OBJECTIVE

Insulin resistance (IR) may mediate heart failure (HF) development. We examined whether IR in people with newly diagnosed type 2 diabetes (T2D) increased their risk of a composite outcome of HF or death or of HF alone.

RESEARCH DESIGN AND METHODS

Insulin resistance (HOMA2-IR) values for UKPDS participants were derived from paired fasting plasma glucose (FPG) and insulin measures. Kaplan-Meier survival curves and multivariable survival models were used to evaluate associations between HOMA2-IR and HF/death or HF alone. We adjusted for potential confounders by including variables with univariate associations (P < 0.1) and by requiring a multivariable P < 0.05.

RESULTS

Of 5,102 UKPDS participants with newly diagnosed T2D, 4,344 had HOMA2-IR measurements. At enrollment, mean (SD) age was 52.5 (8.7) years, with HbA1c 7.2% (1.8%), and BMI 28.8 (5.5) kg/m2, and median (interquartile range) HOMA2-IR was 1.6 (1.1–2.2). HF/death occurred in 1,974 (45.4%) participants (235 first HF events, 1,739 deaths) over a median follow-up of 16.4 years. Multivariable independent associations with HF/death were older age and higher BMI, HOMA2-IR, FPG, waist-to-hip ratio, systolic blood pressure, LDL cholesterol, and heart rate as well as sex, race, smoking status, prior atrial fibrillation, and prior microalbuminuria. A doubling of HOMA2-IR was associated with a 5% greater risk of HF/death (relative risk [RR] 1.05 [95% CI 1.01–1.12], P = 0.0029) and a 14% greater risk of HF (RR 1.14, [95% CI 1.02–1.27], P = 0.017).

CONCLUSIONS

Patients with newly diagnosed T2D and insulin resistance were more likely to develop HF or die than those more sensitive to insulin.

Several studies have shown a relationship between measures of higher insulin resistance (IR) and the development of heart failure (HF) in people with diabetes (13) and obesity (4). Both conditions are associated with IR, which is correlated with adverse cardiac remodeling (cardiac fibrosis, myocardial hypertrophy, and steatosis of the myocardium) (5,6). Additionally, severe IR is a feature of lipodystrophy, a rare genetic disorder characterized by redistribution of body fat, and associated with a threefold increase in myocardial triglyceride content and cardiomyopathy (7). The link between IR and HF is also supported by the observation that IR is associated with HF in patients without diabetes (2). In addition, the Strong Heart Study in Native Americans demonstrated that abnormalities of left ventricle (LV) size and systolic function correlated with fasting plasma insulin values and predated the development of diabetes (8).

In this post hoc observational analysis of the UK Prospective Diabetes Study (UKPDS) (9) and its 10-year posttrial monitoring study (10), we sought to examine possible associations between IR and the incidence of adjudicated HF or death, or HF alone, in patients with newly diagnosed type 2 diabetes (T2D) and whether this was independent of a relationship with BMI. We also evaluated the impact on HF/death and on HF alone of the glucose-lowering therapies allocated at random in the UKPDS.

Study Design and Participants

The UKPDS protocol, design, methods, and impact on diabetes complications of a more intensive glycemic strategy, compared with a conventional glycemic strategy, have been reported in detail (9,11). The UKPDS recruited individuals with newly diagnosed T2D aged 25–65 years between 1977 and 1991. Those with fasting plasma glucose (FPG) concentrations >6 and <15 mmol/L following a 3–4-month dietary run-in period were assigned randomly to a conventional glycemic control strategy (primarily with diet) or to an intensive glycemic control strategy, primarily sulfonylurea, insulin, or metformin (only for those >120% of ideal body weight) monotherapy, and then followed quarterly in UKPDS clinics.

When the interventional trial closed out on 30 September 1997, all surviving participants were entered into a 10-year posttrial monitoring study (10), with no attempt made to maintain them on their previously randomized therapies. Participants were returned to usual care but seen annually for the first 5 years in UKPDS clinics to facilitate continued standardized collection of outcome data, as well as measurements of FPG, glycated hemoglobin (HbA1c), blood pressure, plasma creatinine, and urinary albumin. For the second 5 years, participants were followed remotely via questionnaires. Members of the UKPDS end point committee who were unaware of assignments to study groups adjudicated all predefined clinical outcomes during the trial, including HF and death, as detailed previously (9). Posttrial outcomes were adjudicated in exactly the same way (10).

Microalbuminuria was defined in the UKPDS as a urinary albumin concentration >50 mg/L due to initial storage of urine samples at –20°C between 1979 and 1988 (11). Hypertension at baseline was diagnosed if the mean of the 2- and 9-month blood pressure measurements exceeded 160 mmHg systolic (SBP) or 90 mmHg diastolic or if the participant was already taking antihypertensive therapy (12).

The UKPDS was performed according to the Declaration of Helsinki, and all participants gave written informed consent to participate. The study protocol was approved by the ethics committees from all 23 UKPDS clinical centers.

Estimating IR

IR at baseline for each participant was estimated using the HOMA2 Calculator (13). HOMA uses a structural mathematical model that incorporates the physiological glucose/insulin feedback system, including pancreatic β-cell function and peripheral (muscle) plus hepatic insulin sensitivity, to estimate an individual’s IR (HOMA2-IR) and β-cell function (HOMA2-%B) (14). The HOMA model compares favorably with other models used to estimate IR and has the advantage of requiring only a single fasting plasma sample to be assayed for insulin and glucose (15,16). We compared UKPDS participants with higher, versus lower, HOMA2-IR values, split by the median value.

Clinical Outcomes

We defined the primary outcome as a composite of incident HF (ICD-9 codes 411–428.1) or death, with death included because it is a competing risk. Secondary outcomes were incident HF (first instance) and the impact of the randomly allocated glucose-lowering therapies used in the UKPDS on HF/death or HF alone.

Statistical Analysis

Continuous baseline variables were summarized as mean (SD) or median (interquartile range [IQR]). Categorical variables were summarized as counts and percentages. Comparisons between those with high and low baseline HOMA2-IR values were performed with Student t test, Wilcoxon signed rank test, or χ2 tests as appropriate. We used Kaplan-Meier estimates and, as the assumptions required for proportional hazards models were not met, multivariable accelerated failure time models assuming a Weibull distribution to evaluate possible associations between HOMA2-IR and HF/death or HF alone.

For analysis of the primary HF/death outcome, participants were censored if alive without HF at the end of follow-up or when lost to follow-up. For the secondary analysis of HF alone, participants without HF were censored at the end of follow-up, when lost to follow-up, or when they died. We adjusted for potential confounders by including variables with univariate associations (P < 0.1) and by requiring a multivariable P value <0.05 for retention in the full model. Potential confounders included age, BMI, SBP, heart rate, fasting plasma triglycerides, total cholesterol, LDL cholesterol, HDL cholesterol, plasma creatinine, hemoglobin, and HbA1c as continuous variables and ethnicity, sex, smoking status, prior cardiovascular events, atrial fibrillation (AF), and microalbuminuria as categorical variables.

To examine the relationship of obesity with any associations identified between HOMA2-IR and HF/death or HF, we performed a sensitivity analysis, with conventional BMI cutoffs as defined by the World Health Organization (17) (normal weight ≥18.5–25, overweight ≥25–30, obese ≥30 kg/m2) and waist-to-hip ratio (WHR) split by tertiles.

To evaluate the impact of the randomly allocated diet glycemic control strategy and sulfonylurea, insulin, and metformin glucose-lowering monotherapies used in the UKPDS on HF/death or HF alone, we performed a sensitivity analysis examining the Kaplan-Meier estimates for each therapy (18).

All analyses were conducted using SAS 9.4 statistical software (SAS Institute, Cary, NC). Univariate analyses were performed in complete case data sets. As 20% of participants had missing data for at least one candidate variable, missing data were assigned mean or mode values for the accelerated failure time models with results compared with complete case models. For multivariate statistical comparisons we used two-sided tests at the 0.05 level of significance, with no adjustment for multiple testing.

Participant Characteristics

Of the 5,102 participants enrolled into the UKPDS, 4,344 had the simultaneous fasting plasma insulin and FPG measures required for estimation of HOMA2-IR values and were included in these analyses. Of subjects at baseline, 59% were male and 81% were White Caucasian, with mean (SD) age 52.5 (8.7) years, HbA1c 7.2% (1.8%), BMI 28.8 (5.5) kg/m2, and WHR 0.91 (0.08) and median (IQR) HOMA2-IR 1.6 (1.1–2.2) (Table 1). History of HF at baseline was assessed by study physicians. There were 4,317 (99.4%) participants with no history of HF, 5 (0.1%) thought to have probable/definite HF, 22 (0.5%) for whom this information was not recorded, and 45 (1.0%) who were on diuretic therapy. The diuretic therapy and a history of HF categories were not mutually exclusive.

Table 1

Baseline characteristics of UKPDS participants analyzed

AllHOMA2-IR <1.6HOMA2-IR ≥1.6P
n 4,344 2,099 2,245  
Age (years) 52.5 (8.7) 53.2 (8.5) 51.9 (8.8) <0.0001 
Sex, n (%)    <0.0001 
 Male 2,578 (59) 1,388 (66) 1,190 (53)  
 Female 1,766 (41) 711 (34) 1,055 (47)  
Race, n (%)    <0.0001 
 White Caucasian 3,537 (81.4) 1,737 (82.8) 1,800 (80.2)  
 Afro-Caribbean 343 (7.9) 181 (8.6) 162 (7.2)  
 Asian-Indian 431 (9.9) 162 (7.7) 269 (12.0)  
 Other 33 (0.8) 19 (0.9) 14 (0.6)  
HbA1c (%) 7.2 (1.8) 7.0 (1.7) 7.3 (1.8) <0.0001 
FPG (mmol/L) 8.7 (2.9) 8.3 (2.7) 9.2 (3.1) <0.0001 
Fasting plasma insulin (mU/L) 14.1 (8.1) 8.6 (3.1) 19.3 (7.9) <0.0001 
HOMA2-IR 1.6 (1.1–2.2) 1.1 (0.8–1.3) 2.2 (1.8–2.8) — 
WHR 0.91 (0.08) 0.90 (0.07) 0.92 (0.08) <0.0001 
BMI (kg/m228.8 (5.5) 26.8 (4.2) 30.6 (5.9) <0.0001 
SBP (mmHg) 135.3 (19.4) 133.6 (19.4) 136.9 (19.2) <0.0001 
Heart rate (bpm) 71 (13.3) 70 (13.0) 72 (13.6) <0.0001 
AF, n (%) 30 (0.7) 12 (0.6) 18 (0.8) 0.36 
Fasting triglycerides (mmol/L) 1.5 (1.1–2.1) 1.3 (1.0–1.7) 1.7 (1.3–2.4) <0.0001 
Total cholesterol (mmol/L) 5.4 (1.1) 5.3 (1.1) 5.5 (1.1) <0.0001 
HDL cholesterol (mmol/L) 1.07 (0.24) 1.10 (0.25) 1.04 (0.23) <0.0001 
LDL cholesterol (mmol/L) 3.5 (1.0) 3.5 (1.0) 3.5 (1.0) 0.031 
Smoking status, n (%)    0.16 
 Current 1,339 (30.8) 653 (31.1) 686 (30.6)  
 Ex-smoker 1,463 (33.7) 730 (34.8) 733 (32.7)  
 Never 1,541 (35.5) 716 (34.1) 825 (36.8)  
eGFR (mL/min/1.73 m281 (23) 78 (22) 84 (23) <0.0001 
Urinary albumin (mg/L) 9 (4–20) 7 (4–15) 9 (4–25) <0.0001 
Microalbuminuria, n (%) 462 (11.9) 165 (8.7) 297 (15.0) <0.0001 
Hypertension, n (%) 1,597 (36.8) 702 (33.6) 895 (39.9) <0.0001 
Baseline medication, n (%)     
 Aspirin 834 (19.2) 371 (17.7) 463 (20.6) 0.014 
 Diuretics 389 (9.4) 167 (8.3) 222 (10.5) 0.017 
 Corticosteroids 9 (0.2) 3 (0.1) 6 (0.3) 0.37 
 Digoxin 45 (1.0) 14 (0.7) 31 (1.4) 0.020 
 Antihypertensives 513 (11.8) 242 (11.5) 271 (12.1) 0.57 
 Lipid lowering 11 (0.3) 6 (0.3) 5 (0.2) 0.68 
AllHOMA2-IR <1.6HOMA2-IR ≥1.6P
n 4,344 2,099 2,245  
Age (years) 52.5 (8.7) 53.2 (8.5) 51.9 (8.8) <0.0001 
Sex, n (%)    <0.0001 
 Male 2,578 (59) 1,388 (66) 1,190 (53)  
 Female 1,766 (41) 711 (34) 1,055 (47)  
Race, n (%)    <0.0001 
 White Caucasian 3,537 (81.4) 1,737 (82.8) 1,800 (80.2)  
 Afro-Caribbean 343 (7.9) 181 (8.6) 162 (7.2)  
 Asian-Indian 431 (9.9) 162 (7.7) 269 (12.0)  
 Other 33 (0.8) 19 (0.9) 14 (0.6)  
HbA1c (%) 7.2 (1.8) 7.0 (1.7) 7.3 (1.8) <0.0001 
FPG (mmol/L) 8.7 (2.9) 8.3 (2.7) 9.2 (3.1) <0.0001 
Fasting plasma insulin (mU/L) 14.1 (8.1) 8.6 (3.1) 19.3 (7.9) <0.0001 
HOMA2-IR 1.6 (1.1–2.2) 1.1 (0.8–1.3) 2.2 (1.8–2.8) — 
WHR 0.91 (0.08) 0.90 (0.07) 0.92 (0.08) <0.0001 
BMI (kg/m228.8 (5.5) 26.8 (4.2) 30.6 (5.9) <0.0001 
SBP (mmHg) 135.3 (19.4) 133.6 (19.4) 136.9 (19.2) <0.0001 
Heart rate (bpm) 71 (13.3) 70 (13.0) 72 (13.6) <0.0001 
AF, n (%) 30 (0.7) 12 (0.6) 18 (0.8) 0.36 
Fasting triglycerides (mmol/L) 1.5 (1.1–2.1) 1.3 (1.0–1.7) 1.7 (1.3–2.4) <0.0001 
Total cholesterol (mmol/L) 5.4 (1.1) 5.3 (1.1) 5.5 (1.1) <0.0001 
HDL cholesterol (mmol/L) 1.07 (0.24) 1.10 (0.25) 1.04 (0.23) <0.0001 
LDL cholesterol (mmol/L) 3.5 (1.0) 3.5 (1.0) 3.5 (1.0) 0.031 
Smoking status, n (%)    0.16 
 Current 1,339 (30.8) 653 (31.1) 686 (30.6)  
 Ex-smoker 1,463 (33.7) 730 (34.8) 733 (32.7)  
 Never 1,541 (35.5) 716 (34.1) 825 (36.8)  
eGFR (mL/min/1.73 m281 (23) 78 (22) 84 (23) <0.0001 
Urinary albumin (mg/L) 9 (4–20) 7 (4–15) 9 (4–25) <0.0001 
Microalbuminuria, n (%) 462 (11.9) 165 (8.7) 297 (15.0) <0.0001 
Hypertension, n (%) 1,597 (36.8) 702 (33.6) 895 (39.9) <0.0001 
Baseline medication, n (%)     
 Aspirin 834 (19.2) 371 (17.7) 463 (20.6) 0.014 
 Diuretics 389 (9.4) 167 (8.3) 222 (10.5) 0.017 
 Corticosteroids 9 (0.2) 3 (0.1) 6 (0.3) 0.37 
 Digoxin 45 (1.0) 14 (0.7) 31 (1.4) 0.020 
 Antihypertensives 513 (11.8) 242 (11.5) 271 (12.1) 0.57 
 Lipid lowering 11 (0.3) 6 (0.3) 5 (0.2) 0.68 

Data are mean (SD) or median (IQR) unless otherwise indicated. Up to 10% of data were missing for HbA1c, SBP, heart rate, and microalbuminuria. eGFR, estimated glomerular filtration rate.

At baseline, participants with higher HOMA2-IR values (≥1.6, n = 2,245), compared with those with lower values (<1.6, n = 2,099), were younger (mean 51.9 vs. 53.2 years, P < 0.0001) and more likely to be female (47 vs. 34%, P < 0.0001) and Asian-Indian (12 vs. 7%, P < 0.0001). They had higher mean BMI (30.6 vs. 26.8 kg/m2, P < 0.0001), HbA1c (7.3 vs. 7.0%, P < 0.0001), and urinary albumin values (9 vs. 7 mg/L, P < 0.0001) and median triglycerides (1.7 vs. 1.3 mmol/L, P < 0.0001) and were also more likely to be hypertensive (39.9 vs. 33.6%, P < 0.0001) (Table 1).

Risk Factors for HF/Death and HF Alone

Over a median follow-up period of 16.4 years the primary composite HF/death outcome occurred in 1,974 (45.4%) participants, triggered by 235 HF first events and 1,739 deaths, equating to an incidence rate of 29.6 per 1,000 person-years. Variables found to be associated independently with HF/death in the multivariate analysis were older age; higher BMI, HOMA2-IR, FPG, WHR, SBP, heart rate, and LDL cholesterol; sex, race, and smoking status; and prior history of AF and microalbuminuria. In the secondary HF-alone analysis, independent associations were found for age, BMI, HOMA2-IR, SBP, and prior AF and microalbuminuria. Baseline HbA1c values did not predict the composite HF/death outcome.

Association of HOMA2_IR With HF/Death and HF Alone

During the study there were 145 (6.5%) HF events and 1,095 (48.8%) HF events/deaths among individuals with HOMA2-IR ≥1.6 compared with 90 (4.3%) HF events and 879 (41.9%) HF events/deaths among those with HOMA2-IR <1.6. The unadjusted relative risks (RRs) for HF/death and HF alone for participants with HOMA2-IR ≥1.6 were 1.09 (95% CI 1.04–1.15) and 1.32 (1.10–1.58), respectively. Unadjusted RRs with use of log2 HOMA2-IR as a continuous variable were 1.08 (1.05–1.11) and 1.24 (1.11– 1.39), respectively, suggesting that a twofold increase in HOMA2-IR is associated with corresponding risk increases of 8% for HF/death and 24% for HF alone. The corresponding adjusted RRs in the multivariable analysis were 1.05 (1.01–1.12) for HF/death and 1.14 (1.02–1.27) for HF alone, suggesting that a twofold increase in HOMA2-IR is associated with risk increases of 5% and 14%, respectively (Table 2). Kaplan-Meier estimates confirmed a significant association (log-rank P < 0.0001) between baseline HOMA2-IR and HF/death and HF alone (Fig. 1).

Figure 1

Cumulative incidence plot for time to HF or death (A) and HF alone (B), split by the median HOMA2-IR value, with unadjusted RRs and log-rank P values.

Figure 1

Cumulative incidence plot for time to HF or death (A) and HF alone (B), split by the median HOMA2-IR value, with unadjusted RRs and log-rank P values.

Close modal
Table 2

Multivariable analysis of all risk factors with a univariate P < 0.1 association with the composite outcome of HF or death and with HF alone

VariableHF or death (1,974 events or deaths)HF alone (235 events)
RR95% CIPRR95% CIP
Age (per year) 1.04 (1.03–1.04) <0.0001 1.05 (1.06–1.11) <0.0001 
Female sex (vs. male) 0.86 (0.81–0.91) <0.0001 0.96 (0.79–1.12) 0.52 
Afro-Caribbean (vs. White Caucasian) 0.81 (0.74–0.90) <0.0001 0.84 (0.59–1.20) 0.32 
Asian-Indian (vs. White Caucasian) 0.88 (0.80–0.98) 0.011 0.81 (0.55–1.21) 0.28 
Current smoker (yes vs. no) 1.24 (1.18–1.30) <0.0001 N/A N/A N/A 
BMI (kg/m21.01 (1.01–1.01) 0.0003 1.03 (1.03–1.08) <0.0001 
HOMA2-IR (log21.05 (1.01–1.12) 0.0029 1.14 (1.02–1.27) 0.017 
HbA1c (per 1%) 1.00 (0.98–1.02) 0.78 NA NA NA 
FPG (per 1 mmol/L) 1.02 (1.02–1.03) <0.0001 1.01 (0.99–1.04) 0.36 
WHR 1.74 (1.22–2.58) 0.0039 2.36 (0.81–6.90) 0.12 
SBP (per 10 mmHg) 1.05 (1.04–1.07) <0.0001 1.05 (1.01–1.19) 0.038 
Heart rate (per 5 bpm) 1.02 (1.01–1.03) <0.0001 1.03 (1.00–1.06) 0.081 
LDL cholesterol (mmol/L) 1.03 (1.01–1.06) 0.0096 N/A NA N/A 
HDL cholesterol (mmol/L) NA NA NA 0.75 (0.52–1.08) 0.12 
Fasting triglycerides (log2 mmol/L) 1.03 (0.99–1.07) 0.13 1.03 (0.91–1.16) 0.65 
eGFR (per 10 mL/min/1.73 m21.00 (1.00–1.00) 0.11 NA NA NA 
AF (yes vs. no) 1.84 (1.52–2.25) <0.0001 3.70 (9.48–50.18) <0.0001 
Microalbuminuria (yes vs. no) 1.15 (1.07–1.30) 0.0001 1.44 (1.22–2.79) 0.0027 
VariableHF or death (1,974 events or deaths)HF alone (235 events)
RR95% CIPRR95% CIP
Age (per year) 1.04 (1.03–1.04) <0.0001 1.05 (1.06–1.11) <0.0001 
Female sex (vs. male) 0.86 (0.81–0.91) <0.0001 0.96 (0.79–1.12) 0.52 
Afro-Caribbean (vs. White Caucasian) 0.81 (0.74–0.90) <0.0001 0.84 (0.59–1.20) 0.32 
Asian-Indian (vs. White Caucasian) 0.88 (0.80–0.98) 0.011 0.81 (0.55–1.21) 0.28 
Current smoker (yes vs. no) 1.24 (1.18–1.30) <0.0001 N/A N/A N/A 
BMI (kg/m21.01 (1.01–1.01) 0.0003 1.03 (1.03–1.08) <0.0001 
HOMA2-IR (log21.05 (1.01–1.12) 0.0029 1.14 (1.02–1.27) 0.017 
HbA1c (per 1%) 1.00 (0.98–1.02) 0.78 NA NA NA 
FPG (per 1 mmol/L) 1.02 (1.02–1.03) <0.0001 1.01 (0.99–1.04) 0.36 
WHR 1.74 (1.22–2.58) 0.0039 2.36 (0.81–6.90) 0.12 
SBP (per 10 mmHg) 1.05 (1.04–1.07) <0.0001 1.05 (1.01–1.19) 0.038 
Heart rate (per 5 bpm) 1.02 (1.01–1.03) <0.0001 1.03 (1.00–1.06) 0.081 
LDL cholesterol (mmol/L) 1.03 (1.01–1.06) 0.0096 N/A NA N/A 
HDL cholesterol (mmol/L) NA NA NA 0.75 (0.52–1.08) 0.12 
Fasting triglycerides (log2 mmol/L) 1.03 (0.99–1.07) 0.13 1.03 (0.91–1.16) 0.65 
eGFR (per 10 mL/min/1.73 m21.00 (1.00–1.00) 0.11 NA NA NA 
AF (yes vs. no) 1.84 (1.52–2.25) <0.0001 3.70 (9.48–50.18) <0.0001 
Microalbuminuria (yes vs. no) 1.15 (1.07–1.30) 0.0001 1.44 (1.22–2.79) 0.0027 

eGFR, estimated glomerular filtration rate; NA, not applicable (univariate association ≥0.1).

Relationship With Obesity

We found no heterogeneity for the association of HOMA2-IR with HF/death between normal-weight, overweight, and obese individuals (P for interaction = 0.16), but between thirds of WHR we found possible evidence suggesting a stronger association of HOMA2-IR with the composite outcome of HF or death in those with a lower WHR (P = 0.049) (Fig. 2).

Figure 2

Subgroup analyses of the association of HOMA2-IR with the composite outcome of HF or death, split by BMI categories (kg/m2) and by tertiles of WHR.

Figure 2

Subgroup analyses of the association of HOMA2-IR with the composite outcome of HF or death, split by BMI categories (kg/m2) and by tertiles of WHR.

Close modal

Impact of Randomly Assigned Glucose-Lowering Therapies on HF/Death

The Kaplan-Meier analysis of the probability of HF event/death–free survival in four groups of participants assigned randomly to a conventional glycemic control strategy with diet (reference group) or to an intensive glucose control strategy with sulfonylurea (RR 0.96 [95% CI 0.90–1.02]), insulin (0.94 [0.88–1.01]), or metformin (0.92 [0.83–1.02]) showed no between-group differences over 5 years (log-rank P = 0.24) (Fig. 3).

Figure 3

Cumulative incidence plot for time to the composite outcome of HF or death, split by the four randomly allocated glucose-lowering therapies, and log-rank P value with diet as the reference group.

Figure 3

Cumulative incidence plot for time to the composite outcome of HF or death, split by the four randomly allocated glucose-lowering therapies, and log-rank P value with diet as the reference group.

Close modal

In this large prospective study of patients with newly diagnosed T2D, the incidence of the HF/death composite outcome was higher in those who were more insulin resistant than in those who were less insulin resistant. Specifically, we found that a doubling of HOMA2-IR values was associated with a 5% higher adjusted RR for HF/death and a 14% higher adjusted RR for incident HF. This association was independent of different BMI categories but not thirds of WHR. The somewhat higher risk ratios for incident HF than for HF/death suggest that death may act as a dilutional factor, as might be expected if HF-specific events were driving the observed associations.

We also found that higher HOMA2-IR values were associated independently with HF/death, as were older age and higher BMI, FPG, WHR, SBP, heart rate, and LDL cholesterol, as well as sex, race, smoking status and prior history of AF and microalbuminuria. Higher HOMA2-IR values were also associated independently with incident HF alone, as were older age and higher BMI and SBP, as well as prior AF and microalbuminuria. Our findings confirm that HOMA2-IR, assessed at the time of T2D diagnosis, can help predict the future risk of HF/death, as has been shown for IR and incident HF in people with established T2D (1,19,20).

Our demonstration that the association between HOMA2-IR and HF/death is independent of BMI is consistent with prior reports (1,3). Of note, we show that the association of HOMA2-IR with the composite HF/death outcome was stronger for participants in the lowest WHR category. We cannot, however, exclude that obesity and IR are on the same causal pathway.

In previous UKPDS reports, we have shown a lack of association between baseline fasting hyperinsulinemia and macrovascular disease, and specifically ischemic heart disease (21), and have also demonstrated that measuring insulin sensitivity (estimated using HOMA version 1 as HOMA-%S, the reciprocal of HOMA-IR) at the time of diagnosis of T2D provides no additional value with respect to predicting the future risk of myocardial infarction (22). We did not, however, examine HF outcomes in these previous reports, although others have reported that among participants without antecedent myocardial infarction higher fasting insulin levels are associated with HF risk (2). Accordingly, it may be that promoting insulin sensitization early in the course of T2D could play a role in preventing the development of HF, although we did not see any risk reduction for HF/death in the overweight UKPDS participants randomized to metformin despite a 22% reduction in their median (IQR) fasting plasma insulin, from 15.6 mU/L (9.3–25.9) at baseline to 12.1 mU/L (7.2–20.1) at 6 years (P < 0.01), as reported previously (23).

A causal relationship between IR and the development of HF has not been confirmed thus far in clinical trials with any insulin-sensitizing agent. Thiazolidinediones, despite their well-documented specific effect on peripheral IR, cause fluid retention, and therefore their use in a nonselected population has been associated with an increased risk of HF hospitalization (24).

Several possible mechanisms explaining an increased risk of HF in subjects without diabetes with IR have previously been proposed. These include the link between hyperinsulinemia and sodium retention (25), sympathetic nervous system activation (26), and increased pressor response to angiotensin II (27). Additionally, the metabolic alterations defining IR have been shown to have a direct effect on myocardial structure and function. Studies using various cardiac imaging modalities described concentric LV hypertrophy and increased LV mass in otherwise healthy subjects with IR (19,28). Moreover, IR and dysglycemia have been correlated with an adverse LV remodeling after myocardial infarction (19,29). Nevertheless, the implicated mechanism requires further studies.

Lastly, we would like to highlight the potential utility and relatively low cost of using HOMA2-IR for early risk stratification of people with newly diagnosed diabetes, which could facilitate targeting patients with IR (and at high risk of HF) with treatments particularly effective at reducing HF, e.g., sodium–glucose cotransporter 2 inhibitors. HOMA2-IR estimates have been validated against euglycemic clamp measures, the acute insulin response from the intravenous glucose tolerance test, and the insulinogenic index from oral glucose tolerance testing (16). Estimating IR by HOMA2-IR requires measuring only paired concentrations of fasting insulin and FPG and could therefore be used in clinical practice, in addition to epidemiological studies.

Our study has several limitations. Firstly, information about LV ejection fraction was not collected routinely in the UKPDS, so it is difficult to draw any conclusions regarding the etiology of the HF events that occurred. The UKPDS based the diagnosis of HF primarily on clinical judgement rather than current imaging or biomarker criteria and therefore does not fully represent a contemporary definition of HF. Information about LV ejection fraction was not collected routinely in the UKPDS, so it is difficult to draw any conclusions regarding the etiology of the HF events that occurred. It is also possible that HF preserved ejection fraction, which is frequently observed in people with diabetes, may not have been fully captured in the UKPDS. Secondly, the number of adjudicated HF events in UKPDS is relatively small compared with the number of deaths. We also evaluated a composite end point of HF or death to account for the competing risk of death and present the analysis separately to visualize the HF-specific risk. Thirdly, the substantially increased risk for HF/death in the 0.7% of UKPDS participants with prior AF needs to be evaluated further in a much larger cohort, although it is noteworthy that AF has been included consistently in HF risk score models developed from various large clinical trials (30,31). Finally, we are not able to exclude that the association we found between higher HOMA2-IR values and the future risk of HF/death reflects residual confounding. IR is a feature of a spectrum of cardiometabolic disorders, many of which increase cardiovascular mortality.

In conclusion, this large post hoc analysis showed that participants with IR and newly diagnosed T2D were more likely to develop HF or die than those who were more insulin sensitive. These associations were independent of BMI. Accordingly, we propose that HOMA2-IR measured at the time of the diagnosis of T2D could be an important additional predictor of the future risk of HF and death in people with T2D.

Clinical trial reg. no. ISRCTN75451837, www.isrctn.org

Acknowledgments. The authors thank the participants, without whom this study and these analyses would not have been possible.

Funding. This study was funded by an Oxfordshire Health Services Research Committee grant (OHSRC ref. 1239) awarded to M.W. A.I.A. acknowledges funding from the Oxford Biomedical Research Centre. J.J.V.M. is supported by British Heart Foundation Centre of Research Excellence grant RE/18/6/34217 and reports consulting payments through Glasgow University from Kidney Research UK. R.R.H. is an emeritus U.K. National Institute for Health Research Senior Investigator.

Duality of Interest. J.J.V.M. reports consulting payments through Glasgow University from Bayer, Cardiorentis, Amgen, Theracos, AbbVie, DalCor Pharmaceuticals, Pfizer, Merck, AstraZeneca, GlaxoSmithKline, Bristol-Myers Squibb, Vifor Fresenius Medical Care Renal Pharma, Novartis, and Theracos. R.R.H. reports research support from AstraZeneca, Bayer, and Merck Sharp & Dohme and personal fees from Anji Pharmacueticals, Bayer, Merck Sharp & Dohme, Novartis, and Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.W. wrote the manuscript. M.W. and R.R.H. designed the study. R.L.C. performed the analyses. R.L.C., A.I.A., J.J.V.M., and R.R.H. reviewed and edited the manuscript. R.R.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Ingelsson
E
,
Sundström
J
,
Arnlöv
J
,
Zethelius
B
,
Lind
L
.
Insulin resistance and risk of congestive heart failure
.
JAMA
2005
;
294
:
334
341
2.
Banerjee
D
,
Biggs
ML
,
Mercer
L
, et al
.
Insulin resistance and risk of incident heart failure: Cardiovascular Health Study
.
Circ Heart Fail
2013
;
6
:
364
370
3.
Vardeny
O
,
Gupta
DK
,
Claggett
B
, et al
.
Insulin resistance and incident heart failure the ARIC study (Atherosclerosis Risk in Communities)
.
JACC Heart Fail
2013
;
1
:
531
536
4.
Kenchaiah
S
,
Evans
JC
,
Levy
D
, et al
.
Obesity and the risk of heart failure
.
N Engl J Med
2002
;
347
:
305
313
5.
Rijzewijk
LJ
,
van der Meer
RW
,
Smit
JW
, et al
.
Myocardial steatosis is an independent predictor of diastolic dysfunction in type 2 diabetes mellitus
.
J Am Coll Cardiol
2008
;
52
:
1793
1799
6.
Shah
RV
,
Abbasi
SA
,
Heydari
B
, et al
.
Insulin resistance, subclinical left ventricular remodeling, and the obesity paradox: MESA (Multi-Ethnic Study of Atherosclerosis)
.
J Am Coll Cardiol
2013
;
61
:
1698
1706
7.
Nelson
MD
,
Victor
RG
,
Szczepaniak
EW
,
Simha
V
,
Garg
A
,
Szczepaniak
LS
.
Cardiac steatosis and left ventricular hypertrophy in patients with generalized lipodystrophy as determined by magnetic resonance spectroscopy and imaging
.
Am J Cardiol
2013
;
112
:
1019
1024
8.
Ilercil
A
,
Devereux
RB
,
Roman
MJ
, et al
.
Associations of insulin levels with left ventricular structure and function in American Indians: the Strong Heart Study
.
Diabetes
2002
;
51
:
1543
1547
9.
UK Prospective Diabetes Study (UKPDS)
.
VIII. Study design, progress and performance
.
Diabetologia
1991
;
34
:
877
890
10.
Holman
RR
,
Paul
SK
,
Bethel
MA
,
Matthews
DR
,
Neil
HA
.
10-year follow-up of intensive glucose control in type 2 diabetes
.
N Engl J Med
2008
;
359
:
1577
1589
11.
UK Prospective Diabetes Study (UKPDS) Group
.
Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33)
.
Lancet
1998
;
352
:
837
853
12.
Hypertension in Diabetes Study (HDS)
.
Hypertension in Diabetes Study (HDS): I. Prevalence of hypertension in newly presenting type 2 diabetic patients and the association with risk factors for cardiovascular and diabetic complications
.
J Hypertens
1993
;
11
:
309
317
13.
HOMA2 Calculator
.
Accessed 21 February 2021. Available from https://www.dtu.ox.ac.uk/homacalculator/
14.
Matthews
DR
,
Hosker
JP
,
Rudenski
AS
,
Naylor
BA
,
Treacher
DF
,
Turner
RC
.
Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man
.
Diabetologia
1985
;
28
:
412
419
15.
Hermans
MP
,
Levy
JC
,
Morris
RJ
,
Turner
RC
.
Comparison of insulin sensitivity tests across a range of glucose tolerance from normal to diabetes
.
Diabetologia
1999
;
42
:
678
687
16.
Wallace
TM
,
Levy
JC
,
Matthews
DR
.
Use and abuse of HOMA modeling
.
Diabetes Care
2004
;
27
:
1487
1495
17.
World Health Organization
.
Body mass index - BMI
.
18.
Bland
JM
,
Altman
DG
.
Survival probabilities (the Kaplan-Meier method)
.
BMJ
1998
;
317
:
1572
19.
Velagaleti
RS
,
Gona
P
,
Chuang
ML
, et al
.
Relations of insulin resistance and glycemic abnormalities to cardiovascular magnetic resonance measures of cardiac structure and function: the Framingham Heart Study
.
Circ Cardiovasc Imaging
2010
;
3
:
257
263
20.
Resnick
HE
,
Jones
K
,
Ruotolo
G
, et al.;
Strong Heart Study
.
Insulin resistance, the metabolic syndrome, and risk of incident cardiovascular disease in nondiabetic American Indians: the Strong Heart Study
.
Diabetes Care
2003
;
26
:
861
867
21.
Adler
AI
,
Neil
HA
,
Manley
SE
,
Holman
RR
,
Turner
RC
.
Hyperglycemia and hyperinsulinemia at diagnosis of diabetes and their association with subsequent cardiovascular disease in the United Kingdom prospective diabetes study (UKPDS 47)
.
Am Heart J
1999
;
138
:
S353
S359
22.
Adler
AI
,
Levy
JC
,
Matthews
DR
,
Stratton
IM
,
Hines
G
,
Holman
RR
.
Insulin sensitivity at diagnosis of type 2 diabetes is not associated with subsequent cardiovascular disease (UKPDS 67)
.
Diabet Med
2005
;
22
:
306
311
23.
U.K. Prospective Diabetes Study Group
.
U.K. Prospective Diabetes Study 16: Overview of 6 years’ therapy of type II diabetes: a progressive disease
.
Diabetes
1995
;
44
:
1249
1258
24.
Young
LH
,
Viscoli
CM
,
Schwartz
GG
, et al.;
IRIS Investigators
.
Heart failure after ischemic stroke or transient ischemic attack in insulin-resistant patients without diabetes mellitus treated with pioglitazone
.
Circulation
2018
;
138
:
1210
1220
25.
DeFronzo
RA
,
Cooke
CR
,
Andres
R
,
Faloona
GR
,
Davis
PJ
.
The effect of insulin on renal handling of sodium, potassium, calcium, and phosphate in man
.
J Clin Invest
1975
;
55
:
845
855
26.
Anderson
EA
,
Hoffman
RP
,
Balon
TW
,
Sinkey
CA
,
Mark
AL
.
Hyperinsulinemia produces both sympathetic neural activation and vasodilation in normal humans
.
J Clin Invest
1991
;
87
:
2246
2252
27.
Gaboury
CL
,
Simonson
DC
,
Seely
EW
,
Hollenberg
NK
,
Williams
GH
.
Relation of pressor responsiveness to angiotensin II and insulin resistance in hypertension
.
J Clin Invest
1994
;
94
:
2295
2300
28.
Sundström
J
,
Lind
L
,
Nyström
N
, et al
.
Left ventricular concentric remodeling rather than left ventricular hypertrophy is related to the insulin resistance syndrome in elderly men
.
Circulation
2000
;
101
:
2595
2600
29.
Yang
CD
,
Shen
Y
,
Lu
L
, et al
.
Insulin resistance and dysglycemia are associated with left ventricular remodeling after myocardial infarction in non-diabetic patients
.
Cardiovasc Diabetol
2019
;
18
:
100
30.
Yang
H
,
Negishi
K
,
Otahal
P
,
Marwick
TH
.
Clinical prediction of incident heart failure risk: a systematic review and meta-analysis
.
Open Heart
2015
;
2
:
e000222
31.
Segar
MW
,
Vaduganathan
M
,
Patel
KV
, et al
.
Machine learning to predict the risk of incident heart failure hospitalization among patients with diabetes: the WATCH-DM risk score
.
Diabetes Care
2019
;
42
:
2298
2306
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.