Management of hyperglycemia in type 2 diabetes

ManagementofHyperglycemia inType2Diabetes: A Patient-Centered ApproachPosition Statement of the American Diabetes Association (ADA) andthe European Association for the Study of Diabetes (EASD)

SILVIO E. INZUCCHI, MD1

RICHARD M. BERGENSTAL, MD2

JOHN B. BUSE, MD, PHD3

MICHAELA DIAMANT, MD, PHD4

ELE FERRANNINI, MD5

MICHAEL NAUCK, MD6

ANNE L. PETERS, MD7

APOSTOLOS TSAPAS, MD, PHD8

RICHARD WENDER, MD9

DAVID R. MATTHEWS, MD, DPHIL10,11,12

Glycemic management in type 2 di-abetes mellitus has become increas-ingly complex and, to some extent,

controversial, with a widening array ofpharmacological agents now available (1–5),mounting concerns about their potentialadverse effects and new uncertainties re-garding the benefits of intensive glycemiccontrol on macrovascular complications(6–9). Many clinicians are therefore per-plexed as to the optimal strategies for theirpatients. As a consequence, the AmericanDiabetesAssociation (ADA) and theEuropeanAssociation for the Study of Diabetes (EASD)convened a joint task force to examine theevidence and develop recommendations forantihyperglycemic therapy in nonpregnantadults with type 2 diabetes. Several guide-line documents have been developed bymembers of these two organizations (10)and by other societies and federations(2,11–15). However, an update was

deemed necessary because of contemporaryinformation on the benefits/risks of glycemiccontrol, recent evidence concerning efficacyand safety of several new drug classes(16,17), the withdrawal/restriction of others,and increasing calls for a move toward morepatient-centered care (18,19).

This statement has been written in-corporating the best available evidenceand, where solid support does not exist,using the experience and insight of thewriting group, incorporating an extensivereview by additional experts (acknowl-edged below). The document refers toglycemic control; yet this clearly needs tobe pursued within a multifactorial riskreduction framework. This stems from thefact that patients with type 2 diabetes areat increased risk of cardiovascular morbid-ity and mortality; the aggressive manage-ment of cardiovascular risk factors (bloodpressure and lipid therapy, antiplatelet

treatment, and smoking cessation) is likelyto have even greater benefits.

These recommendations should beconsidered within the context of the needs,preferences, and tolerances of each patient;individualization of treatment is the cor-nerstone of success. Our recommenda-tions are less prescriptive than and not asalgorithmic as prior guidelines. This fol-lows from the general lack of comparative-effectiveness research in this area. Ourintent is therefore to encourage an appre-ciation of the variable and progressivenature of type 2 diabetes, the specific roleof each drug, the patient and diseasefactors that drive clinical decisionmaking(20–23), and the constraints imposed byage and comorbidity (4,6). The implemen-tation of these guidelines will requirethoughtful clinicians to integrate currentevidence with other constraints and im-peratives in the context of patient-specificfactors.

PATIENT-CENTEREDAPPROACHdEvidence-based advicedepends on the existence of primarysource evidence. This emerges onlyfrom clinical trial results in highly selectedpatients, using limited strategies. It doesnot address the range of choices available,or the order of use of additional therapies.Even if such evidence were available, thedata would show median responses andnot address the vital question of whoresponded to which therapy and why (24).Patient-centered care is defined as an ap-proach to “providing care that is respectfulof and responsive to individual patientpreferences, needs, and values and ensur-ing that patient values guide all clinical de-cisions” (25). This should be the organizingprinciple underlying health care for indi-viduals with any chronic disease, but givenour uncertainties in terms of choice or se-quence of therapy, it is particularly appro-priate in type 2 diabetes. Ultimately, it ispatients who make the final decisions re-garding their lifestyle choices and, to somedegree, the pharmaceutical interventionsthey use; their implementation occurs inthe context of the patients’ real lives and

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

From the 1Section of Endocrinology, Yale University School of Medicine and Yale-New Haven Hospital,New Haven, Connecticut; the 2International Diabetes Center at Park Nicollet, Minneapolis, Minnesota; the3Division of Endocrinology, University of North Carolina School of Medicine, Chapel Hill, North Carolina;the 4Diabetes Center/Department of Internal Medicine, VU University Medical Center, Amsterdam, theNetherlands; the 5Department ofMedicine, University of Pisa School ofMedicine, Pisa, Italy; 6DiabeteszentrumBad Lauterberg, Bad Lauterberg im Harz, Germany; the 7Division of Endocrinology, Keck School of Medicine,University of Southern California, Los Angeles, California; the 8Second Medical Department, Aristotle Uni-versity Thessaloniki, Thessaloniki, Greece; the 9Department of Family and Community Medicine, JeffersonMedical College, Thomas Jefferson University, Philadelphia, Pennsylvania; the 10Oxford Centre for Diabetes,Endocrinology and Metabolism, Churchill Hospital, Headington, Oxford, U.K.; the 11National Institute forHealth Research (NIHR), Oxford Biomedical Research Centre, Oxford, U.K.; and the 12Harris ManchesterCollege, University of Oxford, Oxford, U.K.

Corresponding author: Silvio E. Inzucchi, [email protected]: 10.2337/dc12-0413This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc12-0413/-/DC1.

S.E. Inzucchi and D.R. Matthews were co-chairs for the Position Statement Writing Group. R.M. Bergenstal,J.B. Buse, A.L. Peters, and R. Wender were the Writing Group for the ADA. M. Diamant, E. Ferrannini,M. Nauck, and A. Tsapas were the Writing Group for the EASD.

This article is being simultaneously published in 2012 in Diabetes Care and Diabetologia by the AmericanDiabetes Association and the European Association for the Study of Diabetes.

© 2012 by the American Diabetes Association and Springer-Verlag. Readers may use this article as long as thework is properly cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

See accompanying editorial, p. 1201.

1364 DIABETES CARE, VOLUME 35, JUNE 2012 care.diabetesjournals.org

R e v i e w s / C o n s e n s u s R e p o r t s / A D A S t a t e m e n t s

P O S I T I O N S T A T E M E N T

relies on the consumption of resources(both public and private).

Patient involvement in the medicaldecision making constitutes one of thecore principles of evidence-based medi-cine, which mandates the synthesis of bestavailable evidence from the literature withthe clinician’s expertise and patient’s owninclinations (26).During the clinical encoun-ter, the patient’s preferred level of involve-ment should be gauged and therapeuticchoices explored, potentially with the uti-lization of decision aids (21). In a shareddecision-making approach, clinician andpatient act as partners, mutually exchanginginformation and deliberating on options, inorder to reach a consensus on the therapeu-tic course of action (27). There is good ev-idence supporting the effectiveness of thisapproach (28). Importantly, engaging pa-tients in health care decisions may enhanceadherence to therapy.

BACKGROUND

Epidemiology and health careimpactBoth the prevalence and incidence of type 2diabetes are increasing worldwide, particu-larly in developing countries, in conjunctionwith increased obesity rates and westerni-zation of lifestyle. The attendant economicburden for health care systems is skyrocket-ing, owing to the costs associated with treat-ment and diabetes complications. Type 2diabetes remains a leading cause of car-diovascular disorders, blindness, end-stagerenal failure, amputations, and hospitaliza-tions. It is also associatedwith increased riskof cancer, serious psychiatric illness, cogni-tive decline, chronic liver disease, acceler-ated arthritis, and other disabling or deadlyconditions. Effective management strategiesare of obvious importance.

Relationship of glycemic controlto outcomesIt is well established that the risk of micro-vascular and macrovascular complicationsis related to glycemia, as measured byHbA1c; this remains a major focus of ther-apy (29). Prospective randomized trialshave documented reduced rates of micro-vascular complications in type 2 diabeticpatients treated to lower glycemic targets.In the UK Prospective Diabetes Study(UKPDS) (30,31), patients with newly di-agnosed type 2 diabetes were randomizedto two treatment policies. In the standardgroup, lifestyle intervention was the main-stay with pharmacological therapy usedonly if hyperglycemia became severe. In the

more intensive treatment arm, patients wererandomly assigned to either a sulfonylureaor insulin, with a subset of overweightpatients randomized to metformin. Theoverall HbA1c achieved was 0.9% lowerin the intensive policy group comparedwith the conventional policy arm (7.0%vs. 7.9%). Associatedwith this difference inglycemic controlwas a reduction in the riskof microvascular complications (retinopa-thy, nephropathy, neuropathy) with inten-sive therapy. A trend toward reduced ratesof myocardial infarction in this group didnot reach statistical significance (30). Bycontrast, substantially fewer metformin-treated patients experienced myocardialinfarction, diabetes-related and all-causemortality (32), despite a mean HbA1c only0.6% lower than the conventional policygroup. The UKPDS 10-year follow-updemonstrated that the relative benefit ofhaving been in the intensive managementpolicy group was maintained over a de-cade, resulting in the emergence of statisti-cally significant benefits on cardiovasculardisease (CVD) endpoints and totalmortalityin those initially assigned to sulfonylurea/insulin, and persistence of CVD benefitswith metformin (33), in spite of the factthat the mean HbA1c levels between thegroups converged soon after the ran-domized component of the trial hadconcluded.

In 2008, three shorter-term studies[Action to Control Cardiovascular Risk inDiabetes (ACCORD) (34), Action in Dia-betes and Vascular Disease: Preterax andDiamicron Modified-Release ControlledEvaluation (ADVANCE) (35), Veterans Af-fairs Diabetes Trial (VADT) (36)] reportedthe effects of two levels of glycemic controlon cardiovascular end points in middle-aged and older individuals with well-established type 2 diabetes at high risk forcardiovascular events. ACCORD and VADTaimed for an HbA1c,6.0% using complexcombinations of oral agents and insulin.ADVANCE aimed for an HbA1c #6.5%using a less intensive approach based onthe sulfonylurea gliclazide. None of thetrials demonstrated a statistically signif-icant reduction in the primary combinedcardiovascular end points. Indeed, inACCORD, a 22% increase in totalmortalitywith intensive therapy was observed,mainly driven by cardiovascular mortality.An explanation for this finding has re-mained elusive, although rates of hypogly-cemia were threefold higher with intensivetreatment. It remains unclear, however, ifhypoglycemia was responsible for the ad-verse outcomes, or if other factors, such as

more weight gain, or simply the greatercomplexity of therapy, contributed. Therewere suggestions in these trials that patientswithout overt CVD, with shorter durationof disease, and lower baseline HbA1c,benefited from the more intensive strat-egies. Modest improvements in somemicrovascular end points in the studieswere likewise demonstrated. Finally, ameta-analysis of cardiovascular out-comes in these trials suggested that everyHbA1c reduction of ;1% may be associ-ated with a 15% relative risk reduction innonfatal myocardial infarction, butwithout benefits on stroke or all-causemortality (36).

Overview of the pathogenesis oftype 2 diabetesAny rise in glycemia is the net result ofglucose influx exceeding glucose outflowfrom the plasma compartment. In the fast-ing state, hyperglycemia is directly relatedto increased hepatic glucose production.In the postprandial state, further glucoseexcursions result from the combinationof insufficient suppression of this glucoseoutput and defective insulin stimulationof glucose disposal in target tissues, mainlyskeletal muscle. Once the renal tubulartransport maximum for glucose is excee-ded, glycosuria curbs, though does notprevent, further hyperglycemia.

Abnormal islet cell function is a keyand requisite feature of type 2 diabetes. Inearly disease stages, insulin production isnormal or increased in absolute terms,but disproportionately low for the degreeof insulin sensitivity, which is typicallyreduced. However, insulin kinetics, suchas the ability of the pancreatic b-cell torelease adequate hormone in phase withrising glycemia, are profoundly compro-mised. This functional islet incompetenceis the main quantitative determinant ofhyperglycemia (37) and progresses overtime. In addition, in type 2 diabetes, pan-creatic a-cells hypersecrete glucagon, fur-ther promoting hepatic glucose production(38). Importantly, islet dysfunction is notnecessarily irreversible. Enhancing insulinaction relieves b-cell secretory burden, andany intervention that improves glycemiadfrom energy restriction to, most strikingly,bariatric surgerydcan ameliorate b-celldysfunction to an extent (39). More re-cently recognized abnormalities in the in-cretin system (represented by the guthormones, glucagon-like peptide 1 [GLP-1],and glucose-dependent insulinotropicpeptide [GIP]) are also found in type 2diabetes, but it remains unclear whether

care.diabetesjournals.org DIABETES CARE, VOLUME 35, JUNE 2012 1365

Inzucchi and Associates

these constitute primary or secondary de-fects (40). In most patients with type 2diabetes, especially the obese, insulin re-sistance in target tissues (liver, muscle,adipose tissue, myocardium) is a promi-nent feature. This results in both glucoseoverproduction and underutilization.Moreover, an increaseddelivery of fatty acidsto the liver favors their oxidation, whichcontributes to increased gluconeogenesis,whereas the absolute overabundance of lip-ids promotes hepatosteatosis (41).

Antihyperglycemic agents are directedat one or more of the pathophysiologicaldefects of type 2 diabetes, or modifyphysiological processes relating to appetiteor to nutrient absorption or excretion.Ultimately, type 2 diabetes is a diseasethat is heterogeneous in both pathogenesisand in clinical manifestationda point to beconsidered when determining the optimaltherapeutic strategy for individual patients.

ANTIHYPERGLYCEMICTHERAPY

Glycemic targetsThe ADA’s “Standards of Medical Care inDiabetes” recommends lowering HbA1cto ,7.0% in most patients to reduce theincidence of microvascular disease (42).This can be achieved with a mean plasmaglucose of;8.3–8.9 mmol/L (;150–160mg/dL); ideally, fasting and premeal glu-cose should bemaintained at,7.2mmol/L(,130 mg/dL) and the postprandial glu-cose at ,10 mmol/L (,180 mg/dL).More stringent HbA1c targets (e.g., 6.0–6.5%) might be considered in selectedpatients (with short disease duration, longlife expectancy, no significant CVD) if thiscan be achieved without significant hypo-glycemia or other adverse effects of treat-ment (20,43). Conversely, less stringentHbA1c goalsde.g., 7.5–8.0% or evenslightly higherdare appropriate for pa-tientswith ahistory of severe hypoglycemia,limited life expectancy, advanced complica-tions, extensive comorbid conditions andthose inwhom the target is difficult to attaindespite intensive self-management educa-tion, repeated counseling, and effectivedoses of multiple glucose-lowering agents,including insulin (20,44).

The accumulated results from theaforementioned type 2 diabetes cardio-vascular trials suggest that not everyonebenefits from aggressive glucose man-agement. It follows that it is important toindividualize treatment targets (5,34–36).The elements that may guide the clinicianin choosing an HbA1c target for a specific

patient are shown in Fig. 1. As mentionedearlier, the desires and values of the pa-tient should also be considered, since theachievement of any degree of glucose con-trol requires active participation and com-mitment (19,23,45,46). Indeed, any targetcould reflect an agreement between pa-tient and clinician. An important relatedconcept is that the ease with which moreintensive targets are reached influencestreatment decisions; logically, lower tar-gets are attractive if they can be achievedwith less complex regimens and no orminimal adverse effects. Importantly, uti-lizing the percentage of diabetic patientswho are achieving an HbA1c ,7.0% as aquality indicator, as promulgated by vari-ous health care organizations, is inconsis-tentwith the emphasis on individualizationof treatment goals.

Therapeutic optionsLifestyle. Interventions designed to im-pact an individual’s physical activity lev-els and food intake are critical parts oftype 2 diabetes management (47,48). All

patients should receive standardized gen-eral diabetes education (individual orgroup, preferably using an approved cur-riculum), with a specific focus on dietaryinterventions and the importance of in-creasing physical activity. While encourag-ing therapeutic lifestyle change is importantat diagnosis, periodic counseling shouldalso be integrated into the treatmentprogram.

Weight reduction, achieved throughdietary means alone or with adjunctivemedical or surgical intervention, improvesglycemic control and other cardiovascularrisk factors.Modestweight loss (5–10%)con-tributes meaningfully to achieving improvedglucose control. Accordingly, establishing agoal of weight reduction, or at least weightmaintenance, is recommended.

Dietary advice must be personalized(49). Patients should be encouraged to eathealthy foods that are consistent with theprevailing population-wide dietary rec-ommendations and with an individual’spreferences and culture. Foods high in fiber(such as vegetables, fruits,whole grains, and

Figure 1dDepiction of the elements of decision making used to determine appropriate efforts toachieve glycemic targets. Greater concerns about a particular domain are represented by in-creasing height of the ramp. Thus, characteristics/predicaments toward the left justify morestringent efforts to lower HbA1c, whereas those toward the right are compatible with lessstringent efforts. Where possible, such decisions should be made in conjunction with the patient,reflecting his or her preferences, needs, and values. This “scale” is not designed to be appliedrigidly but to be used as a broad construct to help guide clinical decisions. Adapted with per-mission from Ismail-Beigi et al. (20).


Position Statement

Table

1dProperties

ofcurrentlyavailable

glucose-low

eringagents

thatmayguidetreatm

entchoice

inindividualpatients

withtype2diabetes

mellitus

Class

Compound(s)

Cellularmechanism

Primaryphysiological

action(s)

Advantages

Disadvantages

Cost

Biguanides

cMetformin

ActivatesAMP-kinase

c↓Hepaticglucose

production

cExtensiveexperience

cNoweightgain

cNohypoglycemia

cLikely↓CVDevents

(UKPDS)

cGastrointestinalsideeffects

(diarrhea,abdominal

cramping)

cLacticacidosisrisk(rare)

cVitaminB12deficiency

cMultiplecontraindications:

CKD,acidosis,hypoxia,

dehydration,etc.

Low

Sulfonylureas

2ndgeneration

cGlyburide/

glibenclamide

cGlipizide

cGliclazideb

cGlimepiride

ClosesKATPchannels

onb-cellplasma

membranes

c↑Insulinsecretion

cExtensiveexperience

c↓Microvascularrisk

(UKPDS)

cHypoglycemia

cWeightgain

c?Bluntsmyocardialischemic

preconditioning

cLowdurability

Low

Meglitinides

(glinides)

cRepaglinide

cNateglinide

ClosesKATPchannels

onb-cellplasma

membranes


c↓Postprandialglucose

excursions

cDosingflexibility

cHypoglycemia

cWeightgain

c?Bluntsmyocardialischemic

preconditioning

cFrequentdosingschedule

High

Thiazolidinediones

cPioglitazone

cRosiglitazonec

Activatesthenuclear

transcriptionfactor

PPAR-g

c↑Insulinsensitivity

cNohypoglycemia

cDurability

c↑HDL-C

c↓Triglycerides

(pioglitazone)

c?↓CVDevents

(ProACTIVE,

pioglitazone)

cWeightgain

cEdema/heartfailure

cBonefractures

c↑LDL-C(rosiglitazone)

c?↑MI(meta-analyses,

rosiglitazone)

c?↑Bladdercancer

(pioglitazone)

Highe

a-Glucosidase

inhibitorsa

cAcarbose

cMiglitol

cVogliboseb,d

Inhibitsintestinal

a-glucosidase

cSlowsintestinal

carbohydrate

digestion/absorption

cNohypoglycemia


excursions

c?↓CVDevents

(STOP-NIDDM)

cNonsystemic

cGenerallymodestHbA1c

efficacy


(flatulence,diarrhea)


Moderate

DPP-4inhibitors

cSitagliptin

cVildagliptina

cSaxagliptin

cLinagliptin

cAlogliptinb,d

InhibitsDPP-4activity,

increasingpostprandial

activeincretin(GLP-1,

GIP)concentrations


(glucose-dependent)

c↓Glucagonsecretion

(glucose-dependent)

cNohypoglycemia

cWelltolerated


efficacy

cUrticaria/angioedema

c?Pancreatitis

High

Continuedon

p.1368



Tab

le1dCon

tinued

Class

Compound(s)

Cellularmechanism

Primaryphysiological

action(s)

Advantages

Disadvantages

Cost

Bileacid

sequestrantsa

cColesevelam

Bindsbileacidsin

intestinaltract,

increasinghepatic

bileacidproduction;

?activationoffarnesoid

Xreceptor(FXR)inliver

cUnknown

c?↓Hepaticglucose

production

c?↑Incretinlevels

cNohypoglycemia

c↓LDL-C


efficacy

cConstipation

c↑Triglycerides

cMay

↓absorptionofother

medications

High

Dopamine-2

agonistsa

cBromocriptine

(quick-release)d

Activatesdopaminergic

receptors

cModulateshypothalamic

regulationofmetabolism

c↑Insulinsensitivity

cNohypoglycemia

c?↓CVDevents

(CyclosetSafety

Trial)


efficacy

cDizziness/syncope

cNausea

cFatigue

cRhinitis

High

GLP-1receptor

agonists

cExenatide

cExenatide

extended

release

cLiraglutide

ActivatesGLP-1

receptors


(glucose-dependent)


(glucose-dependent)

cSlowsgastricemptying

c↑Satiety

cNohypoglycemia

cWeightreduction

c?Potentialfor

improvedb-cell

mass/function

c?Cardiovascular

protectiveactions


(nausea/vomiting)

c?Acutepancreatitis

cC-cellhyperplasia/medullary

thyroidtumorsinanimals

cInjectable

cTrainingrequirements

High

Amylinmimeticsa

cPramlintided

Activatesamylin

receptors


cSlowsgastricemptying

c↑Satiety


excursions

cWeightreduction


efficacy


(nausea/vomiting)

cHypoglycemiaunless

insulindoseis

simultaneouslyreduced

cInjectable


High

Insulins

cHumanNPH

cHumanRegular

cLispro

cAspart

cGlulisine

cGlargine

cDetemir

cPremixed

(severaltypes)

Activatesinsulin

receptors

c↑Glucosedisposal

c↓Hepaticglucose

production

cUniversallyeffective

cTheoreticallyunlimited

efficacy

c↓Microvascularrisk

(UKPDS)

cHypoglycemia

cWeightgain

c?Mitogeniceffects

cInjectable

cTrainingrequirements

c“Stigma”(forpatients)

Variablef

a LimiteduseintheU.S./Europe.bNotlicensedintheU.S.cPrescribinghighlyrestrictedintheU.S.;withdrawninEurope.dNotlicensedinEurope.e Tobeavailableasagenericproductin2012,withexpectedsignificant

reductionsincost.fDependsontype(analogs

.humaninsulins)anddosage.CKD,chronickidneydisease;CVD,cardiovasculardisease;DPP-4,dipeptidylpeptidase4;GIP,glucose-dependentinsulinotropicpeptide;

GLP-1,glucagon-likepeptide1;HDL-C,HDL-cholesterol;LDL-C,LDL-cholesterol;PPAR,peroxisomeproliferator–activatedreceptor;ProACTIVE,ProspectivePioglitazoneClinicalTrialinMacrovascularEvents(60);

STOP-NIDDM,StudytoPreventNon-Insulin-DependentDiabetesMellitus(134);UKPDS,UKProspectiveDiabetesStudy(29–33).


Position Statement

legumes), low-fat dairy products, and freshfish should be emphasized. High-energyfoods, including those rich in saturatedfats, and sweet desserts and snacks shouldbe eaten less frequently and in loweramounts (50–52). Patients who eventuallylose and keep weight off usually do so afternumerous cycles of weight loss and relapse.The health care team should remain non-judgmental but persistent, revisiting andencouraging therapeutic lifestyle changesfrequently, if needed.

As much physical activity as possibleshould be promoted, ideally aiming for atleast 150 min/week of moderate activityincluding aerobic, resistance, and flexi-bility training (53). In older individuals,or those with mobility challenges, solong as tolerated from a cardiovascularstandpoint, any increase in activity levelis advantageous.

At diagnosis, highly motivated pa-tients withHbA1c already near target (e.g.,,7.5%) could be given the opportunityto engage in lifestyle change for a periodof 3–6 months before embarking onpharmacotherapy (usually metformin).Those with moderate hyperglycemia orin whom lifestyle changes are anticipatedto be unsuccessful should be promptlystarted on an antihyperglycemic agent(also usually metformin) at diagnosis,which can later be modified or possiblydiscontinued if lifestyle changes are suc-cessful.Oral agents and noninsulin injectables.Important properties of antihyperglyce-mic agents that play a role in the choice ofdrug(s) in individual patients are summa-rized in Table 1. Ultimately, the aims ofcontrolling glycemia are to avoid acuteosmotic symptoms of hyperglycemia, toavoid instability in blood glucose overtime, and to prevent/delay the develop-ment of diabetes complications withoutadversely affecting quality of life. Infor-mation on whether specific agents havethis ability is incomplete; an answer tothese questions requires long-term, large-scale clinical trialsdnot available for mostdrugs. Effects on surrogate measures forglycemic control (e.g., HbA1c) generallyreflect changes in the probability of de-veloping microvascular disease but notnecessarily macrovascular complications.Particularly from a patient standpoint,stability of metabolic control over timemay be another specific goal.

Metformin, a biguanide, remains themost widely used first-line type 2 diabetesdrug; its mechanism of action predomi-nately involves reducing hepatic glucose

production (54,55). It is generally consid-ered weight-neutral with chronic use anddoes not increase the risk of hypoglycemia.Metformin is associated with initial gastro-intestinal side effects, and caution is ad-vised to avoid its use in patients at risk forlactic acidosis (e.g., in advanced renal in-sufficiency, alcoholism), a rare complica-tion of therapy. As noted earlier, theremay be some cardiovascular benefitsfrom this drug, but the clinical trial dataare not robust.

The oldest oral agent class is the sulfo-nylurea insulin secretagogues. Throughthe closure of ATP-sensitive potassiumchannels on b-cells, these drugs stimulateinsulin release (56).While effective in con-trolling glucose levels, their use is associ-ated with modest weight gain and risk ofhypoglycemia. In addition, studies havedemonstrated a secondary failure ratethat may exceed other drugs, ascribed toan exacerbation of islet dysfunction (57).Shorter-acting secretagogues, the megliti-nides (or glinides), stimulate insulin re-lease through similar mechanisms butmay be associated with less hypoglycemia(58). They require more frequent dosing,however.

Thiazolidinediones (TZDs) are per-oxisome proliferator–activated receptor gactivators (59) that improve insulin sen-sitivity in skeletal muscle and reduce he-patic glucose production (54,55). Theydo not increase the risk of hypoglycemiaandmay be more durable in their effective-ness than sulfonylureas and metformin(57). Pioglitazone appeared to have a mod-est benefit on cardiovascular events as asecondary outcome in one large trial in-volving patients with overt macrovasculardisease (60). Another agent of this class,rosiglitazone, is no longer widely availableowing to concerns of increased myocardialinfarction risk (61). Pioglitazone has re-cently been associated with a possible in-creased risk of bladder cancer (62).Recognized side effects of TZDs includeweight gain, fluid retention leading toedema and/or heart failure in predisposedindividuals, and increased risk of bonefractures (57,60).

Drugs focused on the incretin systemhave been introduced more recently (63).The injectable GLP-1 receptor agonistsmimic the effects of endogenous GLP-1,thereby stimulating pancreatic insulin se-cretion in a glucose-dependent fashion,suppressing pancreatic glucagon output,slowing gastric emptying, and decreasingappetite. Their main advantage is weightloss, which is modest in most patients but

can be significant in some. A limiting sideeffect is nausea and vomiting, particularlyearly in the course of treatment. Concernsregarding an increased risk of pancreatitisremain unresolved. The oral dipeptidylpeptidase 4 (DPP-4) inhibitors enhancecirculating concentrations of activeGLP-1 and GIP (64). Their major effectappears to be in the regulation of insulinand glucagon secretion; they are weightneutral. Typically, neither of the incretin-based classes cause hypoglycemia bythemselves.

Two agents that are used infrequentlyin theU.S. andEurope are thea-glucosidaseinhibitors (AGIs), which retard gut carbohy-drate absorption (65), and colesevelam, abile acid sequestrant whose mechanism ofglucose-lowering action remains poorlyunderstood and whose major additionalbenefit is LDL-cholesterol reduction (66).Both have gastrointestinal effects, mainlyflatulence with AGIs and constipationwith colesevelam. The dopamine agonistbromocriptine is only available in the U.S.as an antihyperglycemic agent (67). Itsmechanism of action and precise roleare unclear. The amylin agonist, pramlintide,is typically reserved for patients treatedwith intensive insulin therapy, usually intype 1 diabetes mellitus; it decreases post-prandial glucose excursions by inhibitingglucagon secretion and slowing gastricemptying (68).

The glucose-lowering effectiveness ofnoninsulin pharmacological agents is saidto be high for metformin, sulfonylureas,TZDs, and GLP-1 agonists (expectedHbA1c reduction ;1.0–1.5%) (1,69,70),and generally lower for meglitinides,DPP-4 inhibitors, AGIs, colesevelam,and bromocriptine (;0.5–1.0%). How-ever, older drugs have typically beentested in clinical trial participants withhigher baseline HbA1c, which is itself as-sociated with greater treatment emergentglycemic reductions, irrespective of ther-apy type. In head-to-head studies, anydifferential effects on glucose control aresmall. So agent- and patient-specific prop-erties, such as dosing frequency, side-effectprofiles, cost, and other benefits oftenguide their selection.Insulin. Due to the progressive b-celldysfunction that characterizes type 2 di-abetes, insulin replacement therapy is fre-quently required (71). Importantly, mostpatients maintain some endogenous insu-lin secretion even in late stages of disease.Accordingly, the more complex and in-tensive strategies of type 1 diabetes arenot typically necessary (72).



Ideally, the principle of insulin use isthe creation of as normal a glycemic profileas possible without unacceptable weightgain or hypoglycemia (73). As initial ther-apy, unless the patient is markedly hyper-glycemic and/or symptomatic, a “basal”insulin alone is typically added (74). Basalinsulin provides relatively uniform insulincoverage throughout the day and night,mainly to control blood glucose by sup-pressing hepatic glucose production inbetween meals and during sleep. Eitherintermediate-acting (neutral protamineHagedorn [NPH]) or long-acting (insulinglargine [A21Gly,B31Arg,B32Arg hu-man insulin] or insulin detemir [B29Lys(´-tetradecanoyl),desB30 human insulin])formulations may be used. The latter twoare associated with modestly less overnighthypoglycemia (insulin glargine, insulin de-temir) than NPH and possibly slightly lessweight gain (insulin detemir), but aremore expensive (75,76). Of note, the dos-ing of these basal insulin analogsmaydiffer,with most comparative trials showing ahigher average unit requirement with insu-lin detemir (77).

Although the majority of patientswith type 2 diabetes requiring insulintherapy can be successfully treated withbasal insulin alone, some, because of pro-gressive diminution in their insulin secre-tory capacity, will require prandial insulintherapy with shorter-acting insulins. Thisis typically provided in the form of therapid insulin analogs, insulin lispro(B28Lys,B29Pro human insulin), insulinaspart (B28Asp human insulin), or insulinglulisine (B3Lys,B29Glu human insulin),which may be dosed just before the meal.They result in better postprandial glucosecontrol than the less costly human regularinsulin, whose pharmacokinetic profilemakes it less attractive in this setting.

Ideally, an insulin treatment programshould be designed specifically for an in-dividual patient, to match the supply ofinsulin to his or her dietary/exercise hab-its and prevailing glucose trends, as revealedthroughself-monitoring.Anticipatedglucose-lowering effects should be balanced withthe convenience of the regimen, in thecontext of an individual’s specific therapygoals (Fig. 1).

Proper patient education regardingglucose monitoring, insulin injectiontechnique, insulin storage, recognition/treatment of hypoglycemia, and “sickday” rules is imperative. Where available,certified diabetes educators can be in-valuable in guiding the patient throughthis process.

Implementation strategiesInitial drug therapy. It is generallyagreed that metformin, if not contraindi-cated and if tolerated, is the preferred andmost cost-effective first agent (42) (Fig. 2and Supplementary Figs.). It is initiated at,or soon after, diagnosis, especially in pa-tients in whom lifestyle intervention alonehas not achieved, or is unlikely to achieve,HbA1c goals. Because of frequent gastroin-testinal side effects, it should be started at alow dose with gradual titration. Patientswith a high baseline HbA1c (e.g., $9.0%)have a low probability of achieving a near-normal target with monotherapy. It maytherefore be justified to start directlywith a combination of two noninsulinagents or with insulin itself in this circum-stance (78). If a patient presents with sig-nificant hyperglycemic symptoms and/orhas dramatically elevated plasma glucoseconcentrations (e.g., .16.7–19.4 mmol/L[.300–350 mg/dL]) or HbA1c (e.g.,$10.0–12.0%), insulin therapy should bestrongly considered from the outset. Suchtreatment is mandatory when catabolicfeatures are exhibited or, of course, ifketonuria is demonstrated, the latter re-flecting profound insulin deficiency. Im-portantly, unless there is evidence of type 1diabetes, once symptoms are relieved,

glucotoxicity resolved, and the metabolicstate stabilized, it may be possible to taperinsulin partially or entirely, transferring tononinsulin antihyperglycemic agents, per-haps in combination.

If metformin cannot be used, anotheroral agent could be chosen, such as asulfonylurea/glinide, pioglitazone, or aDPP-4 inhibitor; in occasional cases whereweight loss is seen as an essential aspect oftherapy, initial treatment with a GLP-1receptor agonist might be useful. Whereavailable, less commonly used drugs (AGIs,colesevelam, bromocriptine) might also beconsidered in selected patients, but theirmodest glycemic effects and side-effectprofiles make them less attractive candi-dates. Specific patient preferences, char-acteristics, susceptibilities to side effects,potential for weight gain and hypoglycemiashould play a major role in drug selection(20,21). (See Supplementary Figs. for ad-aptations of Fig. 2 that address specificpatient scenarios.)Advancing to dual combination therapy.Figure 2 (and Supplementary Figs.) alsodepicts potential sequences of escalatingglucose-lowering therapy beyond met-formin. If monotherapy alone does notachieve/maintain an HbA1c target over;3 months, the next step would be toadd a second oral agent, a GLP-1 recep-tor agonist, or basal insulin (5,10). No-tably, the higher the HbA1c, the morelikely insulin will be required. On average,any second agent is typically associatedwith an approximate further reduction inHbA1c of ;1% (70,79). If no clinicallymeaningful glycemic reduction (i.e., “non-responder”) is demonstrated, then, adher-ence having been investigated, that agentshould be discontinued, and anotherwith a different mechanism of actionsubstituted. With a distinct paucity oflong-term comparative-effectiveness trialsavailable, uniform recommendations onthe best agent to be combined with metfor-min cannot bemade (80). Thus, advantagesand disadvantages of specific drugs for eachpatient should be considered (Table 1).

Some antihyperglycemic medicationslead to weight gain. This may be associ-ated with worsening markers of insulinresistance and cardiovascular risk. Oneexception may be TZDs (57); weight gainassociated with this class occurs in asso-ciation with decreased insulin resistance.Although there is no uniform evidence thatincreases in weight in the range observedwith certain therapies translate into a sub-stantially increased cardiovascular risk, itremains important to avoid unnecessary

KEY POINTS

c Glycemic targets and glucose-loweringtherapies must be individualized.

c Diet, exercise, and education remainthe foundation of any type 2diabetestreatment program.

c Unless there are prevalent contra-indications, metformin is the op-timal first-line drug.

c After metformin, there are limiteddata to guide us. Combinationtherapywith an additional 1–2 oralor injectable agents is reasonable,aiming to minimize side effectswhere possible.

c Ultimately, many patients will requireinsulin therapy alone or in com-bination with other agents tomaintain glucose control.

c All treatment decisions, where possi-ble, should be made in conjunctionwith the patient, focusing on his/herpreferences, needs, and values.

c Comprehensive cardiovascular riskreduction must be a major focus oftherapy.


Position Statement

weight gain by optimal medication selec-tion and dose titration.

For all medications, considerationshould also be given to overall tolerability.Even occasional hypoglycemia may bedevastating, if severe, or merely irritating,if mild (81). Gastrointestinal side effects

may be tolerated by some, but not others.Fluid retention may pose a clinical ormerely an aesthetic problem (82). Therisk of bone fracturesmay be a specific con-cern in postmenopausal women (57).

It must be acknowledged that costsare a critical issue driving the selection of

glucose-lowering agents in many environ-ments. For resource-limited settings, lessexpensive agents should be chosen. How-ever, due consideration should be alsogiven to side effects and any necessarymonitoring, with their own cost impli-cations. Moreover, prevention of morbid

Figure 2dAntihyperglycemic therapy in type 2 diabetes: general recommendations. Moving from the top to the bottom of the figure, potentialsequences of antihyperglycemic therapy. In most patients, begin with lifestyle changes; metformin monotherapy is added at, or soon after, diagnosis(unless there are explicit contraindications). If the HbA1c target is not achieved after;3months, consider one of the five treatment options combinedwith metformin: a sulfonylurea, TZD, DPP-4 inhibitor, GLP-1 receptor agonist, or basal insulin. (The order in the chart is determined by historicalintroduction and route of administration and is notmeant to denote any specific preference.) Choice is based on patient and drug characteristics, withthe over-riding goal of improving glycemic control while minimizing side effects. Shared decision making with the patient may help in the selection oftherapeutic options. The figure displays drugs commonly used both in the U.S. and/or Europe. Rapid-acting secretagogues (meglitinides) may beused in place of sulfonylureas. Other drugs not shown (a-glucosidase inhibitors, colesevelam, dopamine agonists, pramlintide) may be used whereavailable in selected patients but have modest efficacy and/or limiting side effects. In patients intolerant of, or with contraindications for, metformin,select initial drug from other classes depicted and proceed accordingly. In this circumstance, while published trials are generally lacking, it isreasonable to consider three-drug combinations other than metformin. Insulin is likely to be more effective than most other agents as a third-linetherapy, especially when HbA1c is very high (e.g., $9.0%). The therapeutic regimen should include some basal insulin before moving to morecomplex insulin strategies (Fig. 3). Dashed arrow line on the left-hand side of the figure denotes the option of a more rapid progression from a two-drug combination directly to multiple daily insulin doses, in those patients with severe hyperglycemia (e.g., HbA1c $10.0–12.0%). DPP-4-i, DPP-4inhibitor; Fx’s, bone fractures; GI, gastrointestinal; GLP-1-RA, GLP-1 receptor agonist; HF, heart failure; SU, sulfonylurea. aConsider beginning at thisstage in patients with very high HbA1c (e.g.,$9%). bConsider rapid-acting, nonsulfonylurea secretagogues (meglitinides) in patients with irregularmeal schedules or who develop late postprandial hypoglycemia on sulfonylureas. cSee Table 1 for additional potential adverse effects and risks, under“Disadvantages.” dUsually a basal insulin (NPH, glargine, detemir) in combination with noninsulin agents. eCertain noninsulin agents may becontinued with insulin (see text). Refer to Fig. 3 for details on regimens. Consider beginning at this stage if patient presents with severe hyperglycemia($16.7–19.4 mmol/L [$300–350 mg/dL]; HbA1c $10.0–12.0%) with or without catabolic features (weight loss, ketosis, etc.).



long-term complications will likely reducelong-termexpenses attributed to thedisease.Advancing to triple combination ther-apy. Some studies have shown advantagesof adding a third noninsulin agent to atwo-drug combination that is not yet orno longer achieving the glycemic target(83–86). Not surprisingly, however, atthis juncture, the most robust responsewill usually be with insulin. Indeed, sincediabetes is associated with progressiveb-cell loss, many patients, especially thosewith long-standing disease, will eventuallyneed to be transitioned to insulin, whichshould be favored in circumstances wherethe degree of hyperglycemia (e.g., $8.5%)makes it unlikely that another drug will beof sufficient benefit (87). If triple combina-tion therapy exclusive of insulin is tried, thepatient should be monitored closely, withthe approach promptly reconsidered if itproves to be unsuccessful. Many monthsof uncontrolled hyperglycemia shouldspecifically be avoided.

In using triple combinations the es-sential consideration is obviously to useagents with complementary mechanismsof action (Fig. 2 and Supplementary Figs.).Increasing the number of drugs heightensthe potential for side effects and drug–druginteractions, raises costs, and negatively im-pacts patient adherence. The rationale, ben-efits, and side effects of each newmedicationshould be discussed with the patient. Theclinical characteristics of patients more orless likely to respond to specific combina-tions are, unfortunately, not well defined.Transitions to and titrations of insulin.Most patients express reluctance to be-ginning injectable therapy, but, if thepractitioner feels that such a transition isimportant, encouragement and educationcan usually overcome such reticence. In-sulin is typically begun at a low dose (e.g.,0.1–0.2 U kg21 day21), although largeramounts (0.3–0.4 U kg21 day21) are rea-sonable in the more severely hyperglyce-mic. The most convenient strategy iswith a single injection of a basal insulin,with the timing of administration depen-dent on the patient’s schedule and overallglucose profile (Fig. 3).

Although extensive dosing instruc-tions for insulin are beyond the scope ofthis statement, most patients can be taughtto uptitrate their own insulin dose basedon several algorithms, each essentially in-volving the addition of a small dose increaseif hyperglycemia persists (74,76,88). Forexample, the addition of 1–2 units (or, inthose already on higher doses, incrementsof 5–10%) to the daily dose once or twice

weekly if the fasting glucose levels are abovethe preagreed target is a reasonable ap-proach (89). As the target is neared, dosageadjustments should be more modest andoccur less frequently. Downward adjust-ment is advisable if any hypoglycemia oc-curs. During self-titration, frequent contact(telephone, e-mail) with the clinician maybe necessary. Practitioners themselves can,of course, also titrate basal insulin, but thiswould involve more intensive contact withthe patient than typically available in rou-tine clinical practice. Daily self-monitoringof blood glucose is of obvious importanceduring this phase. After the insulin dose isstabilized, the frequency of monitoringshould be reviewed (90).

Consideration should be given to theaddition of prandial or mealtime insulincoverage when significant postprandial

glucose excursions (e.g., to.10.0 mmol/L[.180 mg/dL]) occur. This is suggestedwhen the fasting glucose is at target butthe HbA1c remains above goal after 3–6months of basal insulin titration (91). Thesame would apply if large drops in glucoseoccur during overnight hours or in be-tween meals, as the basal insulin dose isincreased. In this scenario, the basal insulindose would obviously need to be simulta-neously decreased as prandial insulin is ini-tiated. Although basal insulin is titratedprimarily against the fasting glucose, gen-erally irrespective of the total dose, practi-tioners should be aware that the need forprandial insulin therapywill become likely themore the daily dose exceeds 0.5 U kg21

day21, especially as it approaches 1 U kg21

day21. The aim with mealtime insulin is toblunt postprandial glycemic excursions,

Figure 3dSequential insulin strategies in type 2 diabetes. Basal insulin alone is usually theoptimal initial regimen, beginning at 0.1–0.2 units/kg body weight, depending on the degree ofhyperglycemia. It is usually prescribed in conjunction with one to two noninsulin agents. Inpatients willing to takemore than one injection andwho have higher HbA1c levels ($9.0%), twice-daily premixed insulin or a more advanced basal plus mealtime insulin regimen could also beconsidered (curved dashed arrow lines). When basal insulin has been titrated to an acceptablefasting glucose but HbA1c remains above target, consider proceeding to basal plus mealtime in-sulin, consisting of one to three injections of rapid-acting analogs (see text for details). A lessstudied alternativedprogression from basal insulin to a twice-daily premixed insulindcould bealso considered (straight dashed arrow line); if this is unsuccessful, move to basal plus mealtimeinsulin. Thefigure describes the number of injections required at each stage, togetherwith the relativecomplexity and flexibility. Once a strategy is initiated, titration of the insulin dose is important, withdose adjustments made based on the prevailing glucose levels as reported by the patient. Noninsulinagents may be continued, although insulin secretagogues (sulfonylureas, meglitinides) are typicallystopped once more complex regimens beyond basal insulin are utilized. Comprehensive educationregarding self-monitoring of blood glucose, diet, exercise, and the avoidance of, and response to,hypoglycemia are critical in any patient on insulin therapy. Mod., moderate.


Position Statement

which can be extreme in some individuals,resulting in poor control during the day.Such coverage may be provided by one oftwo methods.

The most precise and flexible prandialcoverage is possible with “basal-bolus”therapy, involving the addition of premealrapid-acting insulin analog to ongoingbasal insulin. One graduated approachis to add prandial insulin before themeal responsible for the largest glucoseexcursiondtypically that with the greatestcarbohydrate content, often, but not always,the evening meal (92). Subsequently, a sec-ond injection can be administered before themeal with the next largest excursion (oftenbreakfast). Ultimately, a third injectionmaybe added before the smallest meal (oftenlunch) (93). The actual glycemic benefitsof these more advanced regimens afterbasal insulin are generallymodest in typicalpatients (92). So, again, individualizationof therapy is key, incorporating the degreeof hyperglycemia needing to be ad-dressed and the overall capacities of thepatient. Importantly, data trends fromself-monitoring may be particularly help-ful in titrating insulins and their doseswithin these more advanced regimens tooptimize control.

A second, perhaps more convenientbut less adaptable method involves “pre-mixed” insulin, consisting of a fixed com-bination of an intermediate insulin withregular insulin or a rapid analog. Tradi-tionally, this is administered twice daily,before morning and evening meals. Ingeneral, when compared with basal insu-lin alone, premixed regimens tend tolower HbA1c to a larger degree, but oftenat the expense of slightly more hypogly-cemia and weight gain (94). Disadvan-tages include the inability to titrate theshorter- from the longer-acting compo-nent of these formulations. Therefore,this strategy is somewhat inflexible butmay be appropriate for certain patientswho eat regularly and may be in needof a simplified approach beyond basal in-sulin (92,93). (An older and less commonlyused variation of this two-injection strategyis known as “split-mixed,” involving a fixedamount of intermediate insulin mixed bythe patient with a variable amount of regu-lar insulin or a rapid analog. This allows forgreater flexibility in dosing.)

The key messages from dozens ofcomparative insulin trials in type 2 diabetesinclude the following:

1. Any insulin will lower glucose andHbA1c.

2. All insulins are associated with someweight gain and some risk of hypo-glycemia.

3. The larger the doses and the moreaggressive the titration, the lower theHbA1c, but often with a greater likeli-hood of adverse effects.

4. Generally, long-acting insulin analogsreduce the incidence of overnight hy-poglycemia, and rapid-acting insulinanalogs reduce postprandial glucoseexcursions as compared with corre-sponding human insulins (NPH, Reg-ular), but they generally do not result inclinically significantly lower HbA1c.

Metformin is often continued whenbasal insulin is added, with studies dem-onstrating less weight gain when the twoare used together (95). Insulin secretago-gues do not seem to provide for additionalHbA1c reduction or prevention of hypo-glycemia or weight gain after insulin isstarted, especially after the dose is titratedand stabilized. When basal insulin isused, continuing the secretagogue mayminimize initial deterioration of glycemiccontrol. However, secretagogues shouldbe avoided once prandial insulin regi-mens are employed. TZDs should be re-duced in dose (or stopped) to avoidedema and excessive weight gain, al-though in certain individuals with largeinsulin requirements from severe insulinresistance, these insulin sensitizers may bevery helpful in lowering HbA1c and mini-mizing the required insulin dose (96).Data concerning the glycemic benefits ofincretin-based therapy combined withbasal insulin are accumulating; combina-tion with GLP-1 receptor agonists may behelpful in some patients (97,98). Onceagain, the costs of these more elaboratecombined regimens must be carefullyconsidered.

OTHER CONSIDERATIONS

AgeOlder adults (.65–70 years) often have ahigher atherosclerotic disease burden, re-duced renal function, and more comor-bidities (99,100). Many are at risk foradverse events from polypharmacy andmay be both socially and economicallydisadvantaged. Life expectancy is reduced,especially in the presence of long-termcomplications. They are also more likelyto be compromised by hypoglycemia; forexample, unsteadiness may result in fallsand fractures (101), and a tenuous cardiacstatus may deteriorate into catastrophic

events. It follows that glycemic targets forelderly with long-standing or more com-plicated disease should be less ambitiousthan for the younger, healthier individuals(20). If lower targets cannot be achievedwith simple interventions, an HbA1c of,7.5–8.0% may be acceptable, transition-ing upward as age increases and capacityfor self-care, cognitive, psychological andeconomic status, and support systemsdecline.

While lifestyle modification can besuccessfully implemented across all age-groups, in the aged, the choice of anti-hyperglycemic agent should focus ondrug safety, especially protecting againsthypoglycemia, heart failure, renal dys-function, bone fractures, and drug–druginteractions. Strategies specifically mini-mizing the risk of low blood glucose maybe preferred.

In contrast, healthier patients withlong life expectancy accrue risk for vas-cular complications over time. Therefore,lower glycemic targets (e.g., an HbA1c,6.5–7.0%) and tighter control of bodyweight, blood pressure, and circulatinglipids should be achieved to prevent ordelay such complications. This usually re-quires combination therapy, the early in-stitution of which may have the bestchance of modifying the disease processand preserving quality of life.

WeightThe majority of individuals with type 2diabetes are overweight or obese (;80%)(102). In these, intensive lifestyle inter-vention can improve fitness, glycemiccontrol, and cardiovascular risk factorsfor relatively small changes in bodyweight (103). Although insulin resistanceis thought of as the predominate driver ofdiabetes in obese patients, they actuallyhave a similar degree of islet dysfunctionto leaner patients (37). Perhaps as a result,the obese may be more likely to requirecombination drug therapy (20,104).While common practice has favored met-formin in heavier patients, because ofweight loss/weight neutrality, this drugis as efficacious in lean individuals (75).TZDs, on the other hand, appear to bemore effective in those with higher BMIs,although their associated weight gainmakes them, paradoxically, a less attractiveoption here. GLP-1 receptor agonists areassociated with weight reduction (38),which in some patients may be substantial.

Bariatric surgery is an increasinglypopular option in severe obesity. Type 2diabetes frequently resolves rapidly after



these procedures. Themajority of patientsare able to stop some, or even all, of theirantihyperglycemic medications, althoughthe durability of this effect is not known(105).

In lean patients, consideration shouldbe given to the possibility of latent autoim-mune diabetes in adults (LADA), a slowlyprogressive form of type 1 diabetes. Theseindividuals, while presenting with mildhyperglycemia, often responsive to oralagents, eventually develop more severehyperglycemia and require intensive insu-lin regimens (106).Measuring titres of islet-associated autoantibodies (e.g., anti-GAD)may aid their identification, encouraging amore rapid transition to insulin therapy.

Sex/racial/ethnic/genetic differencesWhile certain racial/ethnic features thatincrease the risk of diabetes are well recog-nized [greater insulin resistance in Latinos(107), more b-cell dysfunction in EastAsians (108)], using this information tocraft optimal therapeutic strategies is in itsinfancy. This is not surprising given thepolygenic inheritance pattern of the dis-ease. Indeed, while matching a drug’smechanism of action to the underlyingcauses of hyperglycemia in a specific patientseems logical, there are few data that com-pare strategies based on this approach(109). There are few exceptions, mainlyinvolving diabetes monogenic variants of-ten confused with type 2 diabetes, such asmaturity-onset diabetes of the young(MODY), several forms of which respondpreferentially to sulfonylureas (110).While there are no prominent sex differ-ences in the response to various antihyper-glycemic drugs, certain side effects (e.g.,bone loss with TZDs) may be of greaterconcern in women.

ComorbiditiesCoronary artery disease. Given the fre-quency with which type 2 diabetic patientsdevelop atherosclerosis, optimal manage-ment strategies for thosewith or at high riskfor coronary artery disease (CAD) areimportant. Since hypoglycemia may ex-acerbate myocardial ischemia and maycause dysrhythmias (111), it follows thatmedications that predispose patients tothis adverse effect should be avoided, ifpossible. If they are required, however, toachieve glycemic targets, patients shouldbe educated to minimize risk. Because ofpossible effects on potassium channels inthe heart, certain sulfonylureas have beenproposed to aggravate myocardial ischemiathrough effects on ischemic preconditioning

(112), but the actual clinical relevance of thisremains unproven. Metformin may havesome cardiovascular benefits and wouldappear to be a useful drug in the settingof CAD, barring prevalent contraindica-tions (32). In a single study, pioglitazonewas shown to reduce modestly major ad-verse cardiovascular events in patientswith established macrovascular disease.It may therefore also be considered, unlessheart failure is present (60). In very pre-liminary reports, therapy with GLP-1 re-ceptor agonists and DPP-4 inhibitors hasbeen associated with improvement in ei-ther cardiovascular risk or risk factors, butthere are no long-term data regarding clin-ical outcomes (113). There are very limiteddata suggesting that AGIs (114) and bromo-criptine (115) may reduce cardiovascularevents.Heart failure. With an aging populationand recent decreases in mortality aftermyocardial infarction, the diabetic patientwith progressive heart failure is an in-creasingly common scenario (116). Thispopulation presents unique challengesgiven their polypharmacy, frequent hos-pitalizations, and contraindications tovarious agents. TZDs should be avoided(117,118). Metformin, previously contra-indicated in heart failure, can now be usedif the ventricular dysfunction is not se-vere, if patient’s cardiovascular status isstable, and if renal function is normal(119). As mentioned, cardiovascular ef-fects of incretin-based therapies, includ-ing those on ventricular function, arecurrently under investigation (120).Chronic kidney disease. Kidney diseaseis highly prevalent in type 2 diabetes, andmoderate to severe renal functional im-pairment (eGFR ,60 mL/min) occurs inapproximately 20–30% of patients(121,122). The individual with progres-sive renal dysfunction is at increased riskfor hypoglycemia, which is multifactorial.Insulin and, to some degree, the incretinhormones are eliminated more slowly, asare antihyperglycemic drugs with renalexcretion. Thus, dose reduction may benecessary, contraindications need to beobserved, and consequences (hypoglyce-mia, fluid retention, etc.) require carefulevaluation.

Current U.S. prescribing guidelineswarn against the use of metformin inpatients with a serum creatinine $133mmol/L ($1.5 mg/dL) in men or 124mmol/L ($1.4 mg/dL) in women. Metfor-min is eliminated renally, and cases of lacticacidosis have been described in patientswith renal failure (123). There is an

ongoing debate, however, as to whetherthese thresholds are too restrictive andthat those with mild–moderate renal im-pairment would gain more benefit thanharm from using metformin (124,125). Inthe U.K., the National Institute for Healthand Clinical Excellence (NICE) guidelinesare less proscriptive and more evidence-based than those in the U.S., generally al-lowing use down to a GFR of 30 mL/min,with dose reduction advised at 45 mL/min(14). Given the current widespread report-ing of estimated GFR, these guidelinesappear very reasonable.

Most insulin secretagogues undergosignificant renal clearance (exceptions in-clude repaglinide and nateglinide) andthe risk of hypoglycemia is thereforehigher in patients with chronic kidneydisease (CKD). For most of these agents,extreme caution is imperative atmore severedegrees of renal dysfunction. Glyburide(known as glibenclamide in Europe),which has a prolonged duration ofaction and active metabolites, shouldbe specifically avoided in this group.Pioglitazone is not eliminated renally, andtherefore there are no restrictions for usein CKD. Fluid retention may be a concern,however. Among the DPP-4 inhibitors,sitagliptin, vildagliptin, and saxagliptinshare prominent renal elimination. In theface of advanced CKD, dose reduction isnecessary. One exception is linagliptin,which is predominantly eliminated enter-ohepatically. For the GLP-1 receptor ago-nists exenatide is contraindicated in stage4–5 CKD (GFR ,30 mL/min) as it is re-nally eliminated; the safety of liraglutide isnot established in CKD though pharmaco-kinetic studies suggest that drug levels areunaffected as it does not require renal func-tion for clearance.

More severe renal functional impair-ment is associated with slower elimina-tion of all insulins. Thus doses need to betitrated carefully, with some awarenessfor the potential for more prolongedactivity profiles.Liver dysfunction. Individuals with type2 diabetes frequently have hepatosteatosisas well as other types of liver disease(126). There is preliminary evidence thatpatients with fatty liver may benefit fromtreatment with pioglitazone (45,127,128).It should not be used in an individual withactive liver disease or an alanine transami-nase level above 2.5 times the upper limit ofnormal. In those with steatosis but milderliver test abnormalities, this insulin sensi-tizer may be advantageous. Sulfonylureascan rarely cause abnormalities in liver tests


Position Statement

but are not specifically contraindicated;meglitinides can also be used. If hepaticdisease is severe, secretagogues should beavoided because of the increased risk ofhypoglycemia. In patients with mild he-patic disease, incretin-based drugs can beprescribed, except if there is a coexistinghistory of pancreatitis. Insulin has no re-strictions for use in patients with liver im-pairment and is indeed the preferred choicein those with advanced disease.Hypoglycemia. Hypoglycemia in type 2diabetes was long thought to be a trivialissue, as it occurs less commonly than intype 1 diabetes. However, there is emerg-ing concern based mainly on the resultsof recent clinical trials and some cross-sectional evidence of increased risk ofbrain dysfunction in those with repeatedepisodes. In the ACCORD trial, the fre-quency of both minor and major hypo-glycemia was high in intensively managedpatientsdthreefold that associated withconventional therapy (129). It remainsunknown whether hypoglycemia wasthe cause of the increased mortality inthe intensive group (130,131). Clearly,however, hypoglycemia is more danger-ous in the elderly and occurs consistentlymore often as glycemic targets are low-ered. Hypoglycemia may lead to dys-rhythmias, but can also lead to accidentsand falls (which are more likely to be dan-gerous in the elderly) (132), dizziness(leading to falls), confusion (so other ther-apies may not be taken or taken incor-rectly), or infection (such as aspirationduring sleep, leading to pneumonia). Hy-poglycemia may be systematically under-reported as a cause of death, so the trueincidence may not be fully appreciated.Perhaps just as importantly, additional con-sequences of frequent hypoglycemia in-clude work disability and erosion of theconfidence of the patient (and that of familyor caregivers) to live independently. Accord-ingly, in at-risk individuals, drug selectionshould favor agents that do not precipitatesuch events and, in general, blood glucosetargets may need to be moderated.

FUTURE DIRECTIONS/RESEARCH NEEDSdFor antihyper-glycemic management of type 2 diabetes,the comparative evidence basis to date isrelatively lean, especially beyond metfor-minmonotherapy (70). There is a significantneed for high-quality comparative-effectiveness research, not only regardingglycemic control, but also costs and thoseoutcomes that matter most to patientsdquality of life and the avoidance of morbid

and life-limiting complications, especiallyCVD (19,23,70). Another issue about whichmore data are needed is the concept of du-rability of effectiveness (often ascribed tob-cell preservation), which would serveto stabilize metabolic control and decreasethe future treatment burden for patients.Pharmacogenetics may very well informtreatment decisions in the future, guidingthe clinician to recommend a therapy for anindividual patient based on predictors ofresponse and susceptibility to adverse ef-fects. We need more clinical data on howphenotype and other patient/disease char-acteristics should drive drug choices. Asnew medications are introduced to thetype 2 diabetes pharmacopeia, their benefitand safety should be demonstrated in stud-ies versus best current treatment, substan-tial enough both in size and duration toprovide meaningful data on meaningfuloutcomes. It is appreciated, however, thathead-to-head comparisons of all combina-tions and permutations would be impossi-bly large (133). Informed judgment and theexpertise of experienced clinicians willtherefore always be necessary.

AcknowledgmentsdThis position statementwas written by joint request of the ADA andthe EASD Executive Committees, which haveapproved the final document. The process in-volved wide literature review, three face-to-facemeetings of the Writing Group, several tele-conferences, and multiple revisions via e-mailcommunications.We gratefully acknowledge the following

experts who provided critical review of adraft of this statement: James Best, MelbourneMedical School, The University of Melbourne,Melbourne, Australia; Henk Bilo, Isala Clinics,Zwolle, the Netherlands; John Boltri, WayneState University School of Medicine, Detroit,MI; Thomas Buchanan, Keck School ofMedicine, University of Southern California,Los Angeles, CA; Paul Callaway, University ofKansas School of Medicine-Wichita, Wichita,KS; Bernard Charbonnel, University of Nantes,Nantes, France; Stephen Colagiuri, The Uni-versity of Sydney, Sydney, Australia; SamuelDagogo-Jack, The University of Tennessee HealthScience Center, Memphis, TN; Margo Farber,Detroit Medical Center, Detroit, MI; CynthiaFritschi, College of Nursing, University of Illi-nois at Chicago, Chicago, IL; Rowan Hillson,The Hillingdon Hospital, Uxbridge, U.K.;Faramarz Ismail-Beigi, Case Western ReserveUniversity School of Medicine/ClevelandVA Medical Center, Cleveland, OH; DevanKansagara, Oregon Health & Science University/Portland VA Medical Center, Portland, OR;Ilias Migdalis, NIMTS Hospital, Athens,Greece; Donna Miller, Keck School of Medicine,University of Southern California, Los Angeles,

CA; Robert Ratner, MedStar Health ResearchInstitute/Georgetown University School ofMedicine, Washington, DC; Julio Rosenstock,Dallas Diabetes and Endocrine Center at Medi-cal City, Dallas, TX; Guntram Schernthaner,Rudolfstiftung Hospital, Vienna, Austria; RobertSherwin, Yale University School of Medicine,New Haven, CT; Jay Skyler, Miller School ofMedicine, University of Miami, Miami, FL;Geralyn Spollett, Yale University School ofNursing, New Haven, CT; Ellie Strock, In-ternational Diabetes Center, Minneapolis, MN;Agathocles Tsatsoulis, University of Ioannina,Ioannina, Greece; Andrew Wolf, University ofVirginia School of Medicine, Charlottesville,VA; Bernard Zinman, Mount Sinai Hospital/University of Toronto, Toronto, ON, Canada.The American Association of Diabetes Edu-cators, American College of Physicians, andThe Endocrine Society, and several other or-ganizations who wished to remain anony-mous nominated reviewers who providedinput on the final draft. Such feedback doesnot constitute endorsement by these groupsor these individuals. The final draft was alsopeer reviewed and approved by the ProfessionalPracticeCommittee of theADAand the Panel forOverseeing Guidelines and Statements of theEASD. We are indebted to Dr. Sue Kirkman ofthe ADA for her guidance and support duringthis process. We also thank Carol Hill and MaryMerkin for providing administrative assistance.FundingThe three face-to-face meetings and the

travel of some of the writing group were sup-ported by the EASD and ADA. D.R. Matthewsacknowledges support from the National In-stitute for Health Research.Duality of interestDuring the past 12 months, the following

relationships with companies whose productsor services directly relate to the subject matterin this document are declared:R.M. Bergenstal: membership of scientific

advisory boards and consultation for or clinicalresearch support with Abbott Diabetes Care,Amylin, Bayer, Becton Dickinson, BoehringerIngelheim,Calibra,DexCom, Eli Lilly, Halozyme,Helmsley Trust, Hygieia, Johnson & Johnson,Medtronic, NIH, Novo Nordisk, Roche, Sanofi,and Takeda (all under contracts with hisemployer). Inherited stock in Merck (heldby family)J.B. Buse: research and consulting with

Amylin Pharmaceuticals, Inc.; AstraZeneca;Biodel Inc.; Boehringer Ingelheim; Bristol-Myers SquibbCompany; Diartis Pharmaceuticals,Inc.; Eli Lilly and Company; F. Hoffmann-LaRoche Ltd; Halozyme Therapeutics; Johnson& Johnson; Medtronic MiniMed; Merck &Co., Inc.; Novo Nordisk; Pfizer Inc.; Sanofi;and TransPharma Medical Ltd (all undercontracts with his employer)M. Diamant:member of advisory boards of

Abbott Diabetes Care, Eli Lilly, Merck Sharp&Dohme (MSD), Novo Nordisk, Poxel Pharma.Consultancy for: Astra-BMS, Sanofi. Speakerengagements: Eli Lilly, MSD, Novo Nordisk.



Education

Management of hyperglycemia in type 2 diabetes