbal MD,
Seated systolic blood pressure was significantly reduced in the group assigned to dapagliflozin (adjusted mean change S Parikh MD); and Bristol-Myers
from baseline –11·90 mm Hg [95% CI –13·97 to –9·82]) compared with those assigned to placebo (–7·62 mm Hg Squibb, Princeton, NJ, USA
[–9·72 to –5·51]; placebo-adjusted difference for dapagliflozin −4·28 mm Hg [–6·54 to –2·02]; p=0·0002). Reductions (A Ptaszynska MD)
in HbA1c concentrations were also significantly greater in patients assigned to dapagliflozin (adjusted mean change Correspondence to:
from baseline –0·63% [95% CI –0·76 to –0·50]) than in those assigned to placebo (–0·02% [–0·15 to 0·12]; placebo- Dr Michael A Weber, SUNY
Downstate College of Medicine,
adjusted difference –0·61% [–0·76 to –0·46,]; p<0·0001). In a post-hoc analysis, we found difference in blood pressure New York, NY 10021, USA
versus placebo was greater in patients receiving a β blocker (−5·76 mm Hg [95% CI −10·28 to −1·23]) or a calcium- michaelwebermd@cs.com
channel blocker (−5·13 mm Hg, [−9·47 to −0·79]) as their additional antihypertensive drug than in those receiving a
thiazide diuretic (–2·38 mm Hg [–6·16 to 1·40]). Adverse events were similar in the dapagliflozin and placebo groups
(98 [44%] patients vs 93 [42%], respectively, had at least one adverse event), with few adverse events related to renal
function (1% vs <1%) or volume depletion (<1% vs 0%).

Interpretation Dapagliflozin 10 mg significantly improved blood pressure and HbA1c and was tolerated similarly to
placebo. Its blood pressure-lowering properties were particularly favourable in patients already receiving a β blocker
or calcium-channel blocker. Dapagliflozin could benefit patients with type 2 diabetes who need a diuretic-like effect to
optimise control of blood pressure, adding meaningful efficacy to antihypertensive drug regimens.

Funding Bristol-Myers Squibb, AstraZeneca.

Introduction achieve blood pressure goals (mostly defined as

Hypertension is a common comorbidity that affects most <140/90 mm Hg).1 Some glucose-lowering drugs are
patients with type 2 diabetes and contributes to the risk of known to affect blood pressure—eg, glucagon-like
cardiovascular disease and microvascular complications.1 peptide-1 (GLP-1) receptor agonists, thiazolidinediones,
Guidelines recommend that people with diabetes and and (in some studies) dipeptidyl peptidase 4 (DPP-4)
hypertension be treated with an angiotensin-converting inhibitors produce small reductions in blood pressure.2
enzyme (ACE) inhibitor or angiotensin II receptor blocker The most commonly prescribed classes of anti-
and indicate that multidrug therapy is often needed to hypertensive drugs are ACE inhibitors, angiotensin II

www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9 1

 Articles

Research in context
Evidence before this study −3·92 mm Hg and −4·80 mm Hg, respectively; p<0·001 for
On Aug 11, 2015, we searched PubMed for articles with the both), although no information was provided about what type
search terms “SGLT2”, “T2DM”, and either “blood pressure” or or dose of antihypertensive drugs were taken by patients.
“hypertension”, to identify reports of randomised controlled Additionally, the results of a systematic review in which data
trials published in English with no date restrictions. We were pooled from 27 randomised controlled trials of SGLT2
identified 143 articles, 23 of which were randomised trials. All inhibitors (n=12 960) showed reductions in systolic blood
studies reported reductions in blood pressure with pressure of 4·0 mm Hg (95% CI 3·5–4·4) relative to control.
dapagliflozin, canagliflozin, empagliflozin, ipragliflozin,
Added value of this study
remogliflozin, or sotagliflozin, with many studies reporting
Dapagliflozin produced clinically meaningful reductions in
statistically significant decreases. Only three study protocols
blood pressure in patients with type 2 diabetes and
mandated that background drugs for blood pressure should be
hypertension, irrespective of the type of concomitant
stable during the study treatment period. Most studies
antihypertensive therapy, albeit with a smaller
reported blood pressure as an efficacy endpoint (either as a
placebo-adjusted reduction in patients already receiving a
co-primary endpoint [n=1], secondary endpoint [n=11], or
thiazide diuretic. To the best of our knowledge, no previous
exploratory endpoint [n=7]), although four trials included
study with an SGLT2 inhibitor has measured blood pressure
blood pressure as a safety outcome. For most trials, the primary
reductions on the basis of underlying antihypertensive drugs.
endpoint was change from baseline in HbA₁C, with no other
dapagliflozin studies and only one empagliflozin study Implications of all the available evidence
(EMPA-REG BP) reporting change from baseline in blood Our findings suggest a possible approach to identify patients with
pressure as a co-primary outcome. The EMPA-REG BP study, diabetes and hypertension most likely to benefit from blood
which was also the only other dedicated randomised trial in pressure-lowering effects of SGLT2 inhibitor therapy.
patients with type 2 diabetes and hypertension, showed Dapagliflozin might be beneficial in patients with type 2 diabetes
significant reductions at week 12 in ambulatory 24 h systolic who require additional control of blood pressure. Further studies
blood pressure (difference vs placebo −3·44 and −4·16 mm Hg are needed to elucidate long-term outcomes with dapagliflozin in
with empagliflozin 10 mg and 25 mg, respectively; p<0·001 for patients with type 2 diabetes and hypertension.
both) and seated systolic blood pressure (difference vs placebo

receptor blockers, thiazide diuretics, calcium-channel combination regimens of antihyperglycaemic and

blockers, and β adrenergic blocking agents (β blockers), antihypertensive drugs in patients with inadequately
each of which lowers blood pressure via distinct controlled type 2 diabetes mellitus and hypertension.
mechanisms.3
Sodium–glucose cotransporter 2 (SGLT2) inhibitors Methods
reduce glucose reabsorption in the proximal renal tubule, Study design and participants
lowering blood glucose via an insulin-independent In this multicentre, randomised, double-blind, placebo-
mechanism.4 Dapagliflozin is an orally-active, highly- controlled, phase 3 trial, we enrolled patients at 311 centres
selective SGLT2 inhibitor,5 which improves glycaemic in Australia, Canada, Colombia, Czech Republic, Denmark,
control and is well tolerated, with a favourable safety Finland, Germany, Hungary, India, Ireland, Mexico,
profile in patients with type 2 diabetes.6–9 Dapagliflozin Poland, Puerto Rico, Romania, UK, and the USA. Eligible
reduces blood pressure in patients with type 2 diabetes patients were those with type 2 diabetes, inadequate
mellitus;10–13 an effect that has been attributed to osmotic glycaemic control (HbA1c 7·0%–10·5%; 53–91 mmol/mol),
diuresis, mild natriuresis, and weight loss.14 Furthermore, and inadequately controlled hypertension (systolic
despite its diuretic effect, dapagliflozin decreases rather 140–165 mm Hg and diastolic 85–105 mm Hg at both
than increases serum uric acid12 and has not been enrolment and randomisation, and a mean 24 h blood
associated with potassium abnormalities.15 pressure of ≥130/80 mm Hg by ambulatory monitoring
Although the blood pressure-lowering effects of SGLT2 within 1 week of randomisation).16,17 Participants were
inhibitors are already established,10–13 guidance is needed required to be on a stable dose of at least one oral
on how to use these agents in patients already receiving antihyperglycaemic drug for 6 weeks or more (≥12 weeks if
the most commonly prescribed antihypertensive thiazolidinedione) or a stable daily dose of insulin (as
regimens; most typically comprising a renin-angiotensin monotherapy or in combination with another oral
system blocker plus either a diuretic, a calcium channel antihyperglycaemic drug) for 8 weeks before enrolment,
blocker, or a β blocker. In this phase 3 trial, we plus a stable, effective, therapeutic dose of a renin-
investigated the hypothesis that the blood pressure- angiotensin system blocker (ACE inhibitor or angiotensin
lowering and glucose-lowering effects of dapagliflozin II receptor blockers) and an additional antihypertensive
10 mg once a day would be superior to placebo in drug for 4 weeks or longer. The types of additional

2 www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9

 Articles

antihypertensive drugs permitted during the study were each study visit (weeks 1, 2, 4, 8, and 12 after randomisation)
pre-specified as a thiazide diuretic, a thiazide-like diuretic, between 0600 h and 1100 h after an initial resting period of
a calcium-channel blocker, a β blocker, an α adrenergic 10 min. The mean was determined from three replicate
blocker, or a central α adrenergic agonist. Key exclusion measurements obtained at least 1 min apart. Patients had
criteria included serum creatinine of 177 μmol/L or lower fasted for 8 h or longer and had last taken their
(unless the patient was taking metformin, in which case antihypertensive and study drugs the previous day.
exclusionary limits were serum creatinine ≥133 μmol/L for Patients taking insulin who had an HbA1c value at
men and ≥124 μmol/L for women); estimated creatinine enrolment of 7·0–7·4% had to reduce their daily insulin
clearance of less than 60 mL per min, and significant dose by 25% at randomisation and maintain this reduction
hepatic disease. Full inclusion and exclusion criteria are throughout the treatment period to avoid hypoglycaemia
listed in the appendix. and ensure patient safety. Any patients with a greater than For the protocol see http://
The study protocol was approved by the institutional 10% change in mean daily insulin dose after day 1 was astrazenecagrouptrials.
pharmacm.com
review board or independent ethics committee at each discontinued. Rescue treatment for hyperglycaemia was
site and all patients provided written informed consent. not permitted. Any patient with inadequate glycaemic
control (defined as a confirmed fasting plasma glucose
Randomisation and masking >15·0 mmol/L) at any time after randomisation was
Patients were randomly assigned (1:1) to receive discontinued. Open-label antihypertensive rescue treat-
dapagliflozin 10 mg once daily or matched placebo. ment was available for severe (>180/110 mm Hg) or
Randomisation was stratified by type of additional sustained (>165/105 and <180/110 mm Hg on two
antihypertensive medication use and insulin use (yes vs occasions) hypertension. Adverse events and laboratory
no) at baseline in block sizes of two. An exploratory values were assessed at every study visit, including at the
treatment arm in which 133 patients were randomised to follow-up visit 1 week after treatment. Adverse events were
dapagliflozin 5 mg was also included in the study, and either volunteered by the patient or obtained through
patients continued treatment for the full study period general questioning and examination at each study visit.
(data not presented here). However, during the course of Systolic blood pressure was also assessed for a final time
the trial, while the study was still blinded, a protocol at the end of the follow-up period (1 week after treatment
amendment was initiated stipulating that patients would with the study drug ended).
no longer be enrolled to the 5 mg arm. Randomisation
was done by interactive voice response system. Outcomes
Randomisation codes were kept centrally at a Bristol- The co-primary endpoints were changes from baseline to
Myers Squibb facility in Lawrenceville, NJ, USA. week 12 in seated systolic blood pressure and HbA1c.
Investigators and patients were masked to treatment Secondary endpoints included change from baseline to
allocation throughout the treatment period, after which week 12 in 24 h ambulatory systolic blood pressure,
the data were unmasked for reporting purposes. Masking seated diastolic blood pressure, and serum uric acid.
was done through the identical appearance of the tablets, Exploratory endpoints included the proportion of patients
pill bottles, and labels. with improved blood pressure control (defined as <140/90
and <130/80 mm Hg); change from baseline to week 12
Procedures in ambulatory daytime, night-time, and trough systolic
The trial consisted of a qualification period (14 days or and diastolic blood pressures; and change from baseline
less after enrolment), a lead-in period (4 weeks), a double- to week 12 in fasting plasma glucose and bodyweight.
blind treatment period (12 weeks), and a follow-up period To assess safety, we recorded the number of patients who
(1 week). During the qualification period, patients were had at least one adverse event, serious adverse event
given a baseline examination to determine eligibility. (defined as those events meeting the International
During the lead-in period, seated blood pressure and Conference on Harmonisation Good Clinical Practice
heart rate were measured 28 days, 14 days, and 1 day criteria7), and adverse event of special interest. The
before the treatment period. Ambulatory blood pressure intensity of adverse events was classified according to the
was assessed 1 day before treatment begun with investigators’ judgment. We also assessed heart rate,
Spacelabs model 90207 ambulatory blood pressure measured orthostatic hypotension (defined as a decrease
monitoring device (Spacelabs Healthcare, Snoqualmie, of >20 mm Hg in systolic blood pressure or >10 mm Hg in
WA, USA). diastolic blood pressure from a supine to a standing
During the 12 week treatment period, patients were position) and the proportion of patients receiving rescue
dispensed 10 mg of dapagliflozin or placebo tablets to be treatment for severe or sustained hypertension, and took
taken at home orally once a day in the morning. All laboratory findings (serum concentrations of potassium,
antihyperglycaemic and antihypertensive drugs that sodium, calcium, magnesium, phosphorous, chloride,
participants were taking before enrolment were continued bicarbonate and creatinine, estimated glomerular filtration
at the same dose throughout the study, with no new drugs rate [eGFR], haematocrit, aspartate aminotransferase,
permitted. Seated systolic blood pressure was measured at alanine aminotransferase, alkaline phosphatase, total

www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9 3

 Articles

provided the best efficacy. On Aug 28, 2012, the protocol

2245 patients enrolled was also amended further to specify use a longitudinal
repeated measures analysis instead of an ANCOVA model
1032 did not complete the enrolment period to account for missing data in the primary efficacy analysis.
934 did not meet eligibility criteria Due to these changes, 408 patients (204 in each group;
65 withdrew consent assuming 5% do not have a post-baseline assessment)
7 lost to follow-up
2 poor or no compliance were needed for at least 80% power to detect a difference of
2 administrative reason by sponsor 4 mm Hg in mean change from baseline in seated systolic
21 other
1 adverse event blood pressure between the active and control group at a
significance level of 0·05, assuming an SD of 14 mm Hg;
and at least 94% power to detect a difference of 0·4% in
1213 completed the enrolment period
mean change from baseline in HbA1c between the groups
at significance level of 0·05, assuming an SD of 1·1%.
625 not randomised Efficacy was assessed in the full analysis set, which
497 did not meet eligibility criteria included all patients who received at least one dose of
69 withdrew consent
13 lost to follow-up study medication and had both a baseline and at least one
11 poor or no compliance post-baseline efficacy measurement. Data for patients after
8 administrative reason by sponsor
6 adverse event
rescue for severe or sustained hypertension were excluded
2 patient request to discontinue study treatment from blood pressure analyses from after the rescue date,
19 other but all data were included in analyses for change in HbA1c,
serum uric acid, fasting plasma glucose, and bodyweight.
588 randomised to treatment The safety analysis set included all patients who received at
least one dose of study drug, irrespective of rescue
treatment.
139 not analysed A hierarchical closed testing procedure was done
133 randomised to dapagliflozin 5 mg
6 no longer met eligibility criteria across the primary and subsequent secondary endpoints
to control family-wise type 1 error rate at a two-sided
significance level of 0·05. For the primary efficacy
analysis, we used a longitudinal repeated measures
224 randomly assigned to placebo 225 randomly assigned to dapagliflozin 10 mg
analysis using the direct likelihood method with fixed
categorical effects of treatment, week, treatment-by-week
22 not completed 14 not completed interaction, and randomisation strata, and continuous
6 withdrew consent 4 withdrew consent fixed covariates of baseline seated systolic blood pressure
3 lost to follow-up 5 no longer met study criteria
2 administrative reason by sponsor 2 lost to follow-up value and baseline seated systolic blood pressure value-
2 lack of efficacy 1 adverse event by-week interaction. If this primary analysis was
4 adverse events 1 administrative reason by
3 other sponsor
statistically significant at the 0·05 level, the statistical test
1 request to discontinue study 1 other* for the second co-primary endpoint (change from
1 no longer met study criteria baseline in HbA1c) was done. This analysis used a similar
longitudinal repeated measures analysis except that the
202 completed study 211 completed study continuous fixed covariates were baseline HbA1c and
baseline HbA1c-by-week interaction. If the comparisons
Figure 1: Trial profile
between the dapagliflozin 10 mg treatment group and
*Site disbanded before patient discontinuation information was obtained. the placebo group were significant at the 0·05 level for
both co-primary endpoints, then the statistical tests for
the secondary efficacy endpoints were done. All other
bilirubin, urine albumin to creatinine ratio [UACR]). efficacy outcomes were assessed by use of longitudinal
repeated measures analyses, apart from 24 h ambulatory
Statistical analysis systolic blood pressure and daytime ambulatory systolic
This trial (MB102077) was originally planned as a four-arm, blood pressure, for which an ANCOVA model was used,
parallel-group trial to compare the addition of the following with treatment group as an effect and baseline value and
to patients regimens: dapagliflozin 2·5 mg, dapagliflozin randomisation strata as covariates. All analyses were
5 mg, dapagliflozin 10 mg, and placebo. The protocol done with SAS/STAT version 8.2 or higher.
specified that 1104 patients should be assigned into the This study is registered with ClinicalTrials.gov, number
four groups (276 in each) in a 1:1:1:1 ratio. On Aug 18, 2010, NCT01195662.
the protocol was amended to remove the 2·5 mg treatment
arm. On Nov 1, 2011, enrolment to the 5 mg arm was Role of the funding source
discontinued when it became clear that the 10 mg dose The study funders were involved in the study design;

4 www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9

 Articles

Overall population Additional antihypertensive subgroup

Thiazide diuretic* Calcium-channel blocker* β blocker*
Placebo (n=224) Dapagliflozin Placebo (n=77) Dapagliflozin Placebo (n=61) Dapagliflozin Placebo (n=59) Dapagliflozin
(n=225) (n=92) (n=60) (n=57)
Age (years) 57·0 56·0 56·0 54·0 59·0 55·5 57·0 60·0
(51·0–62·0) (50·0–62·0) (51·0–61·0) (48·5–61·0) (51·0–62·0) (49·0–60·0) (51·0–62·0) (52·0–64·0)
Sex
Men 129 (58%) 118 (52%) 42 (55%) 50 (54%) 34 (56%) 30 (50%) 38 (64%) 30 (53%)
Women 95 (42%) 107 (48%) 35 (46%) 42 (46%) 27 (44%) 30 (50%) 21 (36%) 27 (47%)
Ethnic origin
White 157 (70%) 160 (71%) 56 (73%) 68 (74%) 36 (59%) 33 (55%) 43 (72·9) 46 (81%)
Black 17 (8%) 19 (8%) 9 (12%) 7 (8%) 6 (10%) 9 (15%) 0 2 (4%)
Asian 38 (17%) 34 (15%) 8 (10%) 11 (12%) 15 (27%) 15 (25%) 12 (20%) 6 (11%)
Other 12 (5%) 12 (5%) 4 (5%) 6 (7%) 4 (7%) 3 (5%) 4 (7%) 3 (5%)
Bodyweight (kg) 89·9 (18·4) 88·0 (20·5) 89·0 (17·7) 89·3 (22·4) 87·8 (17·4) 87·5 (21·0) 92·0 (20·1) 87·8 (17·5)
Duration of type 2 diabetes (years) 7·3 (5·0) 7·7 (5·9) 6·5 (4·6) 7·5 (6·5) 7·0 (5·6) 7·0 (5·3) 8·1 (5·1) 8·1 (5·3)
HbA₁C (%; mmol/mol) 8·0% (1·0); 64 8·1% (0·9) 65 8·0% (1·0); 8·2% (1·0); 7·8% (1·0); 8·1% (1·0); 8·2% (0·9); 8·0% (0·8);
(10·6) (10·1) 64 (11·3) 66 (10·7) 62 (11·4) 65 (10·7) 66 (9·9) 64 (8·3)
Fasting plasma glucose (mmol/L) 8·9 (2·4) 9·0 (2·5) 9·0 (2·4) 9·5 (2·5) 8·3 (2·4) 8·4 (2·3) 9·4 (2·5) 8·9 (2·5)
Participants taking insulin 16 (7·1) 18 (8·0) 4 (5·2) 5 (5·4) 3 (4·9) 7 (11·7) 7 (11·9) 4 (7·0)
Duration of hypertension (years) 9·3 (7·7) 9·4 (7·8) 8·7 (7·4) 8·5 (8·2) 9·1 (7·8) 8·5 (6·6) 9·3 (7·0) 10·6 (8·0)
Systolic blood pressure (mm Hg) 151·3 (6·7) 151·0 (7·9) 151·8 (6·9) 151·1 (8·6) 150·0 (6·4) 150·1 (7·4) 151·4 (6·9) 152·6 (7·3)
Diastolic blood pressure (mm Hg) 91·4 (4·8) 91·2 (4·8) 91·9 (5·0) 91·5 (4·7) 90·7 (4·8) 91·6 (4·9) 91·1 (4·3) 90·6 (4·8)
Selective vascular history
Coronary artery disease 10 (5%) 12 (5%) 4 (5%) 3 (3%) 2 (3%) 1 (2%) 2 (3%) 7 (12%)
Stable angina 4 (2%) 4 (2%) 2 (3%) 0 0 1 (2%) 2 (3%) 3 (5%)
Peripheral vascular disease 8 (4%) 1 (<1%) 3 (4%) 0 1 (2%) 0 2 (3%) 1 (2%)
Carotid artery disease 2 (1%) 2 (1%) 2 (3%) 0 0 1 (2%) 0 1 (2%)
Amputation 1 (<1%) 0 1 (1%) 0 0 0 0 0
History of dyslipidaemia 113 (50·4) 112 (49·8) 31 (40·3) 37 (40%) 34 (56%) 32 (53%) 32 (54%) 33 (58%)
Recent cardiovascular event† 19 (9%) 21 (9%) 8 (10%) 6 (7%) 4 (7%) 2 (3%) 5 (9%) 12 (21%)
eGFR (mL/min per 1·73 m²) 87·0 (19·5) 84·8 (19·7) 86·7 (21·3) 85·1 (21·6) 84·4 (19·3) 86·9 (17·8) 90·9 (18·7) 83·1 (18·8)

Data are median (IQR), mean (SD), or n (%). ACE=angiotensin-converting enzyme. eGFR=estimated glomerular filtration rate. *Patients who did not take an additional antihypertensive drug from any of the
three subgroups or who received drugs from more than one subgroup were excluded from subgroup analysis. †Previous myocardial infarction, percutaneous coronary intervention, coronary artery bypass graft,
carotid endarterectomy or stenting, peripheral vascular surgery, cerebrovascular accident, transient ischaemic attack, congestive heart failure, or hospitalisation for unstable angina.

Table 1: Baseline demographics and disease characteristics, overall and by additional antihypertensive drug treatment

collection, analysis, and interpretation of data; and the thiazide diuretic, a thiazide-like diuretic, a calcium-
decision to submit for publication. The corresponding channel blocker, a β blocker, or an α adrenergic blocker.
author had full access to all the data in the study and had No participants received a central α adrenergic agonist.
final responsibility for the decision to submit for Antihyperglycaemic drugs included metformin
publication. (203 [90%] of 225 assigned to dapagliflozin vs 206 [92%]
of 224 assigned to placebo), sulfonylurea (105 [47%] vs 105
Results [47%]), DPP-4 inhibitors (16 [7%] vs 20 [9%]),
Between Oct 29, 2010, to Oct 4, 2012, we enrolled 1213 thiazolidinedione (eight [4%] vs nine [4%]), insulin
patients (figure 1). Of 449 patients in the full analysis set, (18 [8%] vs 16 [7%]), acarbose (five [2%] vs three [1%]), and
225 were randomly assigned to receive dapagliflozin meglitinides (three [1%] vs none). Baseline demographics
10 mg and 224 to receive placebo. Before protocol and disease characteristics were generally well balanced
amendment, 133 patients were originally assigned to between treatment groups and additional antihyper-
dapagliflozin 5 mg a day and continued this treatment tensive subgroups (table 1), although history of
for the full study period (data not shown). cardiovascular diseases differed slightly between
About 55% of participants received a stable dose of an additional antihypertensive drug subgroups. Most
ACE inhibitor (126 patients assigned to dapagliflozin vs patients had normal renal function or mild renal
124 assigned to placebo) and about 45% received an impairment, and less than a tenth of patients had an
angiotensin II receptor blocker (99 vs 99), in addition to a eGFR less than 60 mL/min per 1·73 m² at baseline.

www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9 5

 Articles

A Seated systolic blood pressure B HbA1C

0 0·2
Placebo 0·1
–2
Adjusted mean change from

Adjusted mean change from

Dapagliflozin 0
–4 –0·1
baseline (mm Hg)

baseline (%)
–6 –0·2
–0·3
–8 –0·4
–10 –0·5
–0·6
–12
–0·7
–14 –0·8
Baseline 2 4 8 12 Baseline 4 8 12
Time (weeks) Time (weeks)
Number at risk Number at risk
Placebo 219 218 213 205 199 Placebo 217 214 207 197
Dapagliflozin 224 221 220 212 205 Dapagliflozin 220 219 211 204

C Ambulatory systolic blood pressure* D Serum uric acid

0 20
–2 10
Adjusted mean change from

Adjusted mean change from

–4
baseline (mm Hg)

baseline (μmol/L)
0
–6
–10
–8
–20
–10
–30
–12
–14 –40

–16 –50
0 2 4 6 8 10 12 14 16 18 20 22 24 Baseline 4 8 12
Time since recording initiated (h) Time (weeks)
Number at risk Number at risk
Placebo 176 178 181 184 184 184 182 183 184 184 185 151 Placebo 217 214 207 198
Dapagliflozin 175 181 182 184 184 185 185 184 185 185 183 158 Dapagliflozin 220 219 212 204

Figure 2: Change over 12 weeks in seated systolic blood pressure (A), HbA₁C (B), 24 h ambulatory systolic blood pressure (C), and serum uric acid (D) in the full
analysis set
Error bars show 95% CIs. Blood pressure data excluded data after antihypertensive rescue treatment; data for HbA₁C and serum uric acid included data after
antihypertensive rescue treatment. *24 h ambulatory monitoring was initiated between 0600 h and 1100 h for 24 h during week 12.

Patients assigned to dapagliflozin 10 mg had pronounced in patients in the dapagliflozin group than in
significantly greater reductions in mean seated systolic those in the placebo group (placebo-adjusted difference
blood pressure from baseline to week 12 than did those −4·45 mm Hg [95% CI –7·14 to –1·76]; p=0·0012;
assigned to placebo (adjusted mean change from figure 2C). At 12 weeks, dapagliflozin was associated with
baseline –11·90 mm Hg [95% CI –13·97 to –9·82] vs consistently greater reductions in ambulatory systolic
–7·62 mm Hg [–9·72 to –5·51], respectively; placebo- blood pressure versus placebo at every timepoint assessed
adjusted difference −4·28 mm Hg [–6·54 to –2·02]; over a 24 h period. Mean seated diastolic blood pressure
p=0·0002; figure 2A). At week 13, 1 week after the study fell by 6·30 mm Hg in those assigned to dapagliflozin
drug was stopped, systolic blood pressure did not differ (95% CI 5·06 to 7·54) versus 5·33 mm Hg (4·08 to 6·59)
between the dapagliflozin and placebo groups (adjusted in those assigned to placebo, respectively (placebo-
mean change from baseline –10·13 mm Hg [95% CI adjusted difference −0·97 mm Hg [95% CI –2·32 to
–12·26 to –8·01] vs –8·93 mm Hg [–11·08 to –6·78], 0·39]; p=0·16). The mean change from baseline to
respectively; placebo-adjusted difference −1·20 mm Hg week 12 in fasting plasma glucose was −1·0 mmol/L
[95% CI −3·56 to 1·15]). (95% CI –1·4 to −0·6) in the dapagliflozin group
Reductions inHbA1c concentrations were significantly compared with 0·2 mmol/L (–0·2 to 0·6) in the placebo
greater in patients assigned to dapagliflozin 10 mg than group, with a placebo-adjusted difference of –1·2 mmol/L
in those assigned to placebo (adjusted mean change (–1·7 to –0·8). The change from baseline in serum uric
from baseline –0·63% [95% CI –0·76 to –0·50] vs –0·02% acid at week 12 was –25·39 μmol/L (95% CI −35·71 to
[–0·15 to 0·12], respectively; placebo-adjusted difference −15·07) in those assigned to dapagliflozin compared with
–0·61% [–0·76 to –0·46], p<0·0001; figure 2B). –1·72 μmol/L (−12·12 to 8·69) in those assigned to
Mean reductions from baseline values of 24 h placebo (placebo-adjusted difference −23·67 μmol/L
ambulatory systolic blood pressure at week 12 were more [95% CI –33·70 to –13·64; figure 2D).

6 www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9

 Articles

A B
n Baseline mean Difference vs placebo n Baseline mean Difference vs placebo
seated SBP, mm Hg (SD) (95% CI) HbA1C, % (SD) (95% CI)

Overall population Overall population

Placebo 199 151·3 (6·8) –4·28 (–6·54 to –2·02) Placebo 197 8·00 (0·96) –0·61 (–0·76 to –0·46)
Dapagliflozin 205 151·0 (7·9) Dapagliflozin 204 8·09 (0·91)
β blocker subgroup β blocker subgroup
Placebo 52 151·4 (6·9) –5·76 (–10·28 to –1·23) Placebo 51 8·15 (0·92) –0·80 (–1·10 to –0·50)
Dapagliflozin 51 152·6 (7·3) Dapagliflozin 51 8·01 (0·79)
Calcium-channel blocker subgroup Calcium-channel blocker subgroup
Placebo 56 150·0 (6·4) –5·13 (–9·47 to –0·79) Placebo 54 7·87 (1·02) –0·65 (–0·94 to –0·35)
Dapagliflozin 58 150·1 (7·4) Dapagliflozin 58 8·13 (0·97)
Diuretic subgroup Diuretic subgroup
Placebo 69 151·8 (6·9) –2·38 (–6·16 to –1·40) Placebo 70 8·00 (1·01) –0·45 (–0·70 to –0·19)
Dapagliflozin 80 151·1 (8·6) Dapagliflozin 79 8·18 (0·98)

–20 –10 0 –1·0 –0·5 0 0·5

Adjusted mean change (mm Hg) Adjusted mean change (%)

C D
n Baseline mean Difference vs placebo n Baseline mean Difference vs placebo
ambulatory SBP, mm Hg (SD)* (95% CI) serum uric acid, μmol/L (SD) (95% CI)

Overall population Overall population

Placebo 186 149·2 (12·7) –4·45 (–7·14 to 1·76) Placebo 198 325·28 (78·92) –23·67 (–33·70 to –13·64)
Dapagliflozin 187 146·5 (10·4) Dapagliflozin 204 334·95 (92·59)
β blocker subgroup β blocker subgroup
Placebo 46 149·8 (11·6) –6·16 (–11·70 to –0·62) Placebo 51 313·46 (71·97) –36·88 (–56·51 to –17·25)
Dapagliflozin 47 146·1 (11·1) Dapagliflozin 51 333·68 (88·03)
Calcium-channel blocker subgroup Calcium-channel blocker subgroup
Placebo 52 149·3 (13·3) –4·45 (–9·30 to 0·39) Placebo 54 318·81 (75·53) –19·63 (–38·66 to –0·59)
Dapagliflozin 50 146·3 (9·6) Dapagliflozin 58 332·49 (86·84)
Diuretic subgroup Diuretic subgroup
Placebo 67 148·8 (13·8) –4·92 (–9·42 to –0·43) Placebo 70 336·06 (87·43) –13·09 (–29·15 to 2·97)
Dapagliflozin 75 146·9 (9·5) Dapagliflozin 79 331·30 (100·52)

–20 –15 –10 –5 0 –60 –40 –20 0 20 40

Adjusted mean change (mm Hg) Adjusted mean change (μmol/L)

Figure 3: Baseline and placebo-adjusted 12 week difference in seated SBP (A), HbA₁C (B), 24 h ambulatory SBP (C), and serum uric acid (D) by additional antihypertensive drug type, in the full
analysis set
Error bars show 95% CIs. Blood pressure data excluded data after antihypertensive rescue treatment; data for HbA₁C and serum uric acid included data after antihypertensive rescue treatment. Numbers
in antihypertensive subgroups are only those with baseline and week 12 values available.*24 h ambulatory monitoring was initiated between 0600 h and 1100 h for 24 h on day 0 (baseline)
and week 12. SBP=systolic blood pressure.

We did post-hoc subgroup analyses to assess efficacy when measured with 24 h ambulatory blood pressure
variables by the class of additional antihypertensive drug monitoring (figure 3C). Seated diastolic blood pressure
received by patients during the treatment period (table 1). showed no relevant differences between the
These analyses included only data from patients in the antihypertensive subgroups. The proportion of patients
thiazide diuretic, calcium-channel blocker, and β blocker with improved blood pressure control (<140/90 mm Hg)
subgroups because the numbers of patients in the other was 38% (85 of 224) assigned to dapagliflozin versus 33%
subgroups were too small to make meaningful conclusions (72 of 219) assigned to placebo. The proportion of patients
(<10 patients in either treatment group); patients who did with improved blood pressure control (<130/80 mm Hg)
not take an additional antihypertensive drug from any of was 8% (18 of 224) assigned to dapagliflozin versus 5%
these three subgroups or who took drugs from more than (12 of 219) assigned to placebo. Change from baseline in
one subgroup were excluded from this subgroup analysis. ambulatory daytime, night-time and trough systolic and
Seated systolic blood pressure was reduced with diastolic blood pressures are shown in the appendix.
dapagliflozin versus placebo to a greater extent in the Slightly greater reductions in daytime ambulatory
β blocker subgroup and the calcium-channel blocker systolic blood pressure were noted in patients assigned to
subgroup than in the thiazide diuretic subgroup dapagliflozin versus placebo (appendix), with greater
(figure 3A). This difference in systolic blood pressure reductions in the β blocker subgroup than in the calcium-
between the antihypertensive subgroups was reduced channel blocker or thiazide diuretic subgroups.

www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9 7

 Articles

Overall population Additional antihypertensive subgroup

Thiazide diuretic Calcium-channel blocker β blocker
Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin
(n=224) (n=225) (n=77) (n=92) (n=61) (n=60) (n=59) (n=57)
One or more adverse event 93 (42%) 98 (44%) 27 (35%) 46 (50%) 26 (43%) 21 (35%) 27 (46%) 24 (42%)
Adverse events that occurred in at least 10% of participants in either treatment group
Infections and infestations 32 (14%) 44 (20%) 7 (9%) 20 (22%) 13 (21%) 8 (13%) 7 (12%) 12 (21%)
Adverse event leading to discontinuation 4 (2%) 1 (<1%) 0 1 (1%) 2 (3%) 0 1 (2%) 0
One or more serious adverse event 2 (1%) 6 (3%) 1 (1%) 2 (2%) 1 (2%) 0 0 2 (4%)
Special interest categories
Hypoglycaemia* 6 (3%) 13 (6%) 1 (1%) 2 (2%) 1 (2%) 6 (10%) 4 (7%) 5 (9%)
Renal function† 1 (<1%) 3 (1%) 0 2 (2%) 0 0 0 1 (2%)
Volume depletion†‡ 0 1 (<1%) 0 1 (1%) 0 0 0 0
Genital infection† 4 (2%) 6 (3%) 1 (1%) 4 (4%) 3 (5%) 0 0 1 (2%)
Urinary tract infection† 2 (1%) 4 (2%) 0 3 (3%) 1 (2%) 0 0 1 (2%)

Includes data after antihypertensive rescue. No deaths or malignancies were reported in the study. *We noted no major episodes of hypoglycaemia, defined as symptomatic episodes requiring external assistance
with a capillary or plasma glucose value of less than 3 mmol/L and prompt recovery after glucose or glucagon administration; minor episodes of hypoglycaemia were defined as any symptomatic or
non-symptomatic episode with a capillary or plasma glucose measurement of less than 3·5 mmol/L that did not qualify as a major episode; other episodes of hypoglycaemia were defined as investigator-reported
episodes suggestive of hypoglycaemia that did not meet these criteria. †Based on a predefined list of preferred terms from Medical Dictionary for Regulatory Activities (MedDRA) version 15.1. ‡Defined as
hypotension, dehydration, or hypovolaemia.

Table 2: Summary of adverse events for the 12 week treatment period

The effect of dapagliflozin on HbA1c concentrations was patients had infections with dapagliflozin than with
marginally more pronounced in the β blocker and placebo in the thiazide diuretic subgroup (namely
calcium-channel blocker subgroups than in the thiazide asymptomatic bacteriuria, urinary tract infection,
diuretic subgroup (figure 3B). There were no clinically influenza, and nasopharyngitis) and the β blocker
relevant differences in mean change from baseline in subgroup (namely asymptomatic bacteriuria and
mean fasting plasma glucose between the additional nasopharyngitis), whereas these were less common with
antihypertensive subgroups (placebo-adjusted difference dapagliflozin than with placebo in the calcium-channel
for thiazide diuretics –1·1 mmol/L [95% CI –1·9 to –0·4]; blocker subgroup. Few serious adverse events occurred
calcium-channel blocker –1·4 mmol/L [–2·3 to –0·5]; and during the study, but more patients in the dapagliflozin
β blockers –0·9 mmol/L [–1·8 to 0·0]). Reductions in group (n=6; 3%) than in the placebo group (two [1%]) had
serum uric acid were greater in the β blocker subgroup a serious adverse event. Only one serious adverse event (in
than the calcium-channel blocker or thiazide diuretic the placebo group) led to discontinuation.
subgroups (figure 3D). Hypoglycaemia was uncommon in both groups, but
The mean change in bodyweight from baseline to affected slightly more patients assigned to dapagliflozin
week 12 was −1·44 kg (95% CI −1·95 to −0·92) in patients than those assigned to placebo (table 2). No cases of major
assigned to dapagliflozin versus −0·59 kg (95% CI −1·11 to hypoglycaemia were reported. The incidence of
−0·07) in those assigned to placebo (placebo-adjusted hypoglycaemia was higher in patients assigned to
difference −0·85 kg [−1·39 to −0·31]). The timecourse for dapagliflozin than in those assigned to placebo in the
bodyweight reduction seemed similar to that of systolic calcium-channel blocker subgroup whereas it was lower
blood pressure reduction (appendix). Bodyweight was in patients in the thiazide diuretic or β blocker subgroups,
reduced with dapagliflozin compared with placebo to a although this effect might be due to higher insulin use in
greater extent in the β blocker subgroup (placebo-adjusted the calcium-channel blocker subgroup (table 1). Renal
difference −1·51 kg [95% CI −2·64 to −0·39]) than in the function-related adverse events were low in both treatment
calcium-channel blocker subgroup (−0·67 kg [−1·77, 0·43]) groups and across all antihypertensive subgroups (table 2).
or thiazide diuretic subgroup (−0·49 kg [−1·43 to 0·44]). All renal events were mild or moderate in intensity and
The incidence of adverse events was similar between the were mostly due to small changes in creatinine
dapagliflozin (98 [44%] of 225 in the safety set) and placebo concentration. Renal function in the dapagliflozin 10 mg
(93 [42%] of 224) group, with few events leading to group remained stable over 12 weeks, with no clinically
discontinuation (table 2). Adverse events were more meaningful change in mean eGFR or urinary albumin-to-
common in those taking dapagliflozin (n=46; 50%) than creatinine ratio (appendix). Laboratory values did not
in those taking placebo (n=37; 35%) in the thiazide diuretic differ between the dapagliflozin and placebo groups at
subgroup (table 2), whereas events were less common in week 12 (appendix) and mean potassium and sodium
those taking dapagliflozin in the other subgroups. More concentrations remained stable in both groups.

8 www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9

 Articles

At baseline, measured orthostatic hypotension was Dapagliflozin also significantly reduced 24 h ambulatory
reported in two patients in each treatment group. At systolic blood pressure compared with placebo, with no
week 12, measured orthostatic hypotension was noted in difference between the three antihypertensive subgroups
seven (3%) patients in the dapagliflozin 10 mg group and at 12 weeks. Since ambulatory blood pressure is regarded
four (2%) in the placebo group, but no patients reported it as a highly reliable basis for assessing hypertension,19 these
as an adverse event. Only two patients (both in the placebo results further corroborate the clinical benefits of
group) received rescue medication for severe or sustained dapagliflozin for patients with hypertension and type 2
hypertension. Seated heart rate at 12 weeks did not diabetes. Dapagliflozin was not associated with significant
meaningfully differ from that at baseline for either group reductions in diastolic blood pressure. However, systolic
(−1·4 beats per min [bpm] with dapagliflozin 10 mg [baseline blood pressure is regarded as a stronger indicator of
77·1] vs −0·5 bpm with placebo [baseline 77·0]). cardiovascular risk than is diastolic blood pressure,20 which
has led to an increased emphasis on systolic blood pressure
Discussion control in recent guidelines.3,21
In this study, the addition of dapagliflozin to treatment Blood pressure-lowering effects have been reported for
regimens for 12 weeks was associated with lowered blood dapagliflozin in other clinical trials10–12 and in trials of other
pressure and improved glycaemic control in patients with SGLT2 inhibitors.22,23 Furthermore, we have previously
type 2 diabetes and hypertension inadequately controlled reported that dapagliflozin treatment is complementary to
with up to two antihyperglycaemic drugs, a renin- the use of renin–angiotensin system inhibitors.24 The
angiotensin system blocker, and an additional anti- results of the present study expand further our
hypertensive drug. understanding of dapagliflozin and suggest that reduction
Despite the concurrent antihypertensive treatments, of blood pressure is maintained irrespective of the class of
the observed blood pressure effects of dapagliflozin seem additional antihypertensive drug, with a slightly greater
potentially clinically relevant. However, for patients effect in patients already receiving a β blocker or calcium-
already receiving a renin–angiotensin system blocker channel blocker than in patients already on a thiazide
plus a thiazide diuretic, dapagliflozin produced a rather diuretic. Moreover, this trial is the first with dapagliflozin
small reduction in placebo-adjusted systolic blood in which blood pressure was measured as a co-primary
pressure compared with those receiving a renin- endpoint in patients with inadequately controlled
angiotensin system blocker plus a β blocker or calcium- hypertension and type 2 diabetes, with no changes or
channel blocker. This finding is not wholly unexpected, additions of background antihypertensive drugs allowed.
since adding a drug with a predominantly diuretic- We also note that patients in this study had a fairly long
dependent antihypertensive action14 to patients already duration of hypertension (mean 9·4 years) and were
receiving a thiazide drug might be less likely to have a uncontrolled despite dual combination therapy, suggesting
major additive effect. This explanation is supported by that dapagliflozin might provide beneficial blood pressure
our finding that the weight loss produced by dapagliflozin effects in this setting. Moreover, dapagliflozin 10 mg was
in the diuretic subgroup was less than that in the generally well tolerated and, unlike conventional thiazide
β blocker and calcium-channel blocker subgroups. Still, diuretics,25 did not cause hypokalaemia.
we cannot exclude the possibility that the slight placebo- The glucose-lowering actions of dapagliflozin seen in
subtracted effect of dapagliflozin in the diuretic group this study have also been well documented in previous
might have been affected by the chance occurrence of a trials.7–9 The effects of dapagliflozin were similar across
relatively large placebo effect. the antihypertensive subgroups, although variations in
Our results show clinically meaningful, placebo-adjusted, the effects of placebo meant that the placebo-adjusted
decreases in systolic blood pressure of 5·1 mm Hg and reductions in HbA1c were slightly less substantial in the
5·8 mm Hg in patients treated with a β blocker or calcium- thiazide diuretic subgroup than in the β blocker or
channel blocker, respectively, which are of a similar scale to calcium-channel blocker subgroups.
that which might be anticipated with a conventional Reductions in serum uric acid with dapagliflozin have
diuretic added to ongoing dual antihypertensive therapy. been reported previously,12,24 but our results confirm
For example, in an authoritative trial in patients with stage 2 decreases with dapagliflozin in all antihypertensive
hypertension (>160/100 mm Hg) receiving a renin- subgroups (including the thiazide diuretic subgroup, in
angiotensin system blocker plus a calcium-channel blocker, which there might be some concern about diuretic-
the addition of hydrochlorothiazide 25 mg decreased induced increases in uric acid25). Dapagliflozin was also
systolic blood pressure by 6·2 mm Hg and diastolic associated with small reductions in bodyweight and,
blood pressure by 3·3 mm Hg compared with a renin– although it is unclear whether this effect contributed to
angiotensin system blocker plus calcium-channel blocker the reduction in blood pressure, the timecourse for blood
alone,18 an effect only marginally greater than we recorded pressure reduction followed that of bodyweight
in the present study despite the fact that the baseline blood reduction. Weight loss was slightly more pronounced
pressure in the stage 2 patients was clearly higher than in among patients in the β blocker subgroup, although this
our study population. result could have been affected by low patient numbers.

www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9 9

 Articles

Data from previous studies suggest that β blockers might 7 Bailey CJ, Gross JL, Pieters A, Bastien A, List JF. Effect of
dapagliflozin in patients with type 2 diabetes who have inadequate
be associated with weight gain.26 glycaemic control with metformin: a randomised, double-blind,
Our study had some limitations. Because of the strict placebo-controlled trial. Lancet 2010; 375: 2223–33.
criteria for enrolment, we had to include many study 8 Strojek K, Yoon KH, Hruba V, Elze M, Langkilde AM, Parikh S.
sites to recruit enough patients. However, protocol Effect of dapagliflozin in patients with type 2 diabetes who have
inadequate glycaemic control with glimepiride: a randomized,
compliance and data accuracy were ensured by rigorous 24-week, double-blind, placebo-controlled trial.
monitoring. Another possible limitation, as indicated Diabetes Obes Metab 2011; 13: 928–38.
earlier, is that the attenuated antihypertensive efficacy of 9 Wilding JP, Norwood P, T’Joen C, Bastien A, List JF, Fiedorek FT.
A study of dapagliflozin in patients with type 2 diabetes receiving
dapagliflozin in patients already receiving a thiazide high doses of insulin plus insulin sensitizers: applicability of a novel
diuretic might have resulted simply from a relatively insulin-independent treatment. Diabetes Care 2009; 32: 1656–62.
large placebo effect, although a smaller reduction in 10 Woo V, Langkilde AM, Sugg J, Parikh S. Blood pressure reduction
with dapagliflozin in patients with type 2 diabetes and
bodyweight seemed to confirm a lesser diuretic action of cardiovascular disease. Circulation 2013; 128 (suppl 22): A10606.
dapagliflozin in these thiazide-treated patients. A further 11 Woo V, Langkilde AM, Parikh S. Dapagliflozin, a novel
limitation of the study was that very few patients with antihyperglycemic agent that promotes urinary glucose excretion,
reduces systolic blood pressure in patients with type 2 diabetes
moderate renal impairment were included because mellitus. Circulation 2011; 124 (suppl 21): A9520 (abstr).
SGLT2 inhibitors have a lower efficacy in this population.6 12 Basile J, Ptaszynska A, Ying L, Sugg J, Parikh S. The effects of
Additionally, despite the useful reductions in blood dapagliflozin on cardiovascular risk factors in patients with type 2
diabetes mellitus. Circ Cardiovasc Qual Outcomes 2012; 5: A59.
pressure, our study was not designed to measure major
13 Sjostrom CD, Johansson P, Ptaszynska A, List J, Johnsson E.
cardiovascular endpoints. However, an outcomes trial Dapagliflozin lowers blood pressure in hypertensive and
with dapagliflozin is currently in progress (Dapagliflozin non-hypertensive patients with type 2 diabetes. Diab Vasc Dis Res
2015; 12: 352–58.
Effect on CardiovascuLAR Events [DECLARE TIMI-58];
14 Lambers Heerspink HJ, de Zeeuw D, Wie L, Leslie B, List J.
NCT01730534). Furthermore, in the recent EMPA-REG Dapagliflozin a glucose-regulating drug with diuretic properties in
OUTCOME trial in patients with type 2 diabetes at high subjects with type 2 diabetes. Diabetes Obes Metab 2013; 15: 853–62.
risk of cardiovascular events, the SGLT2 inhibitor 15 Yavin Y, Mansfield TA, Ptaszynska A, et al. Hyperkalemia incidence
with the SGLT2 inhibitor dapagliflozin. Diabetes 2014;
empagliflozin reduced the risk of major cardiovascular 63 (suppl 1): 1086-P.
events compared with placebo.27 16 Standards of medical care in diabetes—2009. Diabetes Care 2009;
32 (suppl 1): S13–61.
Contributors
All authors contributed to the conception and design of the study, the 17 Chobanian AV, Bakris GL, Black HR, et al. The seventh report of the
Joint National Committee on Prevention, Detection, Evaluation, and
acquisition and interpretation of the data, and the drafting and critical
Treatment of High Blood Pressure: the JNC 7 report. JAMA 2003;
revision of this report. 289: 2560–72.
Declaration of interests 18 Calhoun DA, Lacourciere Y, Chiang YT, Glazer RD. Triple
MAW received research funding from Bristol-Myers Squibb, and has been antihypertensive therapy with amlodipine, valsartan, and
a consultant to AstraZeneca, Novartis, and Forest Pharmaceuticals. TAM is hydrochlorothiazide: a randomized clinical trial. Hypertension 2009;
an employee of AstraZeneca and a shareholder of AstraZeneca and 54: 32–39.
Bristol-Myers Squibb. VAC, NI, and SP are employees and shareholders of 19 Pickering TG, Shimbo D, Haas D. Ambulatory blood-pressure
AstraZeneca. AP is an employee and shareholder of Bristol-Myers Squibb. monitoring. N Engl J Med 2006; 354: 2368–74.
20 Kannel WB. Elevated systolic blood pressure as a cardiovascular risk
Acknowledgments factor. Am J Cardiol 2000; 85: 251–55.
This study was funded by Bristol-Myers Squibb and AstraZeneca. 21 NICE. The clinical management of primary hypertension in adults.
Editorial assistance was provided by Helen Brereton inScience Clinical Guideline 127. UK National Institute for Clinical Excellence,
Communications, Springer Healthcare, London, UK, funding support 2011. https://www.nice.org.uk/guidance/cg127 (accessed Oct 15, 2015).
for which was provided by AstraZeneca. 22 Baker WL, Smyth LR, Riche DM, Bourret EM, Chamberlin KW,
References White WB. Effects of sodium-glucose co-transporter 2 inhibitors on
1 American Diabetes Association. Standards of medical care in blood pressure: a systematic review and meta-analysis.
diabetes. Diabetes Care 2015; 38 (suppl 1): S1–S94. J Am Soc Hypertens 2014; 8: 262–75, e9.
2 Muskiet MH, Tonneijck L, Smits MM, Kramer MH, 23 Tikkanen I, Narko K, Zeller C, et al. Empagliflozin reduces blood
Lambers Heerspink HJ, van Raalte DH. Pleiotropic effects of type 2 pressure in patients with type 2 diabetes and hypertension.
diabetes management strategies on renal risk factors. Diabetes Care 2015; 38: 420–28.
Lancet Diabetes Endocrinol 2015; 3: 367–81. 24 Weber MA, Mansfield TA, Alessi F, Ptaszynska A. Effects of
3 Weber MA, Schiffrin EL, White WB, et al. Clinical practice guidelines dapagliflozin on blood pressure in diabetic patients with
for the management of hypertension in the community: a statement hypertension inadequately controlled by a renin-angiotensin system
by the American Society of Hypertension and the International blocker. Circulation 2013; 128: A13144 (abstr).
Society of Hypertension. J Clin Hypertens 2014; 16: 14–26. 25 Sica DA, Carter B, Cushman W, Hamm L. Thiazide and loop
4 Bailey CJ, Day C. SGLT2 inhibitors: glucuretic treatment for type 2 diuretics. J Clin Hypertens 2011; 13: 639–43.
diabetes. Br J Diabetes Vasc Dis 2010; 10: 193–99. 26 Sharma AM, Pischon T, Hardt S, Kunz I, Luft FC. Hypothesis:
5 Meng W, Ellsworth BA, Nirschl AA, et al. Discovery of dapagliflozin: β-adrenergic receptor blockers and weight gain: a systematic
a potent, selective renal sodium-dependent glucose cotransporter 2 analysis. Hypertension 2001; 37: 250–54.
(SGLT2) inhibitor for the treatment of type 2 diabetes. J Med Chem 27 Zinman B, Wanner C, Lachin JM, et al. Empagliflozin, cardiovascular
2008; 51: 1145–49. outcomes, and mortality in type 2 diabetes. N Engl J Med 2015;
6 FDA. Farxiga (dapagliflozin) tablets. Highlights of prescribing published online Sept 17. DOI:10.1056/NEJMoa1504720.
information. US Food and Drug Administration, 2014. http://
www1.astrazeneca-us.com/pi/pi_farxiga.pdf (accessed Oct 15, 2015).

10 www.thelancet.com/diabetes-endocrinology Published online November 24, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00417-9