Microvascular Disease in PatientsWith Diabetes With Heart Failureand Reduced Ejection VersusPreserved Ejection FractionDiabetes Care 2019;42:1792–1799 | https://doi.org/10.2337/dc18-2515

OBJECTIVE

Microvascular complications are common among patients with diabetes mellitus(DM). The presence of heart failure (HF) is presumed to be due to macrovasculardisease (typically HF with reduced ejection fraction [HFrEF] following myocardialinfarction). We hypothesized that HF with preserved ejection fraction (HFpEF) inpatients with DMmay be a manifestation of microvascular disease compared withHFrEF. The objective of this study was to examine the prevalence and associationwith clinical outcome of microvascular complications in patients with HF and DM.

RESEARCH DESIGN AND METHODS

We investigated the prevalence, association with clinical outcome, and cardiacstructure and function of microvascular (neuropathy, nephropathy, and retinop-athy) complications of DM in 2,800 prospectively enrolled participants with HF and DM(561 with HFpEF) from the Asian Sudden Cardiac Death In Heart Failure (ASIAN-HF)registry.

RESULTS

A total of 601 (21.5%) participants with DM had microvascular complications.Participants with DM and any (one or more) microvascular complications were morelikely to have HFpEF (odds ratio 1.70 [95% CI 1.15–2.50]; P = 0.008). Furthermore, thelikelihood of having HFpEF increased with an increasing number of microvascularcomplications (Ptrend < 0.001). Microvascular complications were associated withmore left ventricular (LV) hypertrophy and a greater reduction in quality of life inHFpEF than HFrEF (Pinteraction < 0.001 for all). Compared with participants with DMand without microvascular complications, the adjusted hazard ratio for thecomposite outcome of all-cause death or HF hospitalization was 1.35 (95% CI1.04–1.76) for participants with DM and microvascular complications regardless ofHF type (Pinteraction = 0.112).

CONCLUSIONS

Diabetic microvascular disease is more common, and related to greater LV remodel-ing, more impairment of quality in life, and similar adverse outcomes, in participantswith HFpEF compared with HFrEF. HFpEF may be a clinical manifestation of micro-vascular disease in DM.

In patients with diabetes mellitus (DM), microvascular disease is traditionally recog-nized as the presence of neuropathy, retinopathy, or nephropathy (1). In contrast,cardiac involvement in DM is usually categorized as macrovascular disease; namely,epicardial coronary artery disease with myocardial infarction leading to heart failure(HF) with reduced ejection fraction (HFrEF). However, patients with DM are increasingly

1National Heart Centre Singapore, Singapore2UniversityMedical CentreGroningen, Universityof Groningen, Groningen, the Netherlands3Duke-NUS Medical School, Singapore4Department of Cardiology, National UniversityHeart Centre, Singapore5Dayanand Medical College and Hospital,Ludhiana, India6Department of Cardiology, Sir Ganga RamHospital, New Delhi, India7Changi General Hospital, Singapore8Department of Medicine, Yong Loo Lin Schoolof Medicine, National University Singapore,Singapore9Institute of Cardiovascular and Medical Sciencesand School of Medicine, Dentistry and Nursing,University of Glasgow, Glasgow, U.K.10Cardiovascular Research Institute, NationalUniversity Health System, Singapore11Christchurch Heart Institute, University ofOtago, Christchurch, New Zealand12Veterans Affairs Medical Center, Minneapolis,MN

Corresponding author: Carolyn S.P. Lam, carolyn.lam@duke-nus.edu.sg

Received 9 December 2018 and accepted 27May2019

Clinical trial reg. no. NCT01633398, clinicaltrials.gov

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc18-2515/-/DC1.

This article is featured in a podcast available athttp://www.diabetesjournals.org/content/diabetes-core-update-podcasts.

*A complete list of the members of the ASIAN-HF Investigators can be found in the Supple-mentary Data online.

© 2019 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered. More infor-mation is available at http://www.diabetesjournals.org/content/license.

Jasper Tromp,1,2,3 Shir Lynn Lim,4

Wan Ting Tay,1 Tiew-Hwa Katherine Teng,1

Chanchal Chandramouli,1

Wouter Ouwerkerk,1 Gupreet S. Wander,5

Jitendra P.S. Sawhney,6 Jonathan Yap,1

Michael R. MacDonald,7 Lieng Hsi Ling,4,8

Naveed Sattar,9 John J.V. McMurray,9

A. Mark Richards,4,8,10,11 Inder Anand,12

and Carolyn S.P. Lam,1,2,3 on behalf of the

ASIAN-HF Investigators*

1792 Diabetes Care Volume 42, September 2019

CARDIOVASCULA

RANDMETABOLICRISK

recognized to be at risk for HFwith preservedejection fraction (HFpEF), wheremicrovas-cular disease is postulated to play a dom-inant role compared with HFrEF (2,3).Prior clinical studies of diabetic micro-

vascular disease inHFwere limited toonlyone of the HF types (HFrEF or HFpEF) andtherefore could not distinguish the rela-tive association of diabetic microvasculardisease with HFpEF from that with HFrEF.Among participants with DM and HFrEFin the Beta-blocker Evaluation of SurvivalTrial (BEST), 32% had microvascular com-plications (neuropathy, retinopathy, ornephropathy), which were associatedwith worse outcomes (4). A recent reportfrom the Treatment of Preserved CardiacFunction Heart Failure with an Aldoste-rone Antagonist Trial (TOPCAT) showedthat 32% of participants with DM andHFpEF had microvascular complicationsda group that similarly had more adverseoutcomes compared with those withoutmicrovascular disease (5). Neither studiesexamined the association of diabetic mi-crovascular disease with cardiac structureand function.We hypothesized that diabetic micro-

vascular disease in noncardiac organsystems (neuropathy, retinopathy, ornephropathy) may be more prevalentin participants with HFpEF comparedwith HFrEF and associated with greaterleft ventricular (LV) hypertrophy andhigher rates of all-cause mortality andHF hospitalization, thus implying thatHFpEF in patients with DM may be amanifestation of predominant microvas-cular disease. To test these hypotheses,we compared the prevalence and asso-ciation of DM microvascular complica-tions with cardiac structure and functionas well as outcomes in participants withHFrEF and HFpEF enrolled in the sameprospective HF study.

RESEARCH DESIGN AND METHODS

Study PopulationWe studied 6,438 HF participants en-rolled in the Asian Sudden Cardiac DeathIn Heart Failure (ASIAN-HF) registry (6–9),which included participants from 46centers across 11 Asian regions (Taiwan,Hong Kong, China, India, Malaysia, Thai-land, Singapore, Indonesia, Philippines,Japan, and Korea). Inclusion criteriawere age .18 years and symptomaticHF (at least one previous episode ofdecompensated HF in the previous 6months resulting in a hospital admission

or treatment in outpatient clinic). Par-ticipants were excluded if their HF wascaused by valvular disease, presence ofa comorbidity leading to a life expec-tancy,1 year, and if they were unable orunwilling to give consent. ASIAN-HF wasoriginally designed to only include par-ticipants with HFrEF (LV ejection frac-tion [LVEF] ,40%) (6,7), but a protocolamendment was undertaken in 2013 toalso include participants with HFpEF(LVEF $50% and excluding any patientwith a prior LVEF ,50%) (9). Data ondemographics, previous medical history,clinical symptoms, and functional statuswere collected. Health status wasmeasuredusing the Kansas City CardiomyopathyQuestionnaire (KCCQ), a 23-item self-administered HF-specific questionnaire val-idated inmultiple HF-related disease states(10–13). Calculated KCCQ domain scoresranged from 0 to 100; higher scores rep-resent better health status. Participantsunderwent standardized 12-lead electro-cardiography and transthoracic echocardi-ography at inclusion. Participants werefollowed up for 3 years, and outcomeswere adjudicated by an independent com-mittee. All data were captured prospec-tively in an electronic database, withregistry operations and data managementhandled by Quintiles Outcomes as thecontract research organization appointedby the academic Executive Committee.Ethics approvals were obtained from therelevant human ethics committees at allsites. All participants provided informedconsent, and this study adheres to theprinciples ofmedical research as laid downin the Declaration of Helsinki.

Study DefinitionsDM microvascular complications wereascertained by self-reported history andmedical record review. DM was definedas a clinical history ofDMand/or receivingantidiabetes therapy. If a clinical historyofDM was present, the investigators wereasked if the patient had a history ofmicrovascular disease and, if yes, thetype(s) of microvascular disease (neurop-athy, nephropathy, or retinopathy) wasspecified. The definitions of other comor-bidities in ASIAN-HF have been reported(6,7,9). Obesity was defined, in accor-dance with the World Health Organiza-tion, as BMI$30 kg/m2. Coronary arterydisease (CAD)wasdefinedasdocumentedangiographic presence of significant cor-onary obstruction (.50%diameter loss in

at least one major epicardial coronaryartery), history of myocardial infarction,or prior coronary revascularization. Hy-pertension was defined as the clinicaldiagnosis (blood pressure$140/90 mmHg)and/or receiving antihypertensive therapy.Estimated glomerular filtration rate (eGFR)was calculated using the Modification ofDiet in Renal Disease Study equation, andchronic kidney disease was defined aseGFR ,60 mL/min/1.73 m2.

EchocardiographyThe collection and processing of echo-cardiographic data have previously beendescribed (8,14). Echocardiographywas performed at each center accordingto internationally accepted guidelines(15). In addition to LVEF, LV, and leftatrial dimensions, stroke volume, cardiacoutput, and echo Doppler estimates of LVdiastolic dysfunction were collected. TheCardiovascular Imaging Core Laboratoryof the National University Health System,Singapore, provided oversight and imag-ing protocol guidelines as well as qualityassurance of echocardiograms. Accuracyand reproducibility of interpreted resultswere ensured through consistent trainingand systematic analytical processes pro-vided by the Core Laboratory according tointernational guidelines (15). For furthercalculations, LVmass was calculated fromlinear dimensions and indexed to bodysurface area (BSA) (15). Relative wallthickness (RWT) was calculated by theformula [(2 3 diastolic posterior wallthickness)/diastolic LV internal diameter].LV hypertrophy was determined as LVmass index.115 g/m2 in men and.95g/m2 in women (15). Normal LV geom-etry was defined as having no LV hy-pertrophy and an RWT#0.42. AbnormalLV geometry was classified as concentricremodeling (no LV hypertrophy and RWT.0.42), concentric hypertrophy (LV hy-pertrophy and RWT .0.42), and eccen-tric hypertrophy (LV hypertrophy andRWT #0.42). Left atrial size was indexedto BSA (15). In additional analyses, weinvestigated whether microvascularcomplications were associated with di-astolic dysfunction defined as e9 septal,8 and/or e9 lateral ,10 and E/A $0.8and/or deceleration time #200 and/orE/e9 $9 (16).

OutcomesThe primary outcome of this studywas a composite of all-cause death or

care.diabetesjournals.org Tromp and Associates 1793

HF rehospitalization at 1 year. A total of5,831 (90.6%) participants had outcomesdata available, whereas 607 (9.4%) partic-ipants were lost to follow-up. Participantswith ,1 year of outcomes available werecensored at their last known visit date.Secondary outcomes were all-cause mor-tality alone and HF hospitalization alone.

Statistical AnalysisDifferences between participants with DMwith and without microvascular complica-tions were tested with Student t test, x2

test, or theMann-WhitneyU test depend-ing on the nature and distribution of thevariable. Multivariable association be-tween microvascular complications andHFrEF/HFpEF (as thedependent variable)was tested using multivariable logisticregression, correcting for age, sex, his-tory of CAD, previous stroke, peripheralartery disease, hypertension, atrial fibril-lation, ethnicity, New York Heart Asso-ciation (NYHA) class, duration of HF, BMI,usage of b-blockers, ACE inhibitors (ACEi)/angiotensin receptorblockers (ARB), insulin,oral antidiabetes medications, creatinine,hemoglobin, type of DM, and duration ofDM in years. The association between mi-crovascular disease and echocardiographicvariables was tested using multivariablelogistic regression with subjects with DMwithout microvascular complications as areference. Kaplan-Meier curves stratifiedby group membership were depicted, withdifferences between groups tested usingthe log-rank test for survival. Multivari-able analyses were performed using Coxregression analysis correcting for the afore-mentioned model. When analyzing thesecondary outcome of HF hospitalizations,we used all-cause mortality as a competingrisk. All tests were performed two sided,and P values of ,0.05 were consideredstatistically significant. Statistical analy-ses were performed using Stata 15.0(StataCorp, College Station, TX).

RESULTS

Baseline CharacteristicsAmong the 6,438 participants in theASIAN-HF registry, 2,800participants hadDM, of whom 601 (21%) had microvas-cular complications. Among 2,800 par-ticipants with DM, those with HFpEF hada higher prevalence (27%) of microvas-cular complications compared with thosewith HFrEF (20%; P = 0.001) (Fig. 1A).After adjustment for age, sex, ethnic-ity, history of CAD, atrial fibrillation,

previous stroke, peripheral arterial dis-ease, hypertension, NYHA class, durationof HF, HF medications, serum creatinine,hemoglobin, DM type, DMmedication, andduration of DM, participants with any(one or more) microvascular complicationswere more likely to have HFpEF (oddsratio [OR] 1.70 [95% CI 1.15–2.50]; P =0.008). Following further correction forsystolic blood pressure at baseline, par-ticipants with microvascular complica-tions were more likely to have HFpEF(1.80 [95% CI 1.22–2.65]; P = 0.003).Furthermore, there was a linear increasein the OR of having HFpEF for participantswith an increasing number of microvas-cular complications (Fig. 1B) (Ptrend ,0.001), which remained statistically sig-nificant in the fully corrected model (P ,0.01 for all). Among participants withmicrovascular complications, diabeticnephropathy was the most prevalent(71%) followed by retinopathy (42%)and neuropathy (28%). Each of neurop-athy, retinopathy, and nephropathywere all more prevalent among partic-ipants with HFpEF compared with partic-ipants with HFrEF in both univariable andthe fully corrected models (Fig. 1A). Par-ticipants with microvascular complica-tions had higher blood pressure andworse signs and symptoms and weremore often in NYHA class III/IV comparedwith participants with DM without mi-crovascular complications and weremore often on insulin and had a longerduration of DM (Table 1).

EchocardiographyHF type was a significant effect modifierfor the association ofmicrovascular com-plications with LV mass (Pinteraction ,0.001). In participants with DM andHFrEF, agreaternumberofmicrovascularcomplications was associated with a de-crease in LV hypertrophy (Fig. 1C andSupplementary Table 1). In contrast,among participants with DM and HFpEF,those with microvascular complicationshad a higher LV mass indexed to BSA(Supplementary Table 1). Furthermore,the prevalence of LV hypertrophy in-creased with an increasing number ofmicrovascular complications in HFpEF,with 41% having LV hypertrophy amongparticipants with HFpEF, DM, and nomicrovascular complications and 76%having LV hypertrophy among partici-pants with DM, HFpEF, and three micro-vascular complications (Ptrend , 0.001)

(Fig. 1C). In addition, the presence ofDM microvascular complications wasassociated with higher LV filling pres-sures (E/e9 $14: OR 1.71 [95% CI 1.14–2.56]; P = 0.009) regardless of HF type(Pinteraction. 0.1). In addition, microvas-cular complications were associatedwith more diastolic dysfunction (OR 3.83[95% CI 1.1–12.9]; P = 0.029).

Clinical OutcomesParticipants with and without follow-updata available were not different in termsof age, sex, comorbidities, or medicationuse, but those lost to follow-up hadhigher systolic and diastolic blood pres-sures and weremore often in NYHA classIII/IV (Supplementary Table 2). Of 2,567participants with DM and follow-up dataavailable, 609 (23.7%) participants diedor were hospitalized for HF within 1 year.In multivariable analyses comparedwith participants with DM without micro-vascular complications, participants withDM microvascular complications hadhigher rates of the primary combinedoutcome (hazard ratio [HR] 1.34 [95%CI 1.06–1.69]; P = 0.015) after correctionfor age, sex, NYHA class, HF subtype(HFrEF/HFpEF), duration of HF, historyof CAD, hypertension, atrial fibrillation,ethnicity, BMI, usage of b-blockers,mineralocorticoid receptor antagonist,ACEi/ARB, insulin, oral antidiabetes med-ications, creatinine, hemoglobin, and typeof DM (model 3 in Table 2). After correctingfor systolic blood pressure at baseline,microvascular complications remained in-dependently associated with the primarycombined outcome (HR 1.37 [95% CI 1.09–1.73]; P = 0.008) (model not shown in Table2). After further correction for duration ofdiabetes (model4 inTable2),microvascularcomplications were still independently as-sociated with the primary combined out-come (HR 1.35 [95% CI 1.04–1.76]; P 50.024). Microvascular complications weresimilarly associated with the primary com-bined outcome across HF subtypes (HFrEF/HFpEF), as no significant interaction wasfound throughout any of the multivariablemodels between microvascular complica-tions and HF subtype (Pinteraction for allmodels .0.1). Diabetic nephropathyshowed the strongest association with ad-verse outcomes (Supplementary Table 3).However, no significant interaction wasobservedbetweenmicrovascular complica-tion subtype and HF type for the compos-ite outcome (Pinteraction for all .0.2).

1794 Microvascular Complications and Heart Failure Diabetes Care Volume 42, September 2019

Quality of LifeParticipants with DMmicrovascular compli-cations hadworse quality of life as reflectedby lower KCCQ scores across various do-mains, with social limitation, total symp-toms, and physical limitation being mostseverely affected (Supplementary Table4). Compared with participants withoutmicrovascular complications, those withHFpEF and diabetic microvascular com-plications reported a larger reduction inoverall KCCQ scores than participantswith HFrEF and microvascular complica-tions (Pinteraction , 0.001) (Fig. 2D).

CONCLUSIONS

In this first comparison of diabetic micro-vascular complications between HFpEFand HFrEF from the same study popula-tion, we found that DM microvascularcomplications were more prevalent inparticipants with DM and HFpEF (27%)than in participants with DM and HFrEF

(20%), a finding also true for each in-dividual microvascular complication.Diabetic nephropathy was the most com-mon manifestation of microvascularinvolvement (present in 71% of partic-ipants with at least one microvascularcomplication), followed by retinopathy(42%) and neuropathy (28%). Diabeticmicrovascular disease was related togreater LV hypertrophy and worse qualityof life in participants with DM and HFpEFthan HFrEF. Participants with microvas-cular disease had higher LV filling pres-sures and more adverse clinical outcomesregardless of the HF type. These findingssuggest that HFpEF may be a clinical man-ifestation of microvascular disease inpatients with DM.

The prevalence of microvascular com-plications among participants with DMin ASIAN-HF (21%) was lower than thatobserved in participants with DM fromTOPCAT (32%) and BEST (32%) (4,5). This

lower prevalence may be explained bya shorter duration of DM in ASIAN-HF,where participants with DM were almosta decade younger than their counterpartsfromBEST and TOPCAT (4–6,9). In addition,the percentage of participants with DMon insulins was less than half in ASIAN-HF (16%) compared with BEST (41%)and TOPCAT (38%), suggesting a lesserseverity of DM. Furthermore, diabetic ne-phropathy was the most common micro-vascular complication compared withdiabetic neuropathy in TOPCAT and BEST(4,5). Ethnicity has an important influenceon the prevalence and distribution oftypes of microvascular complications(17–23). Asians with DM are at higherrisk for developing diabetic nephropa-thy compared with Europeans (20,21).A separate study found that Indo-Asianimmigrants with DM in the Netherlandshad a 40-fold higher risk of end-stagekidney disease and that overall renal

Figure 1—OR adjusted for age, sex, history of CAD, previous stroke, peripheral artery disease, hypertension, atrial fibrillation, ethnicity, NYHA class, durationof HF, BMI, usage of b-blockers, ACEi/ARB, insulin, oral antidiabetes medications, creatinine, hemoglobin. A and B: Forest plot with ORs and 95% CI for havingHFpEF stratified according to microvascular complication (A) and number of microvascular complications (B). C: Percentage of participants with cardiachypertrophy stratified according to number of microvascular complications and HF subtype.

care.diabetesjournals.org Tromp and Associates 1795

Table 1—Baseline characteristics according to complication status

DM without microvascular complications(N = 2,199)

DM with microvascular complications(N = 601) P value

Age (years) 63.0 (11.4) 64.3 (10.8) 0.012

Women 586 (26.6) 175 (29.1) 0.23

Ethnicity ,0.001Chinese 767 (34.9) 180 (30.0)Indian 673 (30.6) 182 (30.3)Malaysian 415 (18.9) 119 (19.8)Japanese 145 (6.6) 73 (12.1)Korean 103 (4.7) 10 (1.7)Thai 43 (2.0) 27 (4.5)Filipino 12 (0.5) 5 (0.8)Indigenous SEA 34 (1.5) 4 (0.7)Others 6 (0.3) 1 (0.2)

NYHA class ,0.001I 289 (14.9) 50 (9.0)II 1,035 (53.4) 286 (51.6)III 523 (27.0) 195 (35.2)IV 93 (4.8) 23 (4.2)

Systolic BP (mmHg) 122.6 (20.9) 127.1 (22.6) ,0.001

Diastolic BP (mmHg) 72.4 (12.4) 71.1 (12.0) 0.019

BMI (kg/m2) 26.0 (5.1) 26.3 (5.7) 0.16

Heart rate (bpm) 80.1 (16.1) 78.1 (13.9) ,0.001

eGFR (mL/min/1.73 m2) 63.6 (27.4) 46.7 (25.5) ,0.001

LVEF (%) 33.8 (14.9) 35.9 (15.5) 0.002

HFpEF (%) 411 (18.7) 150 (24.9) 0.001

Ischemic etiology 1,202 (58.5) 372 (64.4) 0.011

Duration of HF, years 0.02,1 988 (47.1) 229 (40.2)1–5 621 (29.3) 196 (34.4).5–10 317 (14.9) 97 (17.0)$10 185 (8.7) 48 (8.4)

Signs and symptomsShortness of breath on exertion 1,531 (69.7) 470 (78.2) ,0.001Angina 266 (12.1) 56 (9.3) 0.058Elevated JVP 359 (16.3) 132 (22.0) 0.001Peripheral edema 636 (28.9) 224 (37.3) ,0.001Pulmonary rales 391 (17.8) 165 (27.5) ,0.001

Medical historyObese 380 (18.7) 115 (20.6) 0.329CAD 1,263 (57.5) 383 (63.7) 0.006CKD (eGFR ,60 mL/min/1.73 m2) 877 (48.7) 375 (75.0) ,0.001Prior stroke 166 (7.6) 67 (11.1) 0.005Atrial fibrillation/flutter 399 (18.2) 104 (17.3) 0.63Hypertension 1,532 (69.7) 448 (74.5) 0.022Type of DM 0.214Type 1 60 (2.7) 11 (1.8)Type 2 2,139 (97.3) 590 (98.2)

Duration of DM (years), median (IQR) 7 (3, 12) 11 (6, 20) ,0.001Peripheral arterial disease 73 (3.3) 63 (10.5) ,0.001

Medication and device useACEi or ARB 1,560 (73.2) 392 (66.4) 0.001b-Blockers 1,616 (75.8) 457 (77.5) 0.41MRA 1,078 (50.6) 206 (34.9) ,0.001Any diuretics 1,771 (83.1) 489 (82.9) 0.9Any oral antidiabetes medication 1,291 (60.6) 320 (54.6) 0.008Insulin 335 (15.7) 139 (23.7) ,0.001Device use 0.372None 1,932 (88.0) 539 (89.7)ICD/CRT-D 177 (8.0) 45 (7.5)Pacer/pacemaker 87 (4.0) 17 (2.8)

Continued on p. 1797

1796 Microvascular Complications and Heart Failure Diabetes Care Volume 42, September 2019

disease progression is faster than in theindigenous Dutch population (18,19).Additionally, diabetic neuropathy is con-siderably less common in Asians, whichcould be explained by a smaller stature(22,23). Ethnic differences might alsopotentially influence the associationbetween microvascular complicationsand HFpEF. However, recent resultsfrom the PROMIS-HFpEF (PRevalence OfMIcrovascular dySfunction in Heart Fail-ure with Preserved Ejection Fraction)study showed only marginal ethnic dif-ferences in the prevalence of microvas-cular disease in HFpEF (3).In a study using U.S. Medicare claims,

patients with DM and diabetic nephrop-athy had a higher risk of developing HF(24). A study from the U.K. Clinical Prac-tice Research Datalink showed that DMmicrovascular complications were asso-ciated with a higher risk of developing HF(25). This is further supported by resultsfrom the Atherosclerosis Risk In Com-munities (ARIC) study, where diabeticretinopathy was associated with an ex-cess risk of developing HF (26). Thecurrent study extends on these previousfindings and suggests that microvascular

disease in DM may portend an excessrisk for developing HFpEF specifically(rather than HFrEF). Indeed, data fromthe Multiethnic Study of Atherosclero-sis (MESA) also showed that micro-vascular complications were associatedwith more concentric hypertrophy onechocardiography, a hallmark of HFpEF(27,28). Finally, our findings are alsoconsistent with recently reported resultsof the PROMIS-HFpEF study (3)da largeprospective multicenter study showing ahigh prevalence of coronary microvascu-lar dysfunction in HFpEF in the absenceof unrevascularized macrovascular CAD.The PROMIS-HFpEF study also showedthat microvascular dysfunction in HFpEFwas likely to be systemic, with evidenceof reduced peripheral reactive hyperemiaindex and increased urinary albumin-to-creatinine ratio. Collectively, these datasupport the notion that microvasculardysfunction may be a promising com-posite risk marker and therapeutic tar-get in HFpEF whether in the presence orabsence of DM.

Microvascular disease was similarlyassociated with adverse outcomes inboth participants with HFrEF and HFpEF,

in keeping with prior observations in BESTand TOPCAT (4,5). However, this is the firstreport directly comparing the predictivevalue of microvascular disease in partic-ipants with HFrEF and HFpEF enrolledsimultaneously in the same cohort. Inter-estingly,microvasculardiseasehadastron-ger association with HF hospitalizationsthan all-cause mortality. This could beexplained by the lower statistical powerin the mortality analysis. This is plausiblegiven the significant association betweenmicrovascular disease and all-cause mor-tality alone in participantswithDMandHFas previously reported (4,5). Another po-tential explanation is the relatively largecontribution of diabetic nephropathyto microvascular events in the ASIAN-HFstudy. Renal dysfunction in particular isassociated with sodium and water reten-tion, which leads to worsening of HF andsubsequent hospitalization.

The clinical implications of this studyare twofold. First of all, results of thisstudy suggest that HFpEF might be aclinical manifestation of microvascu-lar disease in patients with DM. Thus,in the routine screening for microvascu-lar complications in patients with DM,

Table 1—Continued

DM without microvascular complications(N = 2,199)

DM with microvascular complications(N = 601) P value

Laboratory, median (IQR)Potassium (mmol/L) 4.2 (3.8, 4.6) 4.3 (3.9, 4.6) 0.008Sodium (mmol/L) 138.0 (136.0, 140.0) 138.0 (136.0, 141.0) 0.88Creatinine (mg/dL) 1.1 (0.9, 1.5) 1.5 (1.2, 2.2) ,0.001

Data are n (%) or median (interquartile range [IQR]). BP, blood pressure; CKD, chronic kidney disease; CRT-D, cardiac resynchronization therapydefibrillator; HR, heart rate; ICD, implantable cardioverter defibrillator; JVP, jugular venous pressure;MRA,mineralocorticoid receptor antagonist; SEA,Southeast Asia.

Table 2—Crude and adjusted associations of microvascular complications with 1-year outcomes

Composite outcome All-cause mortality HF hospitalizations

No. at riskNo. of

events (%) No. at riskNo. of

events (%) No. at riskNo. of

events (%)

DM without microvascularcomplications (referent for HR) 2,024 437 (21.6) 2,024 209 (10.3) 2,024 271 (13.4)

DM with microvascularcomplications 543 172 (31.7) 543 76 (14.0) 543 114 (21.0)

HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value

Univariable 1.55 (1.32–1.94) ,0.001 1.39 (1.07–1.81) 0.014 1.63 (1.31–2.02) ,0.001

Model 1 1.60 (1.34–1.92) ,0.001 1.42 (1.08–1.88) 0.012 1.59 (1.24–2.04) ,0.001

Model 2 1.52 (1.25–1.86) ,0.001 1.43 (1.07–1.91) 0.015 1.46 (1.13–1.90) 0.004

Model 3 1.34 (1.06–1.69) 0.015 1.12 (0.78–1.60) 0.536 1.43 (1.06–1.93) 0.019

Model 4 1.35 (1.04–1.76) 0.024 1.16 (0.78–1.73) 0.456 1.49 (1.06–2.09) 0.020

Model 1: age, sex, ethnicity, NYHA class, HF subtype (HFrEF/HFpEF), duration of HF.Model 2: model 1 + BMI, history of hypertension, atrial fibrillation,CAD, usage of insulins, and oral DM drugs. Model 3: model 2 + usage of ACEi/ARB, b-blockers, mineralocorticoid receptor antagonist, creatininelevels, hemoglobin levels, type of DM. Model 4: model 3 + duration of DM.

care.diabetesjournals.org Tromp and Associates 1797

physicians may also consider evaluat-ing the heart for structural/functionalchanges and not only the eyes, kidneys,and peripheral nerves. Conversely,given that microvascular complicationsare associated with greater LV hyper-trophy and reduction in quality of lifeamong patients with HFpEF, the optimalmanagement of patients with HFpEFmay also include screening for micro-vascular complications in other organsystems and optimizing antidiabetesmedication to prevent microvascularcomplications.

LimitationsThe presence of DM and microvascularcomplications was determined by historyand review of medical records, withoutthe use of glycated hemoglobin levels,disease-specific questionnaires, or spe-cialist testing. Despite rigorous attemptsto identify all participants with DM andmicrovascular complications, some ofthese participants might have been

missed. However, this is more likely tounderestimate the impact of DM andmicrovascular complications on outcomesrather than introduce significant bias. Wewere also unable to account for DMseverity due to the lack of glycated he-moglobin levels. Several uncapturedfactors including quality of health careservices, fitness level, and quality of self-care (adherence to medication, water anddietary restriction, and others) mighthave led to residual confounding notaccounted for in our analyses. Urinaryalbumin-to-creatinine ratio (UACR) wouldhave provided additional insights intoconcurrent microvascular disease in thekidneys; however, this was not avail-able in ASIAN-HF. Nonetheless, UACR hasrecently been shown to directly correlatewith coronarymicrovascular disease (mea-sured as coronary flow reserve) in HFpEF(3). Furthermore, we did not have longi-tudinal assessments of cardiovascular riskfactor control (e.g., lipids, blood pressure,glucose). Finally, ASIAN-HF only included

participants of Asian descent; generaliz-ability to other regions and ethnicities isunclear.

ConclusionDM microvascular complications aremore common and are associated withgreater LV remodeling and worse qualityof life in patients from Asia with HFpEFcompared with those with HFrEF. Mi-crovascular disease portends worse clin-ical outcomes regardless of HF subtype.Our findings suggest that HFpEF may be aclinical manifestation of microvasculardisease in DM. Further investigation isneeded to validate our findings in amultiethnic study and to determine op-timal management strategies in patientswith HFpEF and DM.

Acknowledgments. The contributions of allsite investigators and clinical coordinators areduly acknowledged.Funding. The ASIAN-HF study is partly sup-ported by grants from the Boston Scientific

Figure 2—A–C: Kaplan-Meier curves according to microvascular status for the combined outcome of all-cause mortality or HF hospitalization for thetotal population (A), HFrEF only (B), and HFpEF only (C). D: Box plot of quality of life status (KCCQ overall scores) stratified according to HF type andmicrovascular complication (MC) status.

1798 Microvascular Complications and Heart Failure Diabetes Care Volume 42, September 2019

Investigator Sponsored Research Program, NationalMedical Research Council of Singapore (NMRC/CSA/0052/2013), and A*STAR Biomedical Re-searchCouncil ATTRaCTprogram(SPF2014/003).Duality of Interest. The ASIAN-HF study is alsosupported by Bayer. No other potential conflictsof interest relevant to this article were reported.Author Contributions. J.T., S.L.L., and W.T.T.performed the analyses and wrote the manu-script. T.-H.K.T., C.C., W.O., G.S.W., J.P.S.S., J.Y.,M.R.M., L.H.L., N.S., J.J.V.M., A.M.R., and I.A.contributed to discussion and reviewed andedited the manuscript. C.S.P.L. provided criticalrevisions of the manuscript and interpreted thedata. J.T. and C.S.P.L. are the guarantors of thiswork and, as such, had full access to all the data inthe study and take responsibility for the integrityof the data and the accuracy of the data analysis.

References1. Fowler MJ. Microvascular and macrovascularcomplicationsofdiabetes.ClinDiabetes2008;26:77–822. Paulus WJ, Dal Canto E. Distinct myocardialtargets for diabetes therapy in heart failure withpreserved or reduced ejection fraction. JACCHeart Fail 2018;6:1–73. ShahSJ, LamCSP, SvedlundS, et al. Prevalenceand correlates of coronary microvascular dys-function in heart failure with preserved ejectionfraction: PROMIS-HFpEF. Eur Heart J 2018;39:3439–34504. Kristensen SL, Rørth R, Jhund PS, et al.; BESTInvestigators. Microvascular complications indiabetes patients with heart failure and reducedejection fraction-insights from the Beta-blockerEvaluation of Survival Trial. Eur J Heart Fail 2018;20:1549–15565. Sandesara PB, O’Neal WT, Kelli HM, et al. Theprognostic significance of diabetes and micro-vascular complications in patients with heartfailurewith preservedejection fraction. DiabetesCare 2018;41:150–1556. Lam CSP, Teng TK, Tay WT, et al. Regional andethnic differences among patientswith heart failurein Asia: the Asian sudden cardiac death in heartfailure registry. Eur Heart J 2016;37:3141–31537. LamCSP, Anand I, Zhang S, et al. Asian SuddenCardiac Death in Heart Failure (ASIAN-HF) reg-istry. Eur J Heart Fail 2013;15:928–936

8. Tromp J, Tay WT, Ouwerkerk W, et al.;ASIAN-HF Authors. Multimorbidity in patientswith heart failure from 11 Asian regions: a pro-spective cohort study using the ASIAN-HF registry[published correction appears in PLoS Med 2018;15:e1002583]. PLoS Med 2018;15:e10025419. Tromp J, Teng T-H, TayWT, et al. Heart failurewith preserved ejection fraction in Asia. Eur JHeart Fail 2019;21:23–3610. Green CP, Porter CB, Bresnahan DR, SpertusJA. Development and evaluation of the KansasCity Cardiomyopathy Questionnaire: a new healthstatus measure for heart failure. J Am Coll Cardiol2000;35:1245–125511. Joseph SM, Novak E, Arnold SV, et al. Com-parable performance of the Kansas City Cardio-myopathy Questionnaire in patients with heartfailure with preserved and reduced ejectionfraction. Circ Heart Fail 2013;6:1139–114612. Patidar AB, Andrews GR, Seth S. Prevalenceof obstructive sleep apnea, associated risk fac-tors, and quality of life among Indian congestiveheart failure patients: a cross-sectional survey. JCardiovasc Nurs 2011;26:452–45913. LuoN, Teng TK, TayWT, et al.; ASIAN-HF; HF-ACTION investigators. Multinational and multi-ethnic variations in health-related quality of lifein patients with chronic heart failure. Am Heart J2017;191:75–8114. Chia YMF, Teng TK, Tan ESJ, et al. Disparitybetween indications for and utilization of im-plantable cardioverter defibrillators in Asianpatients with heart failure. Circ CardiovascQual Outcomes 2017;10:e00365115. Lang RM, Badano LP, Mor-Avi V, et al.Recommendations for cardiac chamber quanti-fication by echocardiography in adults: an updatefrom the American Society of Echocardiographyand the European Association of CardiovascularImaging. Eur Heart J Cardiovasc Imaging 2015;16:233–27016. Huttin O, Fraser AG, Coiro S, et al. Impact ofchanges in consensus diagnostic recommenda-tions on the echocardiographic prevalence ofdiastolic dysfunction. J Am Coll Cardiol 2017;69:3119–312117. Low SKM, Sum CF, Yeoh LY, et al. Prevalenceof chronic kidney disease in adults with type 2diabetes mellitus. Ann Acad Med Singapore2015;44:164–171

18. Chandie Shaw PK, Vandenbroucke JP,Tjandra YI, et al. Increased end-stage diabeticnephropathy in Indo-Asian immigrants living inthe Netherlands. Diabetologia 2002;45:337–34119. Chandie Shaw PK, Baboe F, van Es LA, et al.South-Asian type 2 diabetic patients have higherincidence and faster progression of renal diseasecompared with Dutch-European diabetic patients.Diabetes Care 2006;29:1383–138520. Gupta R, Misra A. Epidemiology of micro-vascular complications of diabetes in SouthAsians and comparison with other ethnicities.J Diabetes 2016;8:470–48221. Pan CY, Ho LT, Soegondo S, et al.; DEMANDStudy Investigators. Prevalence of albuminuriaand cardiovascular risk profile in a referred co-hort of patients with type 2 diabetes: an Asianperspective. Diabetes Technol Ther 2008;10:397–40322. Winkley K, Thomas SM, Sivaprasad S, et al.The clinical characteristics at diagnosis of type 2diabetes in a multi-ethnic population: the SouthLondon Diabetes cohort (SOUL-D). Diabetologia2013;56:1272–128123. Abbott CA, Chaturvedi N, Malik RA, et al.Explanations for the lower rates of diabeticneuropathy in Indian Asians versus Europeans.Diabetes Care 2010;33:1325–133024. Bertoni AG, Hundley WG, Massing MW,Bonds DE, Burke GL, Goff DC Jr. Heart failureprevalence, incidence, and mortality in the el-derly with diabetes. Diabetes Care 2004;27:699–70325. Brownrigg JRW,HughesCO, BurleighD, et al.Microvascular disease and risk of cardiovascularevents among individuals with type 2 diabetes:a population-level cohort study. Lancet DiabetesEndocrinol 2016;4:588–59726. Cheung N, Wang JJ, Rogers SL, et al.; ARIC(Atherosclerosis Risk In Communities) Study In-vestigators. Diabetic retinopathy and risk ofheart failure. J Am Coll Cardiol 2008;51:1573–157827. CheungN, BluemkeDA, Klein R, et al. Retinalarteriolar narrowing and left ventricular remod-eling: the multi-ethnic study of atherosclerosis. JAm Coll Cardiol 2007;50:48–5528. Shah AM. Ventricular remodeling in heartfailure with preserved ejection fraction. CurrHeart Fail Rep 2013;10:341–349

care.diabetesjournals.org Tromp and Associates 1799