anti-glycemic agents in 2019: digo · white jr jr. diabetes spectr 2016; 27:82-86. developmental...

Anti-Glycemic Agents in 2019:

Beyond Glucose Control

DrAdrianLiewAssociateProfessorofMedicine

SeniorConsultantDepartmentofRenalMedicine

TanTockSengHospital,Singapore

KDIGO

The Diabetic Epidemic

KDIGO

USRDS ADR 2018.

Diabetic Kidney Disease: Asia leads the way…

KDIGO

White JR Jr. Diabetes Spectr 2016; 27:82-86.

Developmental Milestones for Diabetes Treatment

85Diabetes Spectrum Volume 27, Number 2, 2014

FR

OM

RE

SE

AR

CH

TO

PR

AC

TIC

E/P

HA

RM

AC

OT

HE

RA

PY

OF

DIA

BE

TE

S: P

AS

T, P

RE

SE

NT

, AN

D F

UT

UR

E

by the FDA in 2013. Vildagliptin has been approved for use in Europe but is not available in the United States.

These compounds are associ-ated with an A1C reduction of ~ 0.8%.10 They are weight neutral and do not tend to cause hypogly-cemia.7 However, pancreatitis has been reported in patients treated with DPP-4 inhibitors.11

Amylin AgonistsThe endogenous neuroendocrine hor-mone amylin was discovered in 1987. Amylin is co-secreted with insulin by the β-cells in equimolar amounts.7 Patients with type 2 diabetes have reduced amounts of amylin, whereas patients with type 1 diabetes have essentially no amylin.11 The only amy-lin analog currently on the market is pramlintide, which was approved by the FDA in 2005. Its physiological effect includes weight loss, delayed gastric emptying, and a reduction in both postprandial glucose and gluca-gon. The primary side effect is nausea.

Pramlintide has a modest effect on A1C reduction of ~ 0.5%. This compound is usually reserved for use in patients with type 1 diabetes treated with intensive insulin ther-apy.11 It reduces postprandial glucose excursions via the mechanisms men-tioned above.

BromocriptineBromocriptine is a dopamine ago-nist that was approved for use in the

United States as an antihyperglycemic medication in 2009.12 Its mechanism is not certain but may be related to its dopaminergic activity in the brain and the subsequent inhibition of sym-pathetic tone.11 Its impact on glycemia is modest, with A1C reductions of up to 0.7%.11

ColesevelamColesevelam is an interesting com-pound that has a dual effect of lowering LDL cholesterol and reduc-ing blood glucose levels. This drug was specifically developed for its ability to bind bile acids, effectively removing them from circulation and resulting in reductions in LDL cho-lesterol. The mechanism of action of the glucose lowering observed with this compound is not known. The drug was approved by the FDA for use in patients with type 2 diabetes in 2008.11

Colesevelam is typically associated with an A1C reduction of ~ 0.5% and LDL cholesterol reduction of 13%.7 Its side effects are similar to those encountered with AGIs and are pri-marily gastrointestinal. Also, it should be noted that colesevelam may cause a slight increase in triglycerides.

Sodium Glucose Co-Transporter 2 InhibitorsThe sodium glucose co-transporter 2 (SGLT-2) inhibitors are a novel group of compounds that antagonize a high-capacity, low-affinity glucose

transporter found primarily in the kid-ney.22 This transporter is responsible for ~ 90% of glucose reabsorption in the kidney. When this transporter is antagonized, excess glucose in the renal tubules is not reabsorbed, and glucose is excreted in the urine. This results in a net loss of glucose and a reduction in hyperglycemia.

A recent meta-analysis of placebo-controlled studies evaluating SGLT-2 inhibitors reported A1C reductions of 0.5–0.6% in patients treated with these agents.23 In addition to reduc-ing hyperglycemia, SGLT-2 inhibitors have also been associated with slight reductions in weight and BMI.

The primary side effect of SGLT-2 inhibition is an increase in urinary or genital infections. These infec-tions are much more common than in placebo-treated patients (about four times as many) but are usually mild.23 Canagliflozin was the first SGLT-2 inhibitor to be approved by the FDA, in March 2013.24 Dapagliflozin was approved in the United States in early 2014. Empagliflozin and other SGLT-2 inhibitors are under development.

ConclusionThere are now 11 different categories of medications directed at the man-agement of hyperglycemia in patients with diabetes. These compounds have been developed during the past 90 years (Figure 1), and among these categories, myriad subtypes exist.

Figure 1. History of diabetes medications.

KDIGO

Diabetic Medications: More to come… Bailey CJ et al. Lancet Diabetes Endocrinol 2016; 4:350-359.

KDIGO

Diabetic Treatment: The Era of Personalized Medicine

Davies MJ et al. Diabetologia 2018; 61:2461-2498.

KDIGO

Metformin The Reliable Agent of OldKDIGO

UKPDS Group. Lancet 1998; 352:854-865.

Metformin: First Indication of Benefits beyond Glucose Control

The median duration from the initial randomisationto subsequent randomisation of addition or no additionof metformin was 7·1 years. The median follow-up afterrandomisation was 6·6 years. Vital status was not knownin ten (2%) patients who had emigrated and a furtherfive (1%) who could not be contacted.

Figure 9 shows the median FPG and HbA1c in thecohorts studied for 4 years after second randomisationto addition or no addition of metformin therapycompared with data for all the overweight patients in thecomparison of intensive control with metformin andconventional control. There was a decrease in FPG inpatients on sulphonylurea therapy who were assignedaddition of metformin, whereas FPG concentrations inthose on sulphonylurea therapy alone approached thoseof overweight patients in the conventional treatmentgroup. HbA1c values in patients with addition ofmetformin decreased initially but approached those ofthe patients remaining on sulphonylurea alone after 3

years. The median HbA1c over 4 years in the cohort withaddition of metformin was 7·7% compared with 8·2% inthose on sulphonylurea alone. There were no significantdifferences in bodyweight or plasma insulin between thegroups allocated addition of metformin or continuedsulphonylurea therapy alone.

The patients assigned addition of metformin took thisdrug for 62% of their person-years of follow-up. Forthose randomly assigned continuing sulphonylureaalone, there were 75% of person-years withoutmetformin therapy.

Aggregate and single endpoints (addition of metforminstudy)Figure 10 shows the aggregates of endpoint data andfigure 11 the single endpoint data.

The addition of metformin to sulphonylurea wasassociated with a 96% increased (p=0·039) risk ofdiabetes-related death. Addition of metformin to

ARTICLES

THE LANCET • Vol 352 • September 12, 1998 861

Figure 6: Incidence of clinical endpoints among patients assigned intensive control with metformin (n=342), intensive control withchlorpropamide, glibenclamide, or insulin (intensive; n=951), or conventional control (n=411)Relative risk (RR) is for metformin or intensive group compared with conventional group.

Figure 7: Kaplan-Meier plots in diet/metformin study for microvascular disease (renal failure or death from renal failure, retinopathyrequiring photocoagulation, or vitreous haemorrhage), myocardial infarction (non-fatal and fatal, including sudden death), stroke(non-fatal and fatal) and cataract extractionSimilar plots of data for sulphonylurea/metformin study are superimposed showing relative time of commencement.UKPDS-34Study:Amongstpatientsallocatedtointensiveblood-glucosecontrol,MetformincomparedtoChlorpropamide,glibenclamideorinsulin:-  Greatereffectforanydiabetes-relatedendpoint(p=0.0034)-  All-causemortality(p=0.021)-  Stroke(p=0.032)

KDIGO

Maruthur NM et al. Ann Intern Med 2016; 146:740-751.

Metformin: Purported Benefits

shows the number and design of studies, by outcome.Fifty studies were multicontinental; the others wereconducted in Europe (55 studies), Asia (39 studies), andthe United States (34 studies). Study durations rangedfrom 3 months to 8 years, but only 22 studies (7 RCTs)lasted longer than 2 years. Only 1 RCT specified a car-diovascular outcome as a primary outcome (15).

Study participants were overweight or obese menand women with baseline hemoglobin A1c levels be-tween 7% and 9%. About 45% of the RCTs did not re-port race/ethnicity. When reported, only 10% to 30% ofthe enrolled population was of nonwhite race. Moststudies excluded older persons and those with clini-cally significant comorbid conditions.

Of the randomized trials, approximately one halfreported on their randomization scheme. Sixty-five per-cent reported double-blinding, but most did not reportsteps taken to ensure double-blinding. Of the newlyincluded RCTs, losses to follow-up exceeded 20% inthe majority (>70%) of trials lasting 1 year or more andin 24% of those lasting less than 1 year. Most studiesused the last-observation-carried-forward approach foranalysis of intermediate outcomes. Of the newly in-cluded trials, use of rescue therapy was reported in35%, but 42% did not report on this. All included ob-servational studies were at low risk of bias. Sixty-sevenpercent of studies reported receiving funding frompharmaceutical companies. We did not identify sub-stantive reporting bias that would have affected ourfindings.

All-Cause Mortality and Macrovascular andMicrovascular Outcomes

Although we included 65 new studies (52 RCTs and13 observational studies) for these outcomes in this re-view, the trials were largely 1 year or less in duration,with few or no events; the evidence was insufficient orof low strength for almost all comparisons for these out-comes. However, we found moderate strength of evi-dence that metformin monotherapy was associatedwith lower long-term (≥2 years) cardiovascular mortalitycompared with sulfonylurea monotherapy (Table 1), onthe basis of consistent findings from 2 RCTs (3199 totalparticipants) (15, 16) and 3 observational studies(115 105 total participants) (19, 21, 24) at low risk ofbias.

Both RCTs showed a lower risk for cardiovascularmortality with metformin versus a sulfonylurea (Appen-dix Table 4, available at www.annals.org). The first,ADOPT (A Diabetes Outcome Progression Trial) (16),

was conducted among patients with newly diagnosed,untreated diabetes. The incidence of fatal myocardialinfarction was slightly lower in the metformin group (2of 1454 participants [0.1%]; median follow-up, 4.0years) than the glyburide group [3 of 1441 participants[0.2%]; median follow-up, 3.3 years); losses to follow-upwere greater in the glyburide group than in the met-formin group (44% vs. 38%, respectively) (16). The sec-ond, SPREAD-DIMCAD (Study on the Prognosis and Ef-fect of Antidiabetic Drugs on Type 2 Diabetes Mellituswith Coronary Artery Disease) (15), was a smaller RCTconducted in China among patients with known coro-nary heart disease (clinical evidence of acute myocar-dial infarction or coronary stenosis >50% on angiogra-phy) and also reported a lower risk for cardiovascularmortality for metformin compared with a sulfonylurea(7 of 156 participants [4.5%] vs. 11 of 148 participants[7.4%], respectively) over 5.0 years (15). Losses tofollow-up were 5% in both groups of SPREAD-DIMCAD(15, 29). Cardiovascular mortality was part of the com-posite primary outcome for SPREAD-DIMCAD (15),whereas cardiovascular outcomes were considered ad-verse events and not actively ascertained in ADOPT(16). The 3 high-quality observational studies sup-ported the findings of lower cardiovascular mortalitywith metformin compared with a sulfonylurea (Appen-dix Table 5, available at www.annals.org).

Findings were less consistent across studies of thiscomparison for all-cause mortality and cardiovascularmorbidity. On the basis of the same set of RCTs andadditional observational studies, metformin was associ-ated with lower risk compared with sulfonylureas, butthe evidence was of low strength for these outcomes(Table 1).

All other evidence on these outcomes for all of theother drug comparisons was of low strength or insuffi-cient.

Comparative Effectiveness for IntermediateOutcomesHemoglobin A1c

Most diabetes medications used as monotherapy(metformin, thiazolidinediones, and sulfonylureas) re-duced hemoglobin A1c to a similar degree in the shortterm, except for DPP-4 inhibitors, which were less effec-tive than metformin or sulfonylureas (Figure 1). In the2011 report (6), there were no significant between-group differences in hemoglobin A1c with metforminversus sulfonylureas, and the strength of evidence was

Table 1. Effects of Metformin Compared With Sulfonylurea Monotherapy on Long-Term All-Cause Mortality andCardiovascular Mortality and Morbidity

Outcome Range in RR From RCTs Range in RD From RCTs Adjusted HR From ObservationalStudies

Strength ofEvidence

All-cause mortality 0.5 to 1.0 (2 studies [15, 16]) –5.0% to –0.1% (2 studies [15, 16]) 0.5 to 0.8 (7 studies* [17–23]) LowCVD mortality 0.6 to 0.7 (2 studies [15, 16]) –2.9% to –0.1% (2 studies [15, 16]) 0.6 to 0.9 (3 studies [19, 21, 24]) ModerateCVD morbidity 0.7 to 1.6 (2 studies [15, 16]) –0.4% to 10.1% (2 studies [15, 16]) 0.3 to 0.9 (5 studies† [19, 20, 22, 25, 26]) Low

CVD = cardiovascular disease; HR = hazard ratio; RCT = randomized, controlled trial; RD = risk difference; RR = relative risk.* One additional retrospective cohort study reported an odds ratio of 0.9 (27).† One additional case–control study reported an odds ratio of 0.8 (28).

REVIEW Diabetes Medications as Monotherapy or Metformin-Based Combination Therapy

742 Annals of Internal Medicine • Vol. 164 No. 11 • 7 June 2016 www.annals.org

Downloaded from https://annals.org by National Univ of Singapore user on 02/17/2019

•  EffectiveHbA1creduction•  Lowerriskforseverehypoglycemia•  Beneficialeffectonbodyweight

reduction•  Reduceriskofcardiovasculareventsand

death•  Safeandinexpensive•  Widelyavailableeveninlowresource

settings

KDIGO

ADA Consensus Recommendations


Metabolic surgery

Metabolic surgery is highly effective in improving glucose con-trol [176–178] and often produces disease remission [179–182].The effects can be sustained for at least 5 years [177, 182].Benefits include a reduction in the number of glucose-loweringmedications needed to achieve glycaemic targets [178, 179].

Several clinical practice guidelines and position statementsrecommend consideration of metabolic surgery as a treatmentoption for adults with type 2 diabetes and (1) a BMI ≥40.0 kg/m2 (BMI ≥37.5 kg/m2 in people of Asian ancestry) or (2) aBMI of 35.0–39.9 kg/m2 (32.5–37.4 kg/m2 in people of Asianancestry) who do not achieve durable weight loss and im-provement in comorbidities with reasonable non-surgicalmethods [65, 183]. Because baseline BMI does not predictsurgical benefits on glycaemia or hard outcomes and the im-provement in glyacemic control occurs early through weight-independent mechanisms [183], metabolic surgery may beconsidered for those with a BMI of 30.0–34.9 kg/m2 (27.5–32.4 in people of Asian ancestry) who do not achieve durableweight loss and improvement in comorbidities with reason-able non-surgical methods.

Adverse effects of bariatric surgery which vary by proce-dure include surgical complications (e.g. anastomotic or stapleline leaks, gastrointestinal bleeding, intestinal obstruction, theneed for re-operation), late metabolic complications (e.g. pro-tein malnutrition, mineral deficiency, vitamin deficiency,anaemia, hypoglycaemia) and gastroesophageal reflux [184,185]. Patients who undergo metabolic surgery may be at riskfor substance use, including drug and alcohol use and cigarettesmoking [186]. People with diabetes presenting for metabolicsurgery also have increased rates of depression and other ma-jor psychiatric disorders [187]. These factors should beassessed pre-operatively and during follow-up. Metabolic sur-gery should be performed in high-volume centres with multi-disciplinary teams that are experienced in the management ofdiabetes and gastrointestinal surgery. Long-term lifestyle sup-port and routine monitoring of micronutrient and nutritionalstatus must be provided to patients after surgery [188, 189].

Putting it all together: strategies for implementation

For an increasing number of patients, presence of specific co-morbidities (e.g. ASCVD, HF, CKD, obesity), safety concerns

(e.g. risk of hypoglycaemia) or healthcare environment (e.g.cost of medications) mandate a specific approach to the choiceof glucose-lowering medication. These are considered in Figs2, 3, 4, 5, 6. For patients not reaching their target HbA1c, it isimportant to re-emphasise lifestyle measures, assess adherenceand arrange timely follow-up (e.g. within 3–6 months) (Fig. 1).

Initial monotherapy

Metformin remains the preferred option for initiating glucose-lowering medication in type 2 diabetes and should be added tolifestyle measures in newly diagnosed patients. This recom-mendation is based on the efficacy, safety, tolerability, lowcost and extensive clinical experience with this medication.Results from a substudy of the UKPDS (n = 342) showedbenefits of initial treatment with metformin on clinical out-comes related to diabetes, with less hypoglycaemia andweight gain than with insulin or sulfonylureas [98].

Initial combination therapy compared with stepwiseaddition of glucose-lowering medication

In most patients, type 2 diabetes is a progressive disease, aconsequence generally attributed to a steady decline of insulinsecretory capacity. The practical impact of gradual loss of betacell function is that achieving a glycaemic target with mono-therapy is typically limited to several years. Stepwise therapy(i.e. adding medications to metformin to maintain HbA1c attarget) is supported by clinical trials [3]. While there is somesupport for initial combination therapy due to the greater initialreduction of HbA1c than can be provided by metformin alone[190, 191], there is little evidence that this approach is superiorto sequential addition of medications for maintainingglycaemic control, or slowing the progression of diabetes.However, since the absolute effectiveness of most oral medi-cations rarely exceeds an 11 mmol/mol (1%) reduction inHbA1c, initial combination therapy may be considered in pa-tients presenting with HbA1c levels more than 17 mmol/mol(1.5%) above their target. Fixed-dose formulations can im-provemedication adherencewhen combination therapy is used[192], and may help achieve glycaemic targets more rapidly[100]. Potential benefits of combination therapy need to beweighed against the exposure of patients to multiple

Consensus recommendation Metformin is the preferred initial glucose-lowering medication for most people with type 2 diabetes.

Consensus recommendation The stepwise addition of glucose-lowering medication is generally preferred to initial combination therapy.

Consensus recommendation Metabolic surgery is a recommended treatment option for adults with type 2 diabetes and (1) a BMI ≥ 40.0 kg/m2 (BMI ≥ 37.5 kg/m2 in people of Asian ancestry) or (2) a BMI of 35.0–39.9 kg/m2 (32.5–37.4 kg/m2 in people of Asian ancestry) who do not achieve durable weightloss and improvement in comorbidities with reasonable non-surgical methods.

2484 Diabetologia (2018) 61:2461–2498

KDIGO

The New Wonder Kid on the Block

SGLT2 Inhibitors

KDIGO

Heerspink HJL et al. Circulation 2016; 134:752-772.

Heerspink et al

September 6, 2016 Circulation. 2016;134:752–772. DOI: 10.1161/CIRCULATIONAHA.116.021887754

in HbA1c, in comparison with the sulfonylurea, there was a substantial mean 4.5-kg reduction in body weight and a 90% lower risk of hypoglycemia with the SGLT2 inhibi-tor.16 Similar results were observed for canagliflozin,17 whereas a longer 208-week study of dapagliflozin confirmed this metabolic benefit (HbA1c lowering was 0.3% lower than the active comparator glipizide), with substantial benefits in terms of blood pressure (BP)–lowering and eGFR preservation.18 Placebo-controlled studies of SGLT2 inhibition in patients on complex insu-lin regimens similarly revealed further HbA1c reduction, lower insulin dose requirement, and greater weight loss, each observed without additional hypoglycemia.19–21 This extensive clinical trial record emphasizes the series of metabolic benefits observed when an SGLT2 inhibitor is used particularly in place of an insulin-providing agent such as a sulfonylurea, or when used to attenuate total daily insulin doses.

SUMMARY OF ADVERSE EFFECTS OF SGLT2 INHIBITIONThere is a series of well-recognized and suspected ad-verse events associated with augmenting the presence of glucose and sodium in urine. Although a mild-to-mod-est increase in the incidence of urinary tract infections has been observed in some studies,10 but not other large-scale studies,22 the strong causal relationship with mycotic genital infections, usually with Candida spe-cies, is indisputably the most common adverse event associated with SGLT2 inhibition.10 The magnitude of the absolute risk increment ranges between ≈3% in the pla-cebo versus ≈9% to 18% with the active comparators in women,20 and about half those rates have been reported in men.22

The risk of volume depletion arising from the osmotic diuresis of glucose, and from natriuresis, represents an

infrequent but clear causal adverse event from SGLT2 in-hibition, more common in older adults, among those who use higher doses of the SGLT2 inhibitors, use loop di-uretics, and have kidney dysfunction.20,23 However, such an increase was not observed in EMPA-REG OUTCOME, which was a large-scale controlled clinical trial involving patients with established cardiovascular disease, as dis-cussed below.22

Earlier concerns for a numerically higher case finding of bladder cancer, particularly in the dapagliflozin devel-opment program, and of breast cancer, have not been consistent, and the totality of evidence does not suggest a causal relationship.10,24 Rather, one potential specula-tion is that of a differential surveillance bias induced by SGLT2 inhibition. For example, any increase in urinary symptoms, such as urinary tract or genital infections, may increase the likelihood of investigations that identify bladder cancer, and a loss in weight may increase the likelihood of identifying breast lesions.

Although small mean changes in electrolytes have been described in a controlled trial of canagliflozin,25 such changes have not been observed in a large-scale study in which plasma levels of sodium, potassium, calcium, magnesium, and phosphate did not differ in >2300 participants exposed to each of placebo and low- and high-dose empagliflozin.22 However, hyperka-lemia may be seen among those who develop kidney dysfunction and who are concomitantly exposed to an-tihypertensive medications that may elevate potassium levels, independent of SGLT2 inhibitor use.26 Early re-ports of minor elevations in serum phosphate26 raised concerns of an exaggerated phosphate resorption in-duced by distal sodium delivery, which could explain an increase in parathyroid hormone and its consequent ef-fect on bone turnover, density, and fracture risk. How-ever, observations, made primarily in the canagliflozin trial program, have been inconsistent and inconclusive regarding a risk of bone loss, and the small loss in

Figure 1. The sodium-glucose cotransporter-2 (SGLT2) mecha-nism in the proximal tubule. Modified from Bakris et al4 with per-mission of the publisher. Copyright © 2009, Elsevier.

by guest on March 8, 2018

http://circ.ahajournals.org/D

ownloaded from

SGLT2 Inhibition

KDIGO

Heerspink et al

September 6, 2016 Circulation. 2016;134:752–772. DOI: 10.1161/CIRCULATIONAHA.116.021887756

fect of combining the SGLT2 inhibitor canagliflozin with a thiazide, nor does the combination produce a greater natriuretic effect in comparison with either drug alone.45 Similar to this report using canagliflozin,45 the addition of a thiazide diuretic to dapagliflozin does not yield ad-ditional antihypertensive effect, whereas β-blocker and calcium channel blocker combinations with dapagliflozin did accentuate the degree of BP lowering.46 Similarly, the addition of dapagliflozin to a background of a renin-angiotensin-aldosterone system (RAAS) inhibitor in pa-tients with T2D lowers systolic BP by an additional 3 to 4 mm Hg,47 an effect that has been supported by results from animal studies.48 Whether this apparent synergy is on the basis of plasma volume contraction leading to

RAAS activation, which is then inhibited pharmacological-ly by angiotensin-converting enzyme (ACE) inhibition or an angiotensin receptor blocker, resulting in enhanced BP lowering is not known. Alternatively, angiotensin II in-creases SGLT2 mRNA expression and proximal tubular sodium uptake, perhaps promoting volume expansion and worsening hypertension.49 It is therefore conceiv-able that the greater BP-lowering effect observed with the combination of an SGLT2 inhibitor with a RAAS-block-ing agent versus either agent alone could represent suppressed proximal tubular sodium reabsorption and volume contraction.

Although the mechanisms responsible for the BP-low-ering effects of the SGLT2 inhibitors are still being eluci-

Figure 2. Physiologic mechanisms implicated in the cardiovascular and renal protection with SGLT2 inhibition. HbA1c indicates hemoglobin A1c; and SGLT2, sodium-glucose cotransporter-2.

by guest on March 8, 2018

http://circ.ahajournals.org/Downloaded from

Mechanisms for SGLT2i Protective Effects

Heerspink HJL et al. Circulation 2016; 134:752-772.

KDIGO

SGLT2 Inhibitors: RCTs beyond Glucose Control

Canagliflozin DapagliflozinEmpagliflozin

CANVASTrial(n=10,142)

NEngJMed2017;377:644-657.

DECLARE-TIMITrial(n=10,186)

NEngJMed2019;380:347-357.

EMPA-REGOUTCOMETrial

(n=7,020)

NEnglJMed2015;373:2117-2128.NEngJMed2016;375:323-334.

PrimaryOutcome:1.  CompositeofDeathfromcardiovascularcauses,nonfatalmyocardialinfarction,

nonfatalstroke(EMPA-REGOUTCOMEandCANVASTrials)2.  MACEandacompositeofcardiovasculardeathorhospitalizationforheartfailure.

(DECLARE-TIMITrial)

CREDENCETrial-Canagliflozin

PrimaryOutcome:1.  Compositeofend-stagekidneydisease(dialysis,transplantation,orasustained

estimatedGFR<15ml/min/1.73m2),doublingofserumcreatinine,ordeathfromrenalorcardiovascularcauses.

KDIGO

Efficacy of SGLT2 Inhibitors: CANVAS Study

Neal B et al. N Engl J Med 2017; 377:644–657.

n en

gl j m

ed 377;7

nejm

.org

Au

gu

st 17, 2017650

The n

ew

en

gl

an

d jo

ur

na

l of m

ed

icin

e

Figure 1. Effects of Canagliflozin on Glycated Hemoglobin Level, Body Weight, and Systolic and Diastolic Blood Pressure in the Integrated CANVAS Program.

A total of 10,142 participants were included in the CANVAS Program, which comprised two trials: the Canagliflozin Cardiovascular Assessment Study (CANVAS) and CANVAS–Re-nal (CANVAS-R).

Adju

sted

Mea

n G

lyca

ted

Hem

oglo

bin

(%) 8.4

8.18.0

8.28.3

7.67.77.87.9

7.5

7.37.2

7.0

7.4

7.1

0.0

C Systolic Blood Pressure

A Glycated Hemoglobin

No. of PatientsPlaceboCanagliflozin

42315644

39875329

38545211

35394864

28914228

15612778

10142206

8781965

8992042

7831797

8051889

7261690

Adju

sted

Mea

n Bo

dy W

eigh

t (kg

)

92

91

90

88

87

85

89

86

0Base-line

12/1312/13

26 52 78 104 338

Weeks since Randomization

130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312

6951661

245556

B Body Weight

D Diastolic Blood Pressure


42455651

40245344

39315277

36925044

29774331

16232877

10362247

9352041

9202086

8341902

8261928

7611775

7141669

252567

Adju

sted

Mea

n Sy

stol

ic B

lood

Pre

ssur

e(m

m H

g)

138

136

137

135

134

132

131

127

128

129

133

130

0


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

Adju

sted

Mea

n D

iast

olic

Blo

od P

ress

ure

(mm

Hg)

79

78

75

74

72

76

77

73

0


Base-line

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312


7131675

252567


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

7131675

252567

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

The New

England Journal of Medicine

Dow

nloaded from nejm

.org at NA

TION

AL U

NIV

OF SIN

GA

PORE-CEN

TRAL LIBRA

RY on A

pril 19, 2019. For personal use only. No other uses w

ithout permission.

Copyright © 2017 M

assachusetts Medical Society. A

ll rights reserved.

n en

gl j m

ed 377;7

nejm

.org

Au

gu

st 17, 2017650

The n

ew

en

gl

an

d jo

ur

na

l of m

ed

icin

e



Adju

sted

Mea

n G

lyca

ted

Hem

oglo

bin

(%) 8.4

8.18.0

8.28.3

7.67.77.87.9

7.5

7.37.2

7.0

7.4

7.1

0.0




42315644

39875329

38545211

35394864

28914228

15612778

10142206

8781965

8992042

7831797

8051889

7261690

Adju

sted

Mea

n Bo

dy W

eigh

t (kg

)

92

91

90

88

87

85

89

86

0Base-line

12/1312/13

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312

6951661

245556

B Body Weight



42455651

40245344

39315277

36925044

29774331

16232877

10362247

9352041

9202086

8341902

8261928

7611775

7141669

252567

Adju

sted

Mea

n Sy

stol

ic B

lood

Pre

ssur

e(m

m H

g)

138

136

137

135

134

132

131

127

128

129

133

130

0


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

Adju

sted

Mea

n D

iast

olic

Blo

od P

ress

ure

(mm

Hg)

79

78

75

74

72

76

77

73

0


Base-line

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312


7131675

252567


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

7131675

252567

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

The New


Dow

nloaded from nejm

.org at NA

TION

AL U

NIV

OF SIN

GA

PORE-CEN

TRAL LIBRA

RY on A


ithout permission.

Copyright © 2017 M


ll rights reserved.

n en

gl j m

ed 377;7

nejm

.org

Au

gu

st 17, 2017650

The n

ew

en

gl

an

d jo

ur

na

l of m

ed

icin

e



Adju

sted

Mea

n G

lyca

ted

Hem

oglo

bin

(%) 8.4

8.18.0

8.28.3

7.67.77.87.9

7.5

7.37.2

7.0

7.4

7.1

0.0




42315644

39875329

38545211

35394864

28914228

15612778

10142206

8781965

8992042

7831797

8051889

7261690

Adju

sted

Mea

n Bo

dy W

eigh

t (kg

)

92

91

90

88

87

85

89

86

0Base-line

12/1312/13

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312

6951661

245556

B Body Weight



42455651

40245344

39315277

36925044

29774331

16232877

10362247

9352041

9202086

8341902

8261928

7611775

7141669

252567

Adju

sted

Mea

n Sy

stol

ic B

lood

Pre

ssur

e(m

m H

g)

138

136

137

135

134

132

131

127

128

129

133

130

0


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

Adju

sted

Mea

n D

iast

olic

Blo

od P

ress

ure

(mm

Hg)

79

78

75

74

72

76

77

73

0


Base-line

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312


7131675

252567


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

7131675

252567

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

The New


Dow

nloaded from nejm

.org at NA

TION

AL U

NIV

OF SIN

GA

PORE-CEN

TRAL LIBRA

RY on A


ithout permission.

Copyright © 2017 M


ll rights reserved.

n en

gl j m

ed 377;7

nejm

.org

Au

gu

st 17, 2017650

The n

ew

en

gl

an

d jo

ur

na

l of m

ed

icin

e



Adju

sted

Mea

n G

lyca

ted

Hem

oglo

bin

(%) 8.4

8.18.0

8.28.3

7.67.77.87.9

7.5

7.37.2

7.0

7.4

7.1

0.0




42315644

39875329

38545211

35394864

28914228

15612778

10142206

8781965

8992042

7831797

8051889

7261690

Adju

sted

Mea

n Bo

dy W

eigh

t (kg

)

92

91

90

88

87

85

89

86

0Base-line

12/1312/13

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312

6951661

245556

B Body Weight



42455651

40245344

39315277

36925044

29774331

16232877

10362247

9352041

9202086

8341902

8261928

7611775

7141669

252567

Adju

sted

Mea

n Sy

stol

ic B

lood

Pre

ssur

e(m

m H

g)

138

136

137

135

134

132

131

127

128

129

133

130

0


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

Adju

sted

Mea

n D

iast

olic

Blo

od P

ress

ure

(mm

Hg)

79

78

75

74

72

76

77

73

0


Base-line

26 52 78 104 338


130 156 182 208 234 260 286 312

Base-line

12/13

26 52 78 104 338


130 156 182 208 234 260 286 312


7131675

252567


42475652

40325355

39455293

37075049

29794338

16292883

10382255

9392049

9222092

8361908

8281936

7631782

7131675

252567

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

Placebo

Canagliflozin

The New


Dow

nloaded from nejm

.org at NA

TION

AL U

NIV

OF SIN

GA

PORE-CEN

TRAL LIBRA

RY on A


ithout permission.

Copyright © 2017 M


ll rights reserved.

KDIGO

Cardiovascular Protection with SGLT2 Inhibitors

Zeiniker TA et al. Lancet 2019; 393:31–39.

Articles

www.thelancet.com Vol 393 January 5, 2019 33

of randomised treatment allocation on the primary outcomes across trials overall and within the previously mentioned subgroup strata using fixed effects models. We tested for treatment effect modification by subgroup using random effects models, applying the method of residual maximum likelihood and Hartung-Knapp adjustment.20 All trials met criteria for being well done and had a low risk of bias according to the Cochrane tool for assessing risk of bias in randomised clinical trials (appendix).21

We assessed heterogeneity using Cochrane Q statistic, and Higgins and Thompsons’ I². Heterogeneity was considered to be low if I²=25%, moderate if I²=50%, or high if I²=75%.22 All reported p values are two-sided and we did no adjustments for multiple testing. We did statistical analyses using R version 3.5.1 (R Core Team, Vienna, Austria) and the R package metafor (version 2.0-0).23

Role of the funding sourceThere was no funding source for this study. All authorshad full access to all the data in the study and thecorresponding author had final responsibility for thedecision to submit for publication.

ResultsWe identified a total of three trials1–3 and six secondary analyses7,9–11,24,25 from the same trials that were eligible for inclusion (appendix). The appendix has an overview of the search and the selection process. In total, data from 34 322 patients were included. The mean age was 63·5 years and 35·1% were women (table). A total of 20 650 (60·2%) patients were known to have athero-sclerotic cardiovascular disease and 13 672 (39·8%) had multiple risk factors but without known atherosclerotic cardiovascular disease. The proportion of patients with multiple risk factors differed among the trials, ranging from 0% in the EMPA-REG OUTCOME trial to 34% in the CANVAS Program, and to 59% in DECLARE-TIMI 58 trial. A total of 3891 (11·3%) patients had a history of heart failure, a proportion that was similar across all three trials. Baseline renal function differed among the trials, with the proportion of patients with eGFR

<60 ml/min per 1·73 m² ranging from 25·9% in EMPA-REG OUTCOME to 20·1% in CANVAS, and to 7·4% in DECLARE-TIMI 58 (table).

In total, 3342 (9·7%) of 34 322 patients had a major adverse cardiac event in the trials. Of those events, 2588 (77·4%) occurred in the group with established atherosclerotic cardiovascular disease. Overall, SGLT2i reduced the risk of a major adverse cardiac event by 11% (HR 0·89 [95% CI 0·83–0·96], p=0·0014; appendix). However, this effect was entirely restricted to a 14% reduction in patients with atherosclerotic cardio-vascular disease (0·86 [0·80 to 0·93]), whereas no treatment effect was found in patients with multiple risk factors (1·00 [0·87–1·16], p for interaction=0·0501; figure 1).

1604 (4·7%) patients had a myocardial infarction (80·5% of which occurred in patients with atherosclerotic cardiovascular disease), 1060 (3·1%) had a stroke (73·1% of which occurred in patients with atherosclerotic cardiovascular disease), and 1256 (3·7%) had cardio-vascular death (78·6% of which occurred in patients with the disease). Overall, SGLT2i reduced the risk of myocardial infarction by 11% (HR 0·89 [95% CI 0·80–0·98], p=0·0177) and cardiovascular death by 16% (0·84 [0·75–0·94], p=0·0023, but with high heterogeneity [I²=79·9%]), whereas SGLT2i had no effect on stroke (0·97 [0·86–1·10], p=0·64; appendix). Analogous to the pattern seen for major adverse cardiovascular events overall, SGLT2i reduced myocardial infarction (0·85 [0·76–0·95]) and cardio vascular death (0·80 [0·71–0·91]) in patients with atherosclerotic cardiovascular disease, whereas no treatment effect was found in patients with multiple risk factors. SGLT2i had no effect on stroke, even in patients with atherosclerotic cardiovascular disease (appendix).

Overall, SGLT2i significantly reduced the risk for the composite of cardiovascular death or hospitalisation for heart failure by 23% (HR 0·77 [95% CI 0·71–0·84], p<0·0001), and hospitalisation for heart failure by 31% (0·69 [0·61–0·79], p<0·0001; appendix). In patients with atherosclerotic cardiovascular disease, the HR for

EMPA-REG OUTCOME1 CANVAS Program2 DECLARE-TIMI 583

Drug Empagliflozin Canagliflozin Dapagliflozin

Doses analysed 10 mg, 25 mg (once daily) 100 mg, 300 mg (once daily) 10 mg (once daily)

Median follow-up time, years 3·1 2·4 4·2

Trial participants 7020 10 142 17 160

Age, mean 63·1 63·3 63·9

Women 2004 (28·5%) 3633 (35·8%) 6422 (37·4%)

Patients with established atherosclerotic cardiovascular disease 7020 (100%) 6656 (65·6%) 6974 (40·6%)

Patients with a history of heart failure 706 (10·1%) 1461 (14·4%) 1724 (10·0%)

Patients with eGFR <60 mL/min per 1·73 m² 1819 (25·9%) 2039 (20·1%) 1265 (7·4%)

Data are n (%) unless otherwise specified. The CANVAS Program consisted of two trials, CANVAS and CANVAS-R, but are presented combined. eGFR=estimated glomerular filtration rate.

Table: Randomised controlled phase 3/4 clinical trials of sodium-glucose cotransporter-2 inhibitors

Articles

34 www.thelancet.com Vol 393 January 5, 2019

the composite of cardiovascular death or hospitalisation for heart failure was 0·76 (0·69–0·84) and in patients with multiple risk factors it was 0·84 (0·69–1·01, p for interaction=0·41; figure 2). The effect on hospitalisation for heart failure alone was robust, with an approximately 30% reduction in relative risk in both subgroups (appendix). The reduction in the composite of cardio-vascular death or hospitalisation for heart failure was not statistically different in patients with (HR 0·71 [95% CI 0·61–0·84]) or without (0·79 [0·71–0·88]) a history of heart failure at baseline (p for interaction=0·51; figure 3), nor were the individual component outcomes (appendix).

Overall, SGLT2i significantly reduced the risk for all-cause death by 15% (HR 0·85 [95% CI 0·78–0·93], p=0·0002), but with high heterogeneity (I²=75·2%; appendix). In patients with atherosclerotic cardiovascular disease the HR was 0·83 (0·75–0·92) and in those with multiple risk factors it was 0·90 (0·77–1·05, p for interaction=0·69; appendix). Similarly, in patients with a history of heart failure the HR was 0·80 (0·67–0·95) and

in those without a history of heart failure it was 0·88 (0·80–0·97, p for interaction=0·63; appendix).

Overall, SGLT2i were renoprotective and reduced the composite of worsening of renal function, end-stage renal disease, or renal death by 45% (HR 0·55 [95% CI 0·48–0·64], p<0·0001). This effect was similarly robust both in patients with atherosclerotic cardiovascular disease (HR 0·56 [95% CI 0·47– 0·67]) and those with multiple risk factors, (0·54 [0·42–0·71], p for interaction=0·71; figure 4; appendix). The reduction in the composite renal endpoint was present across all baseline eGFR levels but was greatest in those with preserved renal function at baseline, with a 33% reduction in patients with an eGFR of less than 60 mL/min per 1·73 m², 44% reduction in patients with an eGFR between 60 and 90 mL/min per 1·73 m², and 56% reduction in patients with an eGFR of 90 mL/min per 1·73 m² or higher (p for interaction=0·0258; figure 5A). Conversely, the reduction in hospitalisation for heart failure was 40% in the lowest group, 31% in the middle group, and a non-significant

Figure 2: Meta-analysis of SGLT2i trials on hospitalisation for heart failure and cardiovascular death stratified by the presence of established atherosclerotic cardiovascular diseaseAtherosclerotic cardiovascular disease: Q statistic=3·49, p=0·17, I²=42·7%; multiple risk factors: Q statistic=0·00, p=0·96, I²=0%. The p value for subgroup differences was 0·41. Tests for subgroup differences were based on F tests in a random effect meta-regression estimated using restricted maximum likelihood and Hartung Knapp adjustment. HR=hazard ratio. SGLT2i=sodium-glucose cotransporter-2 inhibitors.

HR (95% CI)Events per 1000 patient-years

Weight(%)

HREventsPatients

Treatment (n) TreatmentPatients with atherosclerotic cardiovascular diseaseEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for atherosclerotic cardiovascular disease (p<0·0001)

Patients with multiple risk factorsCANVAS ProgramDECLARE-TIMI 58Fixed effects model for multiple risk factors (p=0·0634)

4687 3756 3474

2039 5108

Placebo (n)

2333 2900 3500

1447 5078

Placebo

30·1 27·4 23·9

9·8 8·4

30·9 32·8 36·4

30·2 69·8

19·7 21·0 19·9

8·9 7·0

463 524 597

128316

0·66 (0·55–0·79)0·77 (0·65–0·92)0·83 (0·71–0·980·76 (0·69–0·84)

0·83 (0·58–1·19)0·84 (0·67–1·04)0·84 (0·69–1·01)

1·000·500·35 2·50

Favours placeboFavours treatment


Weight(%)

HREventsPatients

Treatment (n) TreatmentPatients with atherosclerotic cardiovascular diseaseEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for atherosclerotic cardiovascular disease (p=0·0002)


4687 3756 3474

2039 5108

Placebo (n)

2333 2900 3500

1447 5078

Placebo

43·9 41·3 41·0

15·5 13·3

29·4 32·4 38·2

25·9 74·1

37·4 34·1 36·8

15·8 13·4

772 796 1020

215539

0·86 (0·74–0·99)0·82 (0·72–0·95)0·90 (0·79–1·02)0·86 (0·80–0·93)

0·98 (0·74–1·30)1·01 (0·86–1·20)1·00 (0·87–1·16)

1·000·500·35 2·50


Figure 1: Meta-analysis of SGLT2i trials on the composite of myocardial infarction, stroke, and cardiovascular death (major adverse cardiovascular events) stratified by the presence of established atherosclerotic cardiovascular diseaseNo heterogeneity was found in terms of between-study variance in the subgroups (atherosclerotic cardiovascular disease: Q statistic=0·94, p=0·63, I²=0%; multiple risk factors: Q statistic=0·03, p=0·86, I²=0%). Tests for subgroup differences were based on F tests in a random effect meta-regression estimated using restricted maximum likelihood and Hartung Knapp adjustment. The p value for subgroup differences was 0·0501. HR=hazard ratio. SGLT2i=sodium-glucose cotransporter-2 inhibitors.

Meta-analysisofSGLT2itrialsonthecompositeofMI,StrokeandCVdeath(MACE)

KDIGO

Cardiovascular Protection with SGLT2 Inhibitors

Zeiniker TA et al. Lancet 2019; 393:31–39.

Articles

34 www.thelancet.com Vol 393 January 5, 2019

the composite of cardiovascular death or hospitalisation for heart failure was 0·76 (0·69–0·84) and in patients with multiple risk factors it was 0·84 (0·69–1·01, p for interaction=0·41; figure 2). The effect on hospitalisation for heart failure alone was robust, with an approximately 30% reduction in relative risk in both subgroups (appendix). The reduction in the composite of cardio-vascular death or hospitalisation for heart failure was not statistically different in patients with (HR 0·71 [95% CI 0·61–0·84]) or without (0·79 [0·71–0·88]) a history of heart failure at baseline (p for interaction=0·51; figure 3), nor were the individual component outcomes (appendix).

Overall, SGLT2i significantly reduced the risk for all-cause death by 15% (HR 0·85 [95% CI 0·78–0·93], p=0·0002), but with high heterogeneity (I²=75·2%; appendix). In patients with atherosclerotic cardiovascular disease the HR was 0·83 (0·75–0·92) and in those with multiple risk factors it was 0·90 (0·77–1·05, p for interaction=0·69; appendix). Similarly, in patients with a history of heart failure the HR was 0·80 (0·67–0·95) and

in those without a history of heart failure it was 0·88 (0·80–0·97, p for interaction=0·63; appendix).

Overall, SGLT2i were renoprotective and reduced the composite of worsening of renal function, end-stage renal disease, or renal death by 45% (HR 0·55 [95% CI 0·48–0·64], p<0·0001). This effect was similarly robust both in patients with atherosclerotic cardiovascular disease (HR 0·56 [95% CI 0·47– 0·67]) and those with multiple risk factors, (0·54 [0·42–0·71], p for interaction=0·71; figure 4; appendix). The reduction in the composite renal endpoint was present across all baseline eGFR levels but was greatest in those with preserved renal function at baseline, with a 33% reduction in patients with an eGFR of less than 60 mL/min per 1·73 m², 44% reduction in patients with an eGFR between 60 and 90 mL/min per 1·73 m², and 56% reduction in patients with an eGFR of 90 mL/min per 1·73 m² or higher (p for interaction=0·0258; figure 5A). Conversely, the reduction in hospitalisation for heart failure was 40% in the lowest group, 31% in the middle group, and a non-significant

Figure 2: Meta-analysis of SGLT2i trials on hospitalisation for heart failure and cardiovascular death stratified by the presence of established atherosclerotic cardiovascular diseaseAtherosclerotic cardiovascular disease: Q statistic=3·49, p=0·17, I²=42·7%; multiple risk factors: Q statistic=0·00, p=0·96, I²=0%. The p value for subgroup differences was 0·41. Tests for subgroup differences were based on F tests in a random effect meta-regression estimated using restricted maximum likelihood and Hartung Knapp adjustment. HR=hazard ratio. SGLT2i=sodium-glucose cotransporter-2 inhibitors.


Weight(%)

HREventsPatients

Treatment (n) TreatmentPatients with atherosclerotic cardiovascular diseaseEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for atherosclerotic cardiovascular disease (p<0·0001)


4687 3756 3474

2039 5108

Placebo (n)

2333 2900 3500

1447 5078

Placebo

30·1 27·4 23·9

9·8 8·4

30·9 32·8 36·4

30·2 69·8

19·7 21·0 19·9

8·9 7·0

463 524 597

128316

0·66 (0·55–0·79)0·77 (0·65–0·92)0·83 (0·71–0·980·76 (0·69–0·84)

0·83 (0·58–1·19)0·84 (0·67–1·04)0·84 (0·69–1·01)

1·000·500·35 2·50



Weight(%)

HREventsPatients

Treatment (n) TreatmentPatients with atherosclerotic cardiovascular diseaseEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for atherosclerotic cardiovascular disease (p=0·0002)


4687 3756 3474

2039 5108

Placebo (n)

2333 2900 3500

1447 5078

Placebo

43·9 41·3 41·0

15·5 13·3

29·4 32·4 38·2

25·9 74·1

37·4 34·1 36·8

15·8 13·4

772 796 1020

215539

0·86 (0·74–0·99)0·82 (0·72–0·95)0·90 (0·79–1·02)0·86 (0·80–0·93)

0·98 (0·74–1·30)1·01 (0·86–1·20)1·00 (0·87–1·16)

1·000·500·35 2·50


Figure 1: Meta-analysis of SGLT2i trials on the composite of myocardial infarction, stroke, and cardiovascular death (major adverse cardiovascular events) stratified by the presence of established atherosclerotic cardiovascular diseaseNo heterogeneity was found in terms of between-study variance in the subgroups (atherosclerotic cardiovascular disease: Q statistic=0·94, p=0·63, I²=0%; multiple risk factors: Q statistic=0·03, p=0·86, I²=0%). Tests for subgroup differences were based on F tests in a random effect meta-regression estimated using restricted maximum likelihood and Hartung Knapp adjustment. The p value for subgroup differences was 0·0501. HR=hazard ratio. SGLT2i=sodium-glucose cotransporter-2 inhibitors.

Meta-analysisofSGLT2itrialsonhospitalizationforheartfailureandCVdeath,stratifiedbyatheroscleroticcardiovasculardiseasesandhistoryofheartfailure.

Articles

www.thelancet.com Vol 393 January 5, 2019 35

12% in the highest group (p for interaction=0·0073; figure 5B). A directionally similar but non-significant trend was found for effect modification for major adverse cardiovascular events, with an 18% reduction in the lowest eGFR group, a 9% reduction in the middle group, and a non-significant 6% reduction in the highest group (p for interaction=0·23; figure 5C).

For safety outcomes, an increased risk of amputations and fractures was observed only in one trial (appendix), resulting in moderate to high percentages of total variation across studies that was due to heterogeneity (I²=79·1% for amputation and I²=42·1% for fracture). Diabetic ketoacidosis showed a consistent increased risk of almost two times higher in patients given SGLT2i than those given placebo (2·20 [1·25–3·87], p=0·0060), but the event rates were low (<one per 1000 patient-years; appendix).

DiscussionThe present meta-analysis of SGLT2i cardiovascular outcome trials substantially expands on previous

meta-analyses,26 and the totality of these data now makes several patterns clear. First, SGLT2i have their greatest and most consistent effect on reducing the relative risk of hospitalisation for heart failure (31%) and of progression of renal disease (45%). Their effect on the composite atherosclerotic outcome of myocardial infarction, stroke, or cardiovascular death (major adverse cardiac events), originally a safety outcome stemming from regulatory guidance, was more modest but still significant with 11% reduction in relative risk. Second, for particular outcomes the clinical effects of SGLT2i depend on the patient population in which they are used. The reduction in major adverse cardiac events was apparent only in patients with established atherosclerotic cardiovascular disease, whereas no effect was observed in patients without atherosclerotic cardiovascular disease. Con-versely, the reduction in hospitalisation for heart failure was robust and of similar magnitude regardless of the presence of established atherosclerotic cardiovascular disease or a history of heart failure. The reduction in

Figure 3: Meta-analysis of SGLT2i trials on hospitalisation for heart failure and cardiovascular death stratified by history of heart failureHistory of heart failure: Q statistic=2·02, p=0·37, I²=0·8%; no history of heart failure: Q statistic=5·89, p=0·0527, I²=66%. The p value for subgroup differences was 0·51. Tests for subgroup differences were based on F tests in a random effect meta-regression estimated using restricted maximum likelihood and Hartung Knapp adjustment. HR=hazard ratio. SGLT2i=sodium-glucose cotransporter-2 inhibitors.


Weight(%)

HREventsPatients

TreatmentPatients with history of heart failureEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for history of heart failure (p<0·0001)

Patients with no history of heart failureEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for no history of heart failure (p<0·0001)

462 803 852

4225 4992 7730

244 658 872

2089 3689 7706

PlaceboTreatment (n) Placebo (n)

85·5 56·8 55·5

24·9 15·2 10·5

23·6 34·1 42·4

30·0 32·4 37·6

63·6 35·4 45·1

15·5 13·6 8·9

124 203 314

339449599

0·72 (0·50–1·04) 0·61 (0·46–0·80) 0·79 (0·63–0·99) 0·71 (0·61–0·84)

0·63 (0·51–0·78) 0·87 (0·72–1·06) 0·84 (0·72–0·99) 0·79 (0·71–0·88)

1·000·500·35 2·50


Figure 4: Meta-analysis of SGLT2i trials on the composite of renal worsening, end-stage renal disease, or renal death stratified by the presence of established atherosclerotic cardiovascular diseaseAtherosclerotic cardiovascular disease: Q statistic=0·19, p=0·91, I²=0%; multiple risk factors: Q statistic=0·52, p=0·47, I²=0%. The p value for subgroup differences was 0·71. Tests for subgroup differences were based on F tests in a random effect meta-regression estimated using restricted maximum likelihood and Hartung Knapp adjustment. HR=hazard ratio. SGLT2i=sodium-glucose cotransporter-2 inhibitors.


Weight(%)

HREventsPatients

TreatmentPatients with atherosclerotic cardiovascular diseaseEMPA-REG OUTCOMECANVAS ProgramDECLARE-TIMI 58Fixed effects model for atherosclerotic cardiovascular disease (p<0·0001)

Patients with multiple risk factorsCANVAS ProgramDECLARE-TIMI 58Fixed effects model for multiple risk factors (p<0·0001)

4645 3756 3474

2039 5108

2323 2900 3500

1447 5078

Placebo

11·5 10·5 8·6

6·6 5·9

31·0 35·6 33·4

29·5 70·5

6·3 6·4 4·7

4·1 3·0

152 179 183

70182

0·54 (0·40–0·75)0·59 (0·44–0·79)0·55 (0·41–0·75)0·56 (0·47–0·67)

0·63 (0·39–1·02)0·51 (0·37–0·69)0·54 (0·42–0·71)

Treatment (n) Placebo (n)

1·000·500·35 2·50


KDIGO

ADA Consensus Recommendations


compelling in this patient group. Thus, we recommend thatproviders consider a history of CVD very early in the processof treatment selection. Other factors affect the choice ofglucose-lowering medications, particularly in the setting ofpatient-centred care. In addition to CVD, we recommend earlyconsideration of weight, hypoglycaemic risk, treatment costand other patient-related factors that may influence treatmentselection (Figs 2, 3, 4, 5, 6).

Implications of new evidence from cardiovascularoutcomes trials

The major change from prior consensus reports is based onnew evidence that specific sodium–glucose cotransporter-2(SGLT2) inhibitors or glucagon-like peptide-1 (GLP-1) recep-tor agonists improve cardiovascular outcomes, as well as sec-ondary outcomes such as HF and progression of renal disease,in patients with established CVD or CKD. Therefore, an im-portant early step in this new approach (Fig. 3) is to considerthe presence or absence of ASCVD, HF and CKD, conditionsin aggregate affecting 15–25% of the population with type 2diabetes. While the new evidence supporting the use of par-ticular medications in patients who also have established CVDor are at high risk of CVD is derived from large cardiovascularoutcomes trials (CVOTs) demonstrating substantial benefitsover 2–5 years, it is important to remember that each trialconstitutes a single experiment.Within each drug class, resultshave been heterogeneous. It is not clear whether there are truedrug class effects with different findings for individual medi-cations due to differences in trial design and conduct, orwhether there are real differences between medications withina drug class due to properties of the individual compounds.Where the current evidence is strongest for a specific medica-tion within a class, it is noted. The ADA’s ‘Standards of med-ical care in diabetes’ will align with this document and will beupdated to reflect new evidence as it emerges from ongoingclinical trials.

ASCVD is defined somewhat differently across trials, but alltrials enrolled individuals with established CVD (e.g. myocar-dial infarction [MI], stroke, any revascularisation procedure)while variably including related conditions compatible withclinically significant atherosclerosis (e.g. transient ischaemicattack, hospitalised unstable angina, amputation, congestive

heart failure New York Heart Association [NYHA] class II–III, >50% stenosis of any artery, symptomatic or asymptomaticcoronary artery disease documented by imaging, CKD withestimated GFR [eGFR] <60 ml min-1 [1.73 m]-2). Most trialsalso included a ‘risk factor only’ group with entry criteria basedon age and usually the presence of two or more cardiac riskfactors [46]. Trials were designed to evaluate cardiovascularsafety (i.e. statistical non-inferiority compared with placebo),but several showed ASCVD outcome benefit (i.e. statisticalsuperiority compared with placebo), including, in some cases,mortality.

Among GLP-1 receptor agonists, liraglutide, studied in theLiraglutide Effect and Action in Diabetes: Evaluation ofCardiovascular Outcomes Results (LEADER) trial (n= 9340)demonstrated an ARR of 1.9% with an HR of 0.87 (95% CI0.78, 0.97; p= 0.01 for superiority) for the primary compositeoutcome of cardiovascular death, non-fatal MI and non-fatalstroke (major adverse cardiac events [MACE]) compared withplacebo over 3.8 years. Each component of the composite con-tributed to the benefit, and the HR for cardiovascular death was0.78 (95%CI 0.66, 0.93;p= 0.007; ARR 1.7%). The LEADERtrial also demonstrated an HR of 0.85 (95% CI, 0.74 to 0.97;p= 0.02; ARR 1.4%) for all-cause mortality [47]. In the Trial toEvaluate Cardiovascular and Other Long-term Outcomes withSemaglutide in Subjects with Type 2 Diabetes (SUSTAIN 6)(n= 3297), semaglutide compared with placebo demonstratedan ARR of 2.3% with HR 0.74 for MACE (95% CI 0.58, 0.95;p= 0.02 for superiority) over 2.1 years, but the reduction inevents appeared to be driven by the rate of stroke, rather thanCVD death [48]. The Exenatide Study of Cardiovascular EventLowering (EXSCEL) compared exenatide extended-releasewith placebo over 3.2 years in 14,752 participants with type 2diabetes. While the medication was safe (non-inferior), the HRfor MACE in the entire trial was 0.91 (95% CI 0.83, 1.0; p=0.06) not reaching the threshold for demonstrated superiority vsplacebo; ARR was 0.8% [49]. All-cause death was lower in theexenatide arm (ARR 1%, HR 0.86 [95% CI 0.77, 0.97]), but itwas not considered to be statistically significant in the hierar-chical testing procedure applied. Lixisenatide, a short-actingGLP-1 receptor agonist, did not demonstrate CVD benefit orharm in a trial of patients recruited within 180 days of an acutecoronary syndrome admission [50]. Taken together, it appearsthat among patients with established CVD, some GLP-1 recep-tor agonists may provide cardiovascular benefit, with the evi-dence of benefit strongest for liraglutide, favourable forsemaglutide, and less certain for exenatide. There is no evi-dence of cardiovascular benefit with lixisenatide. Adverse ef-fects for the class are discussed in the section ‘The full range oftherapeutic options: lifestyle management, medication and obe-sity management’.

Among the SGLT2 inhibitors, empagliflozin compared withplacebo was studied in the Empagliflozin, CardiovascularOutcome Event Trial in Type 2 Diabetes Mellitus Patients

Consensus recommendation Among patients with type 2 diabetes who have established ASCVD, SGLT2 inhibitors or GLP-1 receptor agonists with proven cardiovascular benefit are recommended as part of glycaemic management (Figs 2 and 3).

Diabetologia (2018) 61:2461–2498 2467

fatal MI, non-fatal stroke and cardiovascular death. The ARRwas 2.2% and the HR was 0.62 (95% CI 0.49, 0.77; p< 0.001)for cardiovascular death [51]. The ARR was 2.6% and the HRwas 0.68 (95% CI, 0.57, 0.82; p< 0.001) for death from anycause. Canagliflozin compared with placebo was studied in theCanagliflozin Cardiovascular Assessment Study (CANVAS)Program (comprised of two similar trials, CANVAS andCANVAS-Renal;n= 10,142) in participants with type 2 diabe-tes, 66% of whom had a history of CVD. Participants werefollowed for a median of 3.6 years. In the combined analysisof the two trials, the primary composite endpoint of MI, strokeor cardiovascular death was reduced with canagliflozin (26.9 vs31.5 participants per patient-year with placebo; HR 0.86, 95%CI 0.75, 0.97; p= 0.02) for superiority in the pooled analysis,with consistent findings in the component studies. Thoughthere was a trend towards benefit for cardiovascular death, thedifference from placebo was not statistically significant in theCANVAS Program [52]. For the SGLT2 inhibitors studied todate, it appears that among patientswith established CVD, thereis likely cardiovascular benefit, with the evidence of benefitmodestly stronger for empagliflozin than canagliflozin.Adverse effects for the class are discussed in the section ‘Thefull range of therapeutic options: lifestyle management, medi-cation and obesity management’.

While the evidence of an ASCVD outcomes benefit forGLP-1 receptor agonists and SGLT2 inhibitors has been dem-onstrated for people with established ASCVD, the evidence ofbenefit beyond glucose lowering has not been demonstrated inthose without ASCVD. Indeed, in subgroup analyses of thesetrials, lower risk individuals have not been observed to have anASCVD benefit. While this may be due to the short time frameof the studies and the low event rate in those without ASCVD,the finding is consistent across the reported trials. Overall,CVOTs of dipeptidyl peptidase-4 (DPP-4) inhibitors have dem-onstrated safety, i.e. non-inferiority relative to placebo, for theprimary MACE endpoint, but not cardiovascular benefit.

The available evidence for cardiovascular event reductionin patients with type 2 diabetes and clinical CVD is derivedfrom trials in which the participants were not meetingglycaemic targets (HbA1c ≥53 mmol/mol [≥7%] at baseline).Furthermore, most (~70% across trials) participants weretreated with metformin at baseline. Thus, we recommend thatpatients with clinical CVD not meeting individualisedglycaemic targets while treated with metformin (or in whommetformin is contraindicated or not tolerated) should have anSGLT2 inhibitor or GLP-1 receptor agonist with proven ben-efit for cardiovascular risk reduction added to their treatmentprogramme. There are no clinical trial data that support pre-scribing an SGLT2 inhibitor or GLP-1 receptor agonist withthe intent of reducing cardiovascular risk in patients with anHbA1c <53 mmol/mol (<7%). Limited data suggest that thereis no heterogeneity in the cardiovascular benefits of SGLT2inhibitors or GLP-1 receptor agonists as a function of

background glucose-lowering therapy. Thus, backgroundglucose-lowering therapy in patients with clinical CVD argu-ably is not pertinent in clinical decision making. However,dose adjustment or discontinuation of background medica-tions may be required to avoid hypoglycaemia when addinga new agent to a regimen containing insulin, sulfonylurea orglinide therapy, particularly in patients at or near glycaemicgoals. Full efforts to achieve glycaemic and blood pressuretargets and to adhere to lipid, antiplatelet, antithrombotic andtobacco cessation guidelines [7] should continue after anSGLT2 inhibitor or GLP-1 receptor agonist is added, as suchefforts were integral to all studies that have demonstrated car-diovascular benefit of these agents.

Patients with type 2 diabetes are at increased risk of HF[53]. In the EMPA-REG OUTCOME and CANVAS CVOTstudies testing SGLT2 inhibitors, which enrolled participantswith ASCVD, >85% of participants did not have symptomaticHF at baseline. Yet, in both trials there was a clinically andstatistically significant reduction in hospitalisation for HF forthe SGLT2 inhibitor as compared with placebo. In the EMPA-REG OUTCOME study with empagliflozin [54], the ARRwas 1.4%, and the HR 0.65 (95% CI 0.50, 0.85) and in theCANVAS Program with canagliflozin the HR was 0.67 (95%CI 0.52, 0.87), with a rate of hospitalised HF of 5.5 vs 8.7events per 1000 patient-years [55]. Because HF was neitherwell characterised at baseline nor as carefully adjudicated as itwould have been in a trial specifically designed to evaluate HFoutcomes, and because HF was a secondary endpoint in thetrials, further ongoing studies are required to conclusively ad-dress the issue. That said, the significant reduction inhospitalisation for HF demonstrated in the two study popula-tions and the consistency across two independent trialprogrammes suggest to us that treatment with SGLT2 inhibi-tors in the setting of clinical HF may provide substantial ben-efit and should be specifically considered in people with type2 diabetes and ASCVD and HF.

In the GLP-1 receptor agonist studies LEADER,SUSTAIN 6 and EXSCEL, there was no significant effecton hospitalization for HF with HR 0.86 (95% CI 0.71, 1.06),1.11 (95% CI 0.77, 1.61) and 0.94 (95% CI 0.78, 1.13), re-spectively [47–49]. Two short-term studies of liraglutide inpatients with reduced ejection fraction suggested a lack ofbenefit in this setting [56, 57].

Among the recent cardiovascular safety outcomes trialstesting DPP-4 inhibitors, the Saxagliptin Assessment ofVascular Outcomes Recorded in Patients with Diabetes

Consensus recommendation Among patients with ASCVD in whom HF coexists or is of special concern, SGLT2 inhibitors are recommended (Figs 2 and 3).

2472 Diabetologia (2018) 61:2461–2498

KDIGO

SGLT2i Trials: The Surprisingly Impressive Suggestion of Renoprotection

n engl j med 375;4 nejm.org July 28, 2016 329

Empagliflozin and Kidney Disease in Type 2 Diabetes

disease in patients with type 2 diabetes. How-ever, the renal effects of empagliflozin cannot necessarily be generalized to patients with type 2 diabetes at lower cardiovascular risk. General-ization of the findings to black patients also has limitations because of the small sample size that we studied. Further clinical research is war-ranted to validate our findings in broader popu-lations at risk for adverse renal outcomes.

Empagliflozin was associated with lower rates of hyperglycemia and lower values for weight and blood pressure than was placebo.23 A trial comparing an intensive multifactorial interven-tion with conventional therapy in patients with type 2 diabetes and microalbuminuria did not show significant differences in the decline in renal function after 7.8 years.27 Thus, in our trial, the magnitude of the observed effect on these risk factors over a median follow-up period of 3.1 years is unlikely to fully account for the ob-served difference in renal function among pa-tients receiving empagliflozin.

The mechanisms behind the renal effects of empaglif lozin are probably multifactorial, but direct renovascular effects may play an impor-tant role.21,28,29 Empagliflozin reduces proximal tubular sodium reabsorption, thereby increasing distal sodium delivery to the macula densa, which has been shown to activate tubuloglo-merular feedback, leading to afferent vasomodu-lation and a decrease in hyperfiltration.30 In pa-tients with type 1 diabetes and hyperfiltration, empagliflozin reduces the intraglomerular pres-sure.20 Despite the low-pressure environment of renal glomeruli, a reduction in glomerular hy-pertension of approximately 6 to 8 mm Hg was observed with empagliflozin.20 Other effects, such as those on arterial stiffness,28,29 vascular resistance,28 serum uric acid levels,23 and the systemic and renal neurohormonal systems,29,31,32 may also contribute to the improvements in the progression of renal disease observed with em-pagliflozin. Further research is needed to ex-plore whether empagliflozin-associated changes in blood volume or renal perfusion may alter serum creatinine turnover and renal-function assessments.

Previous trials have focused on the role of RAAS blockade for improving renal outcomes in type 2 diabetes.33-35 Such blockade causes vasodi-lation of the efferent arteriolar system of the glomerulus and a reduction in intraglomerular

pressure.36,37 This vasomodulatory mechanism is known to cause a short-term decrease in the eGFR,38 which is accompanied by a smaller de-cline in renal function during continued treat-ment.39 In our trial, empaglif lozin showed a similar pattern of change in renal function (i.e., a short-term decrease followed by stabilization

Figure 1. Kaplan–Meier Analysis of Two Key Renal Outcomes.

Shown are estimates of the probability of a first occurrence of a prespecified renal composite outcome of incident or worsening nephropathy (Panel A) and of a post hoc renal composite outcome (a doubling of the serum creat-inine level, the initiation of renal-replacement therapy, or death from renal disease) (Panel B) among patients who received at least one dose of either empagliflozin or placebo. The inset in Panel B shows the data on an expand-ed y axis. Hazard ratios are based on Cox regression analyses. Because of the declining numbers of patients at risk, Kaplan–Meier curves have been truncated at 48 months.

Cum

ulat

ive

Prob

abili

tyof

Eve

nt (%

)

30

40

20

10

00 6 12 18 24 30 36 42 48

Month

B Post Hoc Renal Composite Outcome

A Incident or Worsening Nephropathy

Hazard ratio, 0.61 (95% CI, 0.53–0.70)P<0.001

Placebo

Empagliflozin

No. at RiskEmpagliflozinPlacebo

41242061

39941946

38481836

36691703

31711433

22791016

1887833

1219521

290106

Cum

ulat

ive

Prob

abili

tyof

Eve

nt (%

)

80

70

100

90

60

20

30

40

50

0

10

0 6 12 18 24 30 36 42 48

Month

876

2345

01

0 6 12 18 24 30 36 42 48

Hazard ratio, 0.54 (95% CI, 0.40–0.75)P<0.001

Placebo

Empagliflozin

No. at RiskEmpagliflozinPlacebo

46452323

45002229

43772146

42412047

37291771

27151289

22801079

1496680

360144

50

60

70

80

90

100

The New England Journal of Medicine Downloaded from nejm.org at NATIONAL UNIV OF SINGAPORE-CENTRAL LIBRARY on March 8, 2018. For personal use only. No other uses without permission.

Copyright © 2016 Massachusetts Medical Society. All rights reserved.

n engl j med 375;4 nejm.org July 28, 2016 331

Empagliflozin and Kidney Disease in Type 2 Diabetes

Figure 3. Renal Function over Time.

Shown are the adjusted means for the estimated glomerular filtration rate (eGFR) over a period of 192 weeks (Panel A) and at the last measurement during treatment and at follow-up (Panel B) among patients who received empagliflozin (at a dose of 10 mg or 25 mg) or placebo. Baseline values are means, and the I bars indicate standard errors. The eGFR was calculated according to the creatinine formula developed by the Chronic Kidney Disease Epidemiology Collaboration. Among patients in the empagliflozin group, the adjusted mean difference from placebo in the change from baseline at follow-up (Panel B) was 4.7 ml per minute per 1.73 m2 in both the 10-mg and 25-mg groups (P<0.001 for both compari-sons). Panel A is based on prespecified mixed-model, repeated-measures analysis in patients who received at least one dose of a study drug and had a baseline and postbaseline measurement. Panel B is based on a prespecified analysis of a covariance model (with the baseline eGFR and glycated hemoglobin level as linear covariates and baseline body-mass index, region, and study group as fixed effects) in patients who underwent measurements at all three time points.

Placebo

Empagliflozin, 10 mg

Empagliflozin, 25 mg

Placebo (N=1555) Empagliflozin, 10 mg (N=1642) Empagliflozin, 25 mg (N=1686)

Last measurement during treatment

Follow-up Baseline

Median, 3.0 years Median, 34 days

Adj

uste

d M

ean

eGFR

(ml/

min

/1.7

3 m

2 )

78

76

74

72

70

68

66

Adj

uste

d M

ean

eGFR

(ml/

min

/1.7

3 m

2 )

78

76

74

72

70

68

66

Baseline 12 284 52 66 192

Week

A Change in eGFR over 192 Wk

No. at RiskPlaceboEmpagliflozin, 10 mgEmpagliflozin, 25 mgNo. in Follow-up

AnalysisTotal

232323222322

229522902288

7020 7020

226722642269

220522352216

6996 6931

80 94 108 122 136 150 164 178

B Change in eGFR from Baseline to Last Measurement during Treatment and Follow-up

212121622156

6864

206421142111

6765

192720122006

6696

198120642067

6651

176318391871

6068

147915401563

126213141340

5114 4443

112311801207

3961

97710241063

3488

731785838

2707

448513524

1703



Wanner C et al. N Engl J Med 2016; 375:323-334.

KDIGO

The Next Best Thing After RAAS Blockade for Diabetic Kidney Disease

CREDENCETrial-Canagliflozin

PrimaryOutcome:1.  Compositeofend-stagekidneydisease(dialysis,transplantation,orasustained

estimatedGFR<15ml/min/1.73m2),doublingofserumcreatinine,ordeathfromrenalorcardiovascularcauses.

•  CREDENCE began before any CV outcomes trials had reported

•  Renal effects were not the primary focus of the CV outcomes trials

2014

2015 2016 2017 2018 2019 CREDENCE enrollment

CREDENCE ended

DECLARE EMPA-REG OUTCOME

CANVAS Program

KDIGO

Perkovic V et al. N Engl J Med 2019; doi:10.1056/NEJMoa1811744.

CREDENCE: SGLT2i and Renoprotection

Participants continued treatment if eGFR was <30 mL/min/1.73 m2 until chronic dialysis was initiated or kidney transplant occurred.

Key inclusion criteria •  ≥30 years of age •  T2DM and HbA1c 6.5% to 12.0% •  eGFR 30 to 90 mL/min/1.73 m2 •  UACR 300 to 5000 mg/g •  Stable max tolerated labelled dose

of ACEi or ARB for ≥4 weeks

Key exclusion criteria •  Other kidney diseases, dialysis, or kidney transplant •  Dual ACEi and ARB; direct renin inhibitor; MRA •  Serum K+ >5.5 mmol/L •  CV events within 12 weeks of screening •  NYHA class IV heart failure •  Diabetic ketoacidosis or T1DM

2-week placebo run-in Placebo

Canagliflozin 100 mg

R Double-blind

randomization (1:1)

Follow-up at Weeks 3, 13, and 26 (F2F) then every 13 weeks (alternating phone/F2F)

STUDYDESIGN

KDIGO



4401 randomized

15 (0.7%) did not complete 5 (0.2%) withdrew consent

25 (1.1%) did not complete 11 (0.5%) withdrew consent

2199 placebo 2202 canagliflozin

2197 (99.9%) vital status known

2174 (98.9%) completed study

2198 (99.8%) vital status known

2187 (99.3%) completed study

12,900 screened

8499 excluded

KDIGO



-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0 6 12 18 24 30 36 42

Months since randomization

Baseline (%) 8.3 8.3

Canagliflozin Placebo

Mean difference over study –0.25%

(95% CI: –0.31, –0.20) LS m

ean

ch

an

ge (

±S

E)

in

Hb

A1

c (%

)

-5

-4

-3

-2

-1

0

1

2

0 6 12 18 24 30 36 42

LS m

ean

ch

an

ge (

±S

E)

in

sys

tolic

BP

(m

mH

g)


Baseline (mmHg) 139.8 140.2


Mean difference over study –3.30 mmHg

(95% CI: –3.87, –2.73)

0

200

400

600

800

1000

1200

0 6 12 18 24 30 36 42

Geo

metr

ic m

ean

(9

5%

CI)

U

AC

R (

mg

/g

)


Median baseline (mg/g) 914 918


Mean % difference over study –32%

(95% CI: –36, –28)

HbA1c BloodPressure

BodyWeight Albuminuria

KDIGO



0

5

10

15

20

25

0 26 52 78 104 130 156 182

Part

icip

an

ts w

ith

an

eve

nt

(%)


Hazard ratio, 0.70 (95% CI, 0.59–0.82) P = 0.00001

6 12 18 24 30 36 42

340 participants

245 participants

Placebo

Canagliflozin

Primary Outcome: ESKD, Doubling of Serum Creatinine, or Renal or CV Death

KDIGO



ESKD, Doubling of Serum Creatinine, or Renal Death

0

5

10

15

20

25

0 26 52 78 104 130 156 182


Hazard ratio, 0.66 (95% CI, 0.53–0.81) P <0.001

224 participants

153 participants

6 12 18 24 30 36 42

P

art

icip

an

ts w

ith

an

eve

nt

(%)

Placebo

Canagliflozin

KDIGO

-20 -18 -16 -14 -12 -10

-8 -6 -4 -2 0

0 26 52 78 104 130 156 182


56.4 56.0 Canagliflozin Placebo

Chronic eGFR slope Difference: 2.74/year (95% CI, 2.37–3.11)

–4.59/year

6 12 18 24 30 36 42

LS

mean

ch

an

ge (±S

E)

in

eG

FR (

mL/

min

/1

.73

m2)

Baseline

–3.72

Acute eGFR slope (3 weeks) Difference: –3.17 (95% CI, –3.87, –2.47)

–0.55

–1.85/year



No.ofParticipantsPlacebo 2178 2084 1985 1882 1720 1536 1006 583 210

Canagliflozin 2179 2074 2005 1919 1782 1648 1116 652 241

Effects on eGFR

KDIGO

Low Moderately increased High Very high

<30

30-44

45-59

60-90

≥90

GFR

cat

egor

ies

(mL/

min

/1.7

3 m

2 )

Albuminuria categories (mg/g) A1: <30 A2: 30-300 A3: >300

DC E

DECLARE

CANVASProgram

EMPA-REGOUTCOME

CREDENCE

MedianUACR(mg/g)

13

12

18

927

MeaneGFR(mL/min/1.73m2)

85

76

74

56

Sustained RRT Events

DECLARE Not reported CANVAS Program 18 EMPA-REG OUTCOME 11 CREDENCE 176

D

C

E

CREDENCE: Where does it all fit in?

KDIGO


CREDENCE: Putting it in Perspective

Hazardratio(95%CI)

InteractionPvalue

ScreeningeGFR 0.11

30to<45mL/min/1.73m20.75(0.59–0.95)

45to<60mL/min/1.73m20.52(0.38–0.72)

60to<90mL/min/1.73m20.82(0.60–1.12)

Favors Canagliflozin Favors Placebo

0.25 0.5 1.0 2.0 4.0

16

NNT in patients with eGFR 30 to <45 mL/min/1.73 m2

KDIGO

ShouldSGLT2InhibitorsbeinitiatedinallpatientswithType2diabetesmellitusandalbuminuria,witheGFR>30ml/min/1.73m2?KDIGO

The Middle Child

GLP-1 Receptor Agonists

KDIGO

Insulinotropic Glucagonostatic

Muskiet MHA et al. Nat Rev Nephrol 2017; 13:605-628.

Actions of GLP-1

KDIGO

Muskiet MHA et al. Nat Rev Nephrol 2017; 13:605-628.

GLP-1RA and Renal Hemodynamics

KDIGO

Marso SP et al. N Engl J Med 2016; 375:311-322.

LEADER Trial: Liraglutide vs Placebo

n engl j med 375;4 nejm.org July 28, 2016316

T h e n e w e ngl a nd j o u r na l o f m e dic i n e

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54

Months since Randomization

C Nonfatal Myocardial Infarction

A Primary Outcome

Hazard ratio, 0.87 (95% CI, 0.78– 0.97)P<0.001 for noninferiorityP=0.01 for superiority

No. at RiskLiraglutidePlacebo

46684672

44964473

42804237

40724010

45934588

44004352

41724123

39823914

15621543

424407

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


B Death from Cardiovascular Causes

Hazard ratio, 0.78 (95% CI, 0.66– 0.93)P=0.007


46684672

45994601

45054479

43824338

46414648

45584546

44454407

43224267

17231709

484465

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


E Death from Any Cause

Hazard ratio, 0.88 (95% CI, 0.75– 1.03)P=0.11


46684672

45314513

43594301

41814103

46094613

44544407

42634202

41024020

16191594

440424

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


D Nonfatal Stroke

Hazard ratio, 0.89 (95% CI, 0.72– 1.11)P=0.30


46684672

45644558

44264405

42694228

46244622

45044484

43514314

41944141

16621648

465445

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


Hazard ratio, 0.85 (95% CI, 0.74– 0.97)P=0.02


46684672

45994601

45054479

43824338

46414648

45584546

44454407

43224268

17231709

484465

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


F Hospitalization for Heart Failure

Hazard ratio, 0.87 (95% CI, 0.73– 1.05)P=0.14


46684672

45504540

44144372

42584187

46124612

44834464

43374288

41854107

16621647

467442

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

The New England Journal of Medicine Downloaded from nejm.org at NATIONAL UNIV OF SINGAPORE-CENTRAL LIBRARY on May 4, 2019. For personal use only. No other uses without permission.


n engl j med 375;4 nejm.org July 28, 2016316


Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


C Nonfatal Myocardial Infarction

A Primary Outcome

Hazard ratio, 0.87 (95% CI, 0.78– 0.97)P<0.001 for noninferiorityP=0.01 for superiority


46684672

44964473

42804237

40724010

45934588

44004352

41724123

39823914

15621543

424407

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


B Death from Cardiovascular Causes

Hazard ratio, 0.78 (95% CI, 0.66– 0.93)P=0.007


46684672

45994601

45054479

43824338

46414648

45584546

44454407

43224267

17231709

484465

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


E Death from Any Cause

Hazard ratio, 0.88 (95% CI, 0.75– 1.03)P=0.11


46684672

45314513

43594301

41814103

46094613

44544407

42634202

41024020

16191594

440424

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


D Nonfatal Stroke

Hazard ratio, 0.89 (95% CI, 0.72– 1.11)P=0.30


46684672

45644558

44264405

42694228

46244622

45044484

43514314

41944141

16621648

465445

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


Hazard ratio, 0.85 (95% CI, 0.74– 0.97)P=0.02


46684672

45994601

45054479

43824338

46414648

45584546

44454407

43224268

17231709

484465

Patie

nts

with

an

Even

t (%

)

100

80

90

70

60

40

30

10

50

20

00 6 12 18 24 30 36 42 48 54

15

10

20

5

00 6 12 18 24 30 36 42 48 54


F Hospitalization for Heart Failure

Hazard ratio, 0.87 (95% CI, 0.73– 1.05)P=0.14


46684672

45504540

44144372

42584187

46124612

44834464

43374288

41854107

16621647

467442

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

Liraglutide

Placebo

The New England Journal of Medicine Downloaded from nejm.org at NATIONAL UNIV OF SINGAPORE-CENTRAL LIBRARY on May 4, 2019. For personal use only. No other uses without permission.


Randomized 9,340 patients with high cardiovascular risks Median Follow-up of 3.84 years Primary composite outcome of death from cardiovascular causes, nonfatal myocardial infarction, or nonfatal stroke.

KDIGO

n engl j med 377;9 nejm.org August 31, 2017 843

Lir aglutide and Renal Outcomes in Type 2 Diabetes

composite renal outcome occurred in 146 of 1116 patients (13.1%) in the liraglutide group and in 156 of 1042 (15.0%) in the placebo group (hazard ratio, 0.84; 95% CI, 0.67 to 1.05; P = 0.13). In patients with both an estimated GFR of less than 60 ml per minute per 1.73 m2 and microal-buminuria or macroalbuminuria, the composite renal outcome occurred in 131 of 583 patients (22.5%) in the liraglutide group and in 141 of 547 (25.8%) in the placebo group (hazard ratio, 0.81; 95% CI, 0.64 to 1.03; P = 0.09).

Renal Function over TimeThe estimated GFR declined continuously (Fig. 3A), but the decline was slightly slower in the liraglu-tide group than in the placebo group (estimated trial-group ratio at 36 months, 1.02; 95% CI, 1.00 to 1.03; P = 0.01, corresponding to a 2% less decrease with liraglutide). The decrease in the estimated GFR at 36 months was 7.44 ml per minute per 1.73 m2 in the liraglutide group, as compared with 7.82 ml per minute per 1.73 m2 in the placebo group. The urinary albumin-to-

Figure 1. Composite Renal Outcome and Components of the Composite Outcome.

The primary composite renal outcome in the time-to-event analysis was a composite (Panel A) of the first occurrence of persistent macro-albuminuria (Panel B), persistent doubling of the serum creatinine level and an estimated glomerular filtration rate of 45 ml or less per minute per 1.73 m2 of body-surface area (referred to as persistent doubling of the serum creatinine level; Panel C), the need for continu-ous renal-replacement therapy (for end-stage renal disease; Panel D), or death due to renal disease (data not shown). The component of death due to renal disease occurred in 13 patients (8 patients in the liraglutide group and 5 in the placebo group). Cumulative incidences were estimated with the use of the Kaplan–Meier method, and the hazard ratios were calculated with the use of the Cox proportional-hazard regression model. The insets show the same data on an enlarged y axis. The data analyses are truncated at 54 months because less than 10% of the participants had an observation time beyond 54 months. All the events were adjudicated. One patient with macro-albuminuria at baseline had an event of new-onset persistent macroalbuminuria that was confirmed by adjudication after the patient had regression to microalbuminuria earlier in the trial.

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


A Composite Renal Outcome

No. at RiskPlaceboLiraglutide

46724668

45404561

43164400

40944210

16131632

6 18 30 42

46434635

44284492

41964304

39904114

433454

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


B New Onset of Persistent Macroalbuminuria


46724668

45514570

43594437

41624268

16421662

6 18 30 42

46464638

44554508

42524353

40734182

442461

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


C Persistent Doubling of Serum Creatinine Level


46724668

45964591

44474476

42824332

16821692

6 18 30 42

46474639

45294544

43674403

41964264

456475

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


D Continuous Renal-Replacement Therapy


46724668

45904596

44544484

42994349

16991710

6 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

46454640

45274547

43704416

42274282

461483

Hazard ratio, 0.74 (95% CI, 0.60–0.91)P=0.004

Placebo

Liraglutide

Hazard ratio, 0.78 (95% CI, 0.67–0.92)P=0.003

Placebo

Liraglutide

Hazard ratio, 0.87 (95% CI, 0.61–1.24)P=0.44

PlaceboLiraglutide

Hazard ratio, 0.89 (95% CI, 0.67–1.19)P=0.43

Placebo

Liraglutide





composite renal outcome occurred in 146 of 1116 patients (13.1%) in the liraglutide group and in 156 of 1042 (15.0%) in the placebo group (hazard ratio, 0.84; 95% CI, 0.67 to 1.05; P = 0.13). In patients with both an estimated GFR of less than 60 ml per minute per 1.73 m2 and microal-buminuria or macroalbuminuria, the composite renal outcome occurred in 131 of 583 patients (22.5%) in the liraglutide group and in 141 of 547 (25.8%) in the placebo group (hazard ratio, 0.81; 95% CI, 0.64 to 1.03; P = 0.09).

Renal Function over TimeThe estimated GFR declined continuously (Fig. 3A), but the decline was slightly slower in the liraglu-tide group than in the placebo group (estimated trial-group ratio at 36 months, 1.02; 95% CI, 1.00 to 1.03; P = 0.01, corresponding to a 2% less decrease with liraglutide). The decrease in the estimated GFR at 36 months was 7.44 ml per minute per 1.73 m2 in the liraglutide group, as compared with 7.82 ml per minute per 1.73 m2 in the placebo group. The urinary albumin-to-

Figure 1. Composite Renal Outcome and Components of the Composite Outcome.

The primary composite renal outcome in the time-to-event analysis was a composite (Panel A) of the first occurrence of persistent macro-albuminuria (Panel B), persistent doubling of the serum creatinine level and an estimated glomerular filtration rate of 45 ml or less per minute per 1.73 m2 of body-surface area (referred to as persistent doubling of the serum creatinine level; Panel C), the need for continu-ous renal-replacement therapy (for end-stage renal disease; Panel D), or death due to renal disease (data not shown). The component of death due to renal disease occurred in 13 patients (8 patients in the liraglutide group and 5 in the placebo group). Cumulative incidences were estimated with the use of the Kaplan–Meier method, and the hazard ratios were calculated with the use of the Cox proportional-hazard regression model. The insets show the same data on an enlarged y axis. The data analyses are truncated at 54 months because less than 10% of the participants had an observation time beyond 54 months. All the events were adjudicated. One patient with macro-albuminuria at baseline had an event of new-onset persistent macroalbuminuria that was confirmed by adjudication after the patient had regression to microalbuminuria earlier in the trial.

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


A Composite Renal Outcome


46724668

45404561

43164400

40944210

16131632

6 18 30 42

46434635

44284492

41964304

39904114

433454

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


B New Onset of Persistent Macroalbuminuria


46724668

45514570

43594437

41624268

16421662

6 18 30 42

46464638

44554508

42524353

40734182

442461

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


C Persistent Doubling of Serum Creatinine Level


46724668

45964591

44474476

42824332

16821692

6 18 30 42

46474639

45294544

43674403

41964264

456475

Pat

ien

ts w

ith

an

Eve

nt

(%)

100

80

60

40

20

00 12 24 36 48 54


D Continuous Renal-Replacement Therapy


46724668

45904596

44544484

42994349

16991710

6 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

10

8

6

4

2

00 12 24 36 48 546 18 30 42

46454640

45274547

43704416

42274282

461483

Hazard ratio, 0.74 (95% CI, 0.60–0.91)P=0.004

Placebo

Liraglutide

Hazard ratio, 0.78 (95% CI, 0.67–0.92)P=0.003

Placebo

Liraglutide

Hazard ratio, 0.87 (95% CI, 0.61–1.24)P=0.44

PlaceboLiraglutide

Hazard ratio, 0.89 (95% CI, 0.67–1.19)P=0.43

Placebo

Liraglutide



Pre-specified secondary renal outcomes which was a composite of new-onset persistent albuminuria, doubling of serum creatinine, ESRD or death to renal causes

Mann JFE et al. N Engl J Med 2017; 377:839-848.

LEADER Trial: Liraglutide vs Placebo

KDIGO


LEADER Trial: Liraglutide



in the liraglutide group and the placebo group (15.1 events and 16.5 events per 1000 patient-years), including the rate of acute kidney injury (7.1 events and 6.2 events per 1000 patient-years, respectively), according to adverse-event report-ing. The similarity in the rates of adverse events also applies to subgroups defined according to the estimated GFR at baseline. Details are pro-vided in Table S12A and S12B in the Supplemen-tary Appendix.

Discussion

Among patients who were receiving usual care, liraglutide resulted in significantly lower rates of renal outcomes than placebo among patients with type 2 diabetes who were at high cardiovas-cular risk. This result was driven mainly by a lower incidence of macroalbuminuria in the lira-glutide group than in the placebo group. There were nonsignificantly lower risks of the doubling of the serum creatinine level and of end-stage renal disease with liraglutide than with placebo during up to 5 years of follow-up.

New-onset persistent macroalbuminuria is an effect that is typically associated with subsequent progressive reductions in the GFR in patients with type 2 diabetes.17 For example, in ONTARGET (Ongoing Telmisartan Alone and in Combina-tion with Ramipril Global Endpoint Trial) and TRANSCEND (Telmisartan Randomized Assess-ment Study in ACE Intolerant Subjects with Cardiovascular Disease), new-onset macroalbu-minuria was associated with a risk of end-stage renal disease or doubling of the serum creati-nine level that was 3 to 5 times as high as the risk among patients in whom new-onset macro-albuminuria did not develop.18 Similar renal out-comes associated with new-onset macroalbu-minuria were reported in the ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evalua-tion) trial19 and supported by meta-regression analyses.17 Macroalbuminuria is also a risk fac-tor for cardiovascular events.17,20

In this present trial, the risks of doubling of the serum creatinine level and end-stage renal disease did not differ significantly between the liraglutide group and the placebo group, possi-bly owing to the moderate decline in the esti-mated GFR observed in this cohort and the few patients who had advanced kidney disease at

randomization. The initial changes in the esti-mated GFR are difficult to interpret,21 but there may be a potential slowing in the reduction in the estimated GFR with liraglutide in patients with a low baseline estimated GFR.

The mechanism behind the effect of liraglu-tide on the renal outcomes is unclear. Statistical adjustment for the glycated hemoglobin level at

Figure 3. Changes in the Estimated GFR and Urinary Albumin-to-Creatinine Ratio.

Panel A shows the estimated GFR, and Panel B the urinary albumin-to- creatinine ratio (with albumin measured in milligrams and creatinine mea-sured in grams). Geometric means were estimated for the urinary albumin-to-creatinine ratio with the use of a linear mixed model for log-transformed assessment, with accounting for repeated measures. Trial-group ratios were estimated with the use of a mixed-effect model for repeated measures on log-transformed values. Interaction between visit and, respectively, trial group, sex, geographic region, and use of antidiabetic therapy at baseline were included as fixed effects, and interaction between visit and baseline log-estimated GFR or baseline urinary albumin-to-creatinine ratio and age at baseline were included as covariates. The values for the urinary albumin-to-creatinine ratio that were outside the range of quantification were imputed (see the Supplementary Methods section in the Supplementary Appendix).

Mea

n Es

timat

ed G

FR(m

l/m

in/1

.73

m2 )

85

75

80

70

65

60

00 6 12 24 36 48


B Urinary Albumin-to-Creatinine Ratio

A Estimated GFR

Estimated trial-group ratio at 36 mo, 1.02 (95% CI, 1.00–1.03)P=0.01


46724668

43564349

42374288

39114031

36343806

755812

Liraglutide

PlaceboU

rinar

y A

lbum

in-to

-Cre

atin

ine

Ratio

35

25

30

20

15

10

00 12 24 36 48


Estimated trial-group ratio at 36 mo, 0.83 (95% CI, 0.79–0.88)P<0.001


45594578

41034167

37893934

35093686

730786

Liraglutide

Placebo



KDIGO


LEADER Trial: Liraglutide

n engl j med 377;9 nejm.org August 31, 2017844


creatinine ratio increased less in the liraglutide group, yielding a 17% lower urinary albumin-to-creatinine ratio at 36 months in favor of liraglu-tide (estimated trial-group ratio, 0.83; 95% CI, 0.79 to 0.88; P<0.001) (Fig. 3B). The estimated increase in the urinary albumin-to-creatinine ratio at 36 months was 1.8 mg of albumin per gram of creatinine in the liraglutide group, as compared with 6.3 mg of albumin per gram of creatinine in the placebo group (Fig. 3B); in an analysis that excluded values outside the quanti-fication range, the results were similar (Table S10A and S10B in the Supplementary Appendix). The smaller increase in albuminuria at 36 months in the liraglutide group than in the placebo group was independent of the baseline estimated GFR and baseline albuminuria (Table S10A and S10B and Fig. S1A and S1B in the Supplementary Ap-pendix).

When results were stratified according to the estimated GFR at baseline, the decrease in the es-timated GFR in patients with a baseline estimated GFR of 30 to 59 ml per minute per 1.73 m2 was 2 ml per minute per 1.73 m2 in the liraglutide group, as compared with 4 ml per minute per 1.73 m2 in the placebo group (estimated trial-group ratio in favor of liraglutide, 1.07; 95% CI,

1.04 to 1.10; P <0.001) (Fig. 4, and Fig. S2 in the Supplementary Appendix). Results did not differ significantly between groups in patients with a baseline estimated GFR of 60 ml or more per minute per 1.73 m2 or with an estimated GFR of less than 30 ml per minute per 1.73 m2 (there were few patients in the latter group). When analyzed according to albuminuria at baseline, the decrease in the estimated GFR was smaller in the liraglutide group than in the placebo group in patients with macroalbuminuria (P = 0.01) but did not differ significantly in those with micro-albuminuria (P = 0.24) or normoalbuminuria (P = 0.60) (data not shown).

Other Outcomes and Adverse EventsOutcomes that were not prespecified included the composite of the doubling of the serum cre-atinine level or the use of renal-replacement therapy; there were no significant differences between the randomized groups (Table S11 in the Supplementary Appendix). New-onset micro-albuminuria occurred in fewer patients in the liraglutide group than in the placebo group (2293 patients [49.1%] vs. 2498 [53.5%]; hazard ratio, 0.87; 95% CI, 0.83 to 0.93; P<0.001).

The rates of renal adverse events were similar

Figure 2. Composite Renal Outcome, According to Baseline Renal Risk.

The estimated glomerular filtration rate (GFR) was calculated with the use of the Modification of Diet in Renal Dis-ease formula. Hazard ratios and P values were estimated with the use of a Cox proportional-hazards model with in-teraction between trial group and renal risk at baseline. The composite renal outcome consisted of new-onset per-sistent macroalbuminuria, persistent doubling of the serum creatinine level and an estimated GFR of 45 ml or less per minute per 1.73 m2, the need for continuous renal-replacement therapy (end-stage renal disease), or death due to renal disease.

Placebo BetterLiraglutide Better

All patientsEstimated GFR

<60 ml/min/1.73 m2

≥60 ml/min/1.73 m2

Microalbuminuria or macroalbuminuriaYesNo

Combined estimated GFR and albuminuria statusEstimated GFR <60 ml/min/1.73 m2

and microalbuminuria or macroalbuminuriaEstimated GFR ≥60 ml/min/1.73 m2

or no microalbuminuria or macroalbuminuria

No. of Patients Hazard Ratio (95% CI)Subgroup

0.70 (0.56– 0.87)

0.69 (0.46– 1.04)

0.81 (0.64– 1.03)

0.81 (0.68– 0.96)

0.68 (0.54– 0.86)

0.78 (0.67– 0.92)

0.84 (0.67– 1.05)

P Value forInteraction

9340

21587182

34225918

1130

8210

—0.20

0.50

0.36

0.4 1.0 2.0



KDIGO

The Final of the CV Trio

DPP-4 Inhibitors

KDIGO

0.2–0.3% reduction in HbA1c obtainedwith saxagliptin compared with placebothroughout the trial (23). However, itmust be made clear that preliminarydata demonstrate that GLP-1RA havestronger efficacy in terms of correctionof themajor risk factors for CVD (includingblood pressure and lipids) (135); indeedSAVOR, EXAMINE, and other smaller

studies did not show any significant effecton both blood pressure and lipids. If thesepurported protective effects of DPP4-Itranslate into better outcomes in peoplewith diabetes, they should be verified bythe several ongoing clinical trials. Indeed,caution should be paid when trying totranslate findings obtained in animalmodels and small clinical studies to the

heterogeneous population of diabetic pa-tients, as long as results from specificallydesigned randomized controlled trialsare not available. In addition to the afore-mentioned aspects, the effect of DPP4-Ion BM stem cells is also promising toachieve microvascular protection at dis-tant sites. Ultimately, reducing the burdenof microangiopathy may translate into

Figure 1—A schematic representation summarizing the roles of DPP-4 inhibition on diabetic microangiopathy. Experimental evidence indicates thatDPP4-I affects inflammation, vascular responses, lipids, blood pressure, and BM function. In combination with increased levels of GLP-1 andimproved glucose control, these effects can mediate protection from microvascular diabetes complications. BNP, B-type natriuretic factor; GFR,glomerular filtration rate; HIF, hypoxia inducible factor; NO, nitric oxide; TG, triglycerides; VEGF, vascular endothelial growth factor.

Table 2—Potential methodological evidence that the effects of DPP4-I on microvascular protection may be conveyedindependently of glucose control

Mechanisms References

DPP-4 inhibition improves microvascular end points in vitro (41,134)

DPP-4 inhibition improves microvascular end points in animal models of T1D (90,92,107–109)

Short-term (4–6weeks) DPP-4 inhibition in T2D has provided effects that occur before fulldevelopment of the antihyperglycemic effect (44,106)

Prevention of microvascular end point is not fully explained by improved glucose control (23)

care.diabetesjournals.org Avogaro and Fadini 2891

Avogaro A et al. Diabetes Care 2014; 37:2884-2894.

Actions of DPP-4 Inhibitors

KDIGO

Groop P-H et al. Diabetes Care 2013; 36:3460-3468.

significantly different. Safety assessmentswere performed on the pooled treated set.All analyses were performed using SASsoftware, version 9.2 (SAS Institute Inc.,Cary, NC).

RESULTSdPatient disposition isshown in Fig. 1. Of the 2,472 subjectswho were included in the treated set ofthe four clinical studies, 564 subjects hadalbuminuria (UACR 3023,000 mg/g Cr)and an eGFR $30 mL/min/1.73 m2 atscreening. Furthermore, of these 564 sub-jects, 217 subjects were receiving stabledoses of ACEIs and/or ARBs at baselineand during the 24-week treatment periodand were used in the primary analysis set(Fig. 1). Of note, only seven patients wereexcluded from the analysis because of aUACR .3,000 mg/g Cr. The sensitivityanalysis set included 249 subjects withtype 2 diabetes and prevalent albuminuriawho were not previously treated withRAAS inhibitors (Fig. 1).

Baseline demographic, clinical, andbiochemical characteristics, as well asconcomitant background therapies, werebalanced between the two treatmentgroups (Table 1). Overall, the majorityof the subjects (71%) were white. Meanage and baseline HbA1c of the study pop-ulation were 60.7 6 9.6 years and 8.3 60.9% (67.2 6 9.8 mmol/mol), respec-tively, and 68% of subjects had type 2

diabetes for more than 5 years. At studyentry, most individuals hadmicroalbumin-uria (84%) and mild or no renal impair-ment (88%); 68 and 35% of participantsreceived ACEIs or ARBs at screening, re-spectively, with only 3% of participants re-ceiving dual RAAS blockade.

EfficacyMedian UACR values were similar be-tween treatment groups at baseline: 73.8(30.122,534.4) and 80.5 (30.921,538.2)mg/g Cr in the linagliptin and placebogroups, respectively (Table 1). After 24weeks of treatment, the percentage changein adjusted geometric mean UACR frombaseline was significantly higher withlinagliptin (232% [95% CI 242 to 221];P, 0.05) compared with placebo (26%[95% CI 227 to 123]), with a between-group difference of 228% (95% CI 247to22; P= 0.0357) (Fig. 2A and B). Nota-bly, the magnitude of the albuminuria-lowering effect of linagliptin was alreadyseen after 12 weeks of treatment (229%[95%CI240 to217];P, 0.05; between-group difference: 225% [95% CI 246 to13]; P= 0.0750) (Fig. 2A and B).

Further subgroup analyses of theprimary end point were performed forrace, baseline HbA1c, and baseline SBP.For each analysis, there was no statisti-cally significant interaction between treat-ment and the relevant subgroup. The

overall effect of linagliptin was consistentwith the results from the primary analysis(Fig. 2C and Supplementary Fig. 2).

We further explored the effect oflinagliptin on the UACR in patients withrenal dysfunction who were not previ-ously treated with RAAS inhibitors. Thissensitivity analysis showed a significantreduction in the UACR from baseline toweek 24 with linagliptin (230% [95% CI240 to 219]; P, 0.05; n = 183). Thebetween-group difference of217% (95%CI 238 to 112) showed a similar trendas the primary analysis set but did notreach statistical significance (P= 0.2301;n = 249).

Important potential confounding fac-tors for the primary end point that wereconsidered included improvements inglycemic control and alterations in bloodpressure or renal function during the 24weeks of treatment in either group. Asexpected, linagliptin led to significantreductions in HbA1c. Adjusted meanchanges in HbA1c from baseline to week24 were 20.66% (27.2 mmol/mol) withlinagliptin compared with 20.05%(20.5 mmol/mol) with placebo (Fig.3A), with a between-group difference of20.61% (26.7 mmol/mol) in favor oflinagliptin (95% CI 20.88 to 20.34%[29.6 to 23.7 mmol/mol]; P, 0.0001).HbA1c alsowas stratified intoquartiles basedon the change frombaseline toweek24. Thepercentage changes in the geometric meanUACR atweek 24 for linagliptin across thesecategories were not statistically significantlydifferent (Fig. 3B). Furthermore, significantresults were seen in adjustedmean changesin FPG from baseline to week 24 (28.88mg/dL with linagliptin versus116.22 mg/dL with placebo; between-group difference225.1mg/dL [95%CI238.55 to211.65];P= 0.0003).

We found no clinically meaningfulchanges in SBP from baseline to last valueon treatment in subjects treated witheither linagliptin or placebo (Fig. 3C).SBP was stratified into three categoriesbased on the change from baseline tolast value on treatment. The percentagechanges in the geometric mean UACR atweek 24 for linagliptin across these cate-gories were not statistically significantlydifferent (Fig. 3D). There were no rele-vant differences between the distribu-tion of different RAAS inhibitors atbaseline. Moreover, changes in other bloodpressure–lowering medications duringthe study were rare and were balancedoverall between the two treatment groups(data not shown).

Figure 1dPatient disposition. *Patients might have had more than one exclusion criterion. †183participants receiving linagliptin; 66 participants receiving placebo. #For at least 4 weeks beforethe study and from baseline to the date of the last UACR measurement within the 24-weektreatment period.

3462 DIABETES CARE, VOLUME 36, NOVEMBER 2013 care.diabetesjournals.org

Albuminuria-lowering effect of linagliptin

Linagliptin and Diabetic Kidney Disease

KDIGO

the percentage changes in geometricmean UACR across the categories ofSBP for the last value on treatment. Thisindicates that blood pressure does nothave a significant influence on the effect oflinagliptin on the UACR in our analysis.

In addition, changes in hyperglyce-mia are a potentially important con-founder. In fact, large-scale interventionstudies have demonstrated that sustainedimprovements in glycemic control reducethe development and progression ofmicrovascular complications in patientswith type 2 diabetes after long-term treat-ment (30–32). Hence, it is tempting tospeculate that any reduction in albu-minuria may simply mirror concomitantimprovements in glucose control aftershort-term treatment. The evidence for apotential effect of relatively short-termglucose control (over months ratherthan years) on albuminuria in diabetesis, however, inconclusive. An improve-ment in HbA1c after 52 weeks of insulininfusiondfrom 9.5 to 7.3% (80.3 to 56.3mmol/mol)dwas not associated with anysignificant changes in renal parameters,such as GFR or urinary albumin excre-tion, in patients with insulin-dependentdiabetes and elevated urinary albuminexcretion (30–300 mg/24 h) (33).Moreover, a previous study by Tuttleet al. (34) reported that strict glycemiccontrol did lower HbA1c levels from8.4 to 6.9% (68 to 52 mmol/mol) after3 weeks of intensive insulin therapy.Although renal hemodynamic responsesto increased plasma amino acid concentra-tions were improved, the rapid HbA1creduction did not lead to significantchanges in urinary albumin excretion. Inline with these findings, improvement inglycemic control did not improve micro-albuminuria in an adolescent populationwith insulin-dependent diabetes usingeither intensive conventional therapy orinsulin infusion up to 8 months (35).These results suggest no major influenceof short-term glucose control on urinaryalbumin excretion. However, an exacttimely separation between short- andlonger-term interdependencies betweenglucose control and progression of renaldisease in type 2 diabetes is difficult. Ouranalysis does not support a direct relation-ship between short-term glucose controland changes in UACR. We found no sta-tistically significant difference between thepercentage changes in geometric meanUACR and changes in HbA1c at week 24for linagliptin. In fact, patients with onlymodest reductions in HbA1c showed

Figure 2dA:Adjusted geometric mean of percentage change in UACR from baseline to 12 and 24weeks (-, linagliptin; ▫, placebo); P , 0.05 versus baseline. Error bars represent 95% CIs.B: Adjusted geometric mean of placebo-corrected percentage change in UACR from baseline to 12(n = 226; P = 0.0750;C) and 24 weeks (n = 217; P = 0.0357;▲). Error bars represent 95% CIs.C: Adjusted geometric mean of percentage change in UACR from baseline to week 24 stratified byrace, mean baseline HbA1c, andmean baseline SBP in the linagliptin group.C, white subjects (n =113); ▲, Asian subjects (n = 45; treatment 3 race interaction, P = 0.7397);-, mean baselineHbA1c ,8.25% (n = 97); ◆, mean baseline HbA1c $8.25% (n = 65; treatment 3 baseline HbA1cinteraction, P = 0.8100); ▼, mean baseline SBP ,137.4 mmHg (n = 88); , mean baseline SBP$137.4mmHg(n=74; treatment3 baseline SBP interaction, P = 0.6475). Error bars represent 95%CIs.

3464 DIABETES CARE, VOLUME 36, NOVEMBER 2013 care.diabetesjournals.org

Albuminuria-lowering effect of linagliptin

similar changes in UACR compared withthose havingmore profound reductions inHbA1c (.1.1% [12.0 mmol/mol]) after 24weeks of treatment with linagliptin.

The mechanisms by which linagliptinmay additionally improve the effects ofRAAS inhibitors in the kidney remain tobe fully elucidated. The hypothesis of apotential albuminuria-lowering effect oflinagliptin was first raised as a result offindings from an experimental animalstudy assessing the renal effects of coad-ministration of linagliptin with telmisartan

in diabetic endothelial nitric oxide syn-thase knockout mice (16). In this modelof vascular renal damage, 11 weeks ofcombination therapy significantly reducedurinary albumin excretion, independentof changes in blood glucose, and theeffects were greater than those seen withRAAS blockade alone. The available evi-dence suggests that the albuminuria-lowering effect of linagliptin may be dueto inhibition of podocyte damage and myo-fibroblast transformation (36) as well as aconsequence of improvement in renal

inflammatory responses mediated by in-creased glucagon-like peptide-1 (GLP-1)activity (16,36) or inhibition of tumornecrosis factor-a (16).Moreover, treatmentwith linagliptin reduced plasma levels ofosteopontin, a marker of vascular calcifi-cation and progression of renal disease(16). These results are further supportedby several other experimental studies ofDPP-4 inhibitors showing beneficialeffects of sitagliptin and vildagliptin on al-buminuria and renal function in modelsof diabetic nephropathy (18–20). There is

Figure 3dA: Adjusted mean change in HbA1c from baseline to week 24 (-, linagliptin;▫, placebo). There was a between-group difference of20.61%(26.7 mmol/mol) in favor of linagliptin (95% CI20.88 to20.34% [29.6 to23.7 mmol/mol]; P, 0.0001). Error bars represent SE. B: Adjustedgeometricmean of percentage change inUACRby quartiles of HbA1c reduction in the linagliptin group:-,,0.1% (,1.1mmol/L)HbA1c reduction;▫,0.1020.59% (1.126.5mmol/L)HbA1c reduction;E, 0.6020.99% (6.6210.8mmol/L)HbA1c reduction;D,$1.0% ($10.9mmol/L)HbA1c reduction;P$ 0.05 (ANOVA F test). Error bars represent 95%CIs. C:Adjusted mean change in SBP from baseline to last value during treatment (-, linagliptin;▫, placebo). Error bars represent SE. D:Adjusted geometric mean of percentage change in UACR by categories of SBP change in the linagliptin group:-, SBP increase.1.0 mmHg;▫,21.0 mmHg# SBP change#11.0 mmHg;E, SBP decrease.1.0 mmHg; P$ 0.05 (ANOVA F test). Error barsrepresent 95% CIs.

care.diabetesjournals.org DIABETES CARE, VOLUME 36, NOVEMBER 2013 3465

Groop and Associates

Urine ACR at week 24 was reduced by 32% with Linagliptin compared with 6% with placebo, between-group difference of 28%.

Albuminuria lowering effect of Linagliptin was not influenced by race or HbA1c and systolic blood pressure values at baseline or after treatment.

Groop P-H et al. Diabetes Care 2013; 36:3460-3468.

Linagliptin and Diabetic Kidney Disease

KDIGO

Summary

KDIGO

Summary

Widerangeofanti-glycemicagents,wherebenefitshavegonebeyondjustglucosecontrolandHbA1creduction.Metforminremainsthemostwidelyrecommendeddrugforinitialmonotherapy,forbenefitsonweightreductionandCVprotection.SGLT2inhibitorsaregoingtobegamechangersforthemanagementofpatientswithType2diabetesmellitus,withimpressiveresultsoncardio-andrenalprotectivebenefits.Long-actingGLP-1ReceptorAgonist,Liraglutide,couldbeanalternativeinpatientswhoarenotabletotolerateSGLT2inhibitors.DPP-4inhibitorsmaybeaddedinpatientswithdiabetickidneydisease,whohaspoorglucosecontrolaftermetforminandSGLT2inhibitors/GLP-1receptoragonist.

KDIGO

anti-glycemic agents in 2019: digo · white jr jr. diabetes spectr 2016; 27:82-86. developmental...

Documents