Diagnosis of Nonischemic Stage B Heart Failure in Type 2 Diabetes Mellitus: Optimal Parameters for Prediction of Heart Failure
Original Research
Graphical abstract
Abstract
Objectives:
This study sought to identify whether impaired global longitudinal strain (GLS), diastolic dysfunction (DD), or left atrial enlargement (LAE) should be added to stage B heart failure (SBHF) criteria in asymptomatic patients with type 2 diabetes mellitus.
Background:
SBHF is a precursor to clinical heart failure (HF), and its recognition justifies initiation of cardioprotective therapy. However, original definitions of SBHF were based on LV hypertrophy and impaired ejection fraction.
Methods:
Patients with asymptomatic type 2 diabetes mellitus ≥65 years of age (age 71 ± 4 years; 55% men) with preserved ejection fraction and no ischemic heart disease were recruited from a community-based population. All underwent a standard clinical evaluation, and a comprehensive echocardiogram, including assessment of left ventricular hypertrophy (LVH), LAE, DD (abnormal E/e′), and GLS (<16%). Over a median follow-up of 1.5 years (range 0.5 to 3), 20 patients were lost to follow-up, and 290 individuals were entered into the final analyses.
Results:
In this asymptomatic group, LV dysfunction was identified in 30 (10%) by DD, 68 (23%) by LVH, 102 (35%) by LAE, and 68 (23%) by impaired GLS. New-onset HF developed in 45 patients and 4 died, giving an event rate of 112/1,000 person-years. Survival free of the composite endpoint (HF and death) was about 1.5-fold higher in patients without a normal, compared with an abnormal echocardiogram. LVH, LAE, and GLS <16% were associated with increased risk of the composite endpoint, independent of ARIC risk score and glycosylated hemoglobin, but abnormal E/e′ was not. The addition of left atrial volume and GLS provided incremental value to the current standard of clinical risk (ARIC score) and LVH. In a competing-risks regression analysis, LVH (hazard ratio: 2.90; p < 0.001) and GLS <16% (hazard ratio: 2.26; p = 0.008), but not DD and LAE were associated with incident HF.
Conclusions:
Subclinical left ventricular systolic dysfunction is prevalent in asymptomatic elderly patients with type 2 diabetes mellitus, and impaired GLS is independent and incremental to LVH in the prediction of incident HF.
Introduction
Type 2 diabetes mellitus (T2DM) is a potent risk factor for the development of nonischemic heart failure (HF) (1,2). The development of HF may be preceded by stage B HF (SBHF) (3), the recognition of which justifies the use of cardioprotective medication to prevent or retard the progression of HF. The definition of SBHF on the basis of valvular disease, evidence of previous infarction, left ventricular hypertrophy (LVH), and impaired ejection fraction (EF) (3) reflects an era when the entity was dominated by ischemic HF with reduced EF. Unfortunately, the cutoff for “low EF” cited in the guidelines is not clearly defined, with a previous review identifying previous studies of asymptomatic LV dysfunction (LVD) to include an EF range of 30% to 54% (4). Moreover, the evidence base for prognostic benefit of treating asymptomatic LVD almost exclusively involves EF <40% (5). Because 83% of elderly patients with T2DM and newly diagnosed HF have HF with preserved EF (6) standard criteria seem unsuited to this group.
LVD is highly prevalent in T2DM, with up to one-half of the patients involved (7). Global longitudinal strain (GLS) is the most robust LV strain parameter, which may now be readily measured using speckle-tracking echocardiography (8). This is more sensitive and specific than conventional 2-dimensional EF as a measure of systolic function, and can be used to identify subclinical systolic LVD in cardiomyopathies (9). Previous studies have demonstrated that early detection of subclinical LVD by strain imaging is independently associated with long-term adverse outcome in asymptomatic patients with T2DM (10,11). Likewise, diastolic dysfunction (DD) and left atrial enlargement (LAE) are potent prognostic markers in HF, although this evidence has focused on HF-free survival. However, the value of screening of asymptomatic T2DM patients for SBHF is undefined, and a vital step is to identify the optimal parameters. We hypothesized that in addition to LVH, the presence of impaired GLS, DD, and LAE predict incident HF in asymptomatic patients with T2DM.
Methods
Patient selection
Patient selection
We prospectively recruited 310 asymptomatic T2DM patients aged ≥65 years with preserved LVEF from a community-based population in Tasmania from 2013 and 2015. The recruitment exclusion criteria were existing HF or known ischemic heart disease, more than moderate valve disease, history of HF or LVEF <40%, and inability to acquire adequate echocardiographic images for speckle-tracking imaging analysis at baseline. All participants provided written, informed consent, and the study protocol was approved by the Tasmanian Human Research Ethics Committee.
Clinical features
Socioeconomic status, demographics, medical history, family history, and use of medication were obtained from a baseline survey. The diagnosis of T2DM was based on self-report including diabetes treatment. Obesity was defined as body mass index ≥30 kg/m2. Waist circumference was measured to the nearest millimeter by a trained examiner using a tape measure at the midpoint between the lower costal margin and the iliac crest. Supine resting blood pressure was measured twice and averaged in each patient after at least 10 min rest in a quiet room. Hypertension was defined by averaged systolic blood pressure ≥140 mm Hg or a diastolic blood pressure ≥90 mm Hg. International Federation of Clinical Chemistry standardized hemoglobin A1c (HbA1c) was obtained from local pathology records, and a cutoff of 64 mmol/mol was used to denote adequate control, based on the age and risk status of the study group (12). In 85 patients, missing values for HbA1c were estimated by imputation using linear regression. The ARIC (Atherosclerosis Risk In Communities) HF risk scores were used to estimate the absolute risk of HF at 3 years. The online ARIC Heart Failure Risk Calculator was used and 10 common clinical variables were included (age, sex, race, systolic blood pressure, heart rate, body mass index, smoking status, previous coronary heart disease, current use of blood pressure–lowering medication, and diabetes mellitus) (13).
Echocardiography
Patients all underwent a comprehensive echocardiogram including standard transthoracic 2-dimensional, Doppler echocardiographic studies, and speckle-tracking echocardiogram in accordance with the American Society of Echocardiography guidelines (ACUSON SC2000, 4V1c, and 4Z1c probes, Siemens Healthcare, Mountain View, California) (14,15). LV internal dimensions and wall thickness, chamber volumes, and valvular morphology were assessed. LVEF was measured using the modified Simpson biplane method. Left atrial (LA) volume was calculated using the biplane method of disks and indexed to body surface area (BSA); left atrial volume index ≥34 ml/m2 was used to define LAE (14). Left ventricular mass index (LVMi) was obtained from M-mode LV mass measurement using standard criteria and normalized for body size (BSA, height to the power of 1.7 or 2.7) (14). LVH was defined as LVMi (normalized for BSA) >115 g/m2 for men and >95 g/m2 for women. LV inflow was obtained using pulsed wave Doppler in the apical 4-chamber view; peak early (E) and late (A) diastolic velocities, deceleration time, and E/A ratio were obtained. Peak early diastolic medial and lateral mitral annular velocity (e′) and the ratio of mitral inflow early diastolic velocity to average e′ velocity were obtained from pulsed tissue Doppler; E/e′ >13 was used as a cutoff of DD (16). For deformation imaging, standard grayscale 2-dimensional images were acquired in conventional 4-chamber, 2-chamber, 3-chamber, parasternal short-axis views at the mid, basal, and apical level. GLS was calculated by average of 3 apical views using standard software (17,18). Although in deformation analysis shortening is described as a negative number, for computational simplicity (and because there were no positive GLS) we express GLS in this paper without this information. Our laboratory uses cutoffs of 16% to designate impaired GLS and 18% to designate abnormal GLS, and we evaluated both cutoffs.
Accordingly, SBHF was defined by: 1) DD (E/e′ >13); 2) LAE (>34 ml/m2); 3) LVH (>115 g/m2 for men, >95 g/m2 for women); and 4) impaired GLS (cutoff 16%).
Outcomes
Regular telephone calls and surveys were performed to identify potential HF symptoms during follow-up, and symptom surveillance questionnaires and clinical visits were followed. Records of all-cause hospitalization were obtained and collected. Patients were examined, and the diagnosis of incident HF was made by consensus of 3 independent cardiologists, using the Framingham HF criteria (19). The primary composite endpoint for study was new-onset of HF and all-cause mortality. The primary endpoint was incident HF, and all-cause mortality was considered as a competing risk.
Statistical analysis
Continuous variables are presented as mean ± SD and categorical data as frequencies and percentages. Cox regression models were used to model time to composite endpoint and reported as hazard ratio (HR) and 95% confidence interval (20). The cumulative survival free of HF incidence and death during the follow-up period was estimated using the Kaplan-Meier method and survival curves were compared with the log-rank test. Univariable Cox regression was used to assess the predictive power of composite endpoint among clinical, demographic, biochemical, and echocardiographic variables. Significant variables and echocardiographic parameters of interest were selected for entry into multivariate analyses to determine independent correlates. To account for the competing risk of all-cause death during follow-up, competing risk methods were used to generate HR and 95% confidence intervals for the associations between each echocardiographic parameter and incident HF (20). The cumulative incidences of HF were calculated and graphically displayed separately for patients with and without impaired GLS. Gray K-sample test was used to compare the cumulative incidence estimates of HF between patients with and without impaired GLS (21). All data were analyzed using standard statistical computer software SPSS version 22 (IBM, Chicago, Illinois; and Stata version 12.0, StataCorp, College Station, Texas); p < 0.05 was deemed to be statistically significant.
Results
Patient characteristics
Patient characteristics
Of 310 eligible asymptomatic T2DM patients ≥65 years of age with preserved EF from the community who underwent baseline tests, 2 (0.7%) were lost to follow-up and 18 participants (5.8%) were alive but unable to attend for clinic review, after a median follow-up time of 1.5 years (range 0.5 to 3.0 years) (Figure 1). This group was no different from the remaining 290 individuals who completed follow-up (Online Table 1). The baseline clinical and echocardiographic characteristics of 290 individuals with and without events are summarized in Table 1. In the remaining 290 T2DM participants (71 ± 4 years of age; 56% male), the mean HbA1c level was 53.7 ± 10.3 mmol/mol. HF risk factors were prevalent among these patients: 77% had hypertension, 49% had obesity, 31% had family history of HF, 8% had exposure to chemotherapy, and 7% had history of heart disease.
Flowchart
Original study group, reasons for exclusion, and patients enrolled in the study.
| Age, yrs | 70.9 ± 4.3 |
| Male | 163 (56.2) |
| Weight, kg | 85.9 ± 17.0 |
| Height, cm | 168.0 ± 10.3 |
| BMI, kg/m2 | 30.3 ± 5.9 |
| Waist circumference, cm | 103.4 ± 13.0 |
| HbA1c, mmol/mol | 53.7 ± 10.3 |
| Poor HbA1c∗ | 38 (13.1) |
| Obesity | 142 (49.0) |
| Heart rate, beats/min | 69 ± 11 |
| Systolic blood pressure, mm Hg | 139 ± 14 |
| Diastolic blood pressure, mm Hg | 81 ± 10 |
| Hypertension | 222 (76.6) |
| Family history of HF | 90 (31.0) |
| Past chemotherapy | 24 (8.3) |
| Past heart disease | 20 (6.9) |
| ARIC HF risk score (3 yrs), % | 7.5 ± 6.4 |
| Medication | |
|---|---|
| Insulin | 69 (23.8) |
| Metformin | 196 (67.6) |
| ACE inhibitor/ARB | 201 (69.3) |
| Beta-blockers | 16 (5.5) |
| Calcium antagonists | 68 (23.4) |
| Diuretics | 33 (11.4) |
| Lipid-lowering medications | 148 (51.0) |
| Echocardiography | |
| LV ejection fraction, % | 62.9 ± 6.5 |
| Mitral early diastolic inflow velocity (E-wave), m/s | 0.65 ± 0.17 |
| Mitral late-diastolic inflow velocity (A-wave), m/s | 0.83 ± 0.19 |
| Transmitral diastolic flow velocity ratio, E/A | 0.80 ± 0.21 |
| Early diastolic mitral annular velocity (e'), m/s | 0.08 ± 0.02 |
| Mitral E/e′ septal-lateral ratio, E/e′ | 9.2 ± 2.7 |
| Deceleration time, s | 246.4 ± 52.4 |
| LA volume, ml/m2 | 32.3 ± 10.1 |
| LVMi, g/m2 | 92.4 ± 23.8 |
| LVMi, g/m1.7 | 74.7 ± 21.4 |
| LVMi, g/m2.7 | 44.4 ± 12.6 |
| GLS, % | 17.7 ± 2.6 |
| Echocardiography categorical variables | |
| E/e′ ratio >13 | 30 (10.3) |
| LAE† | 102 (35.2) |
| LVH‡ | 68 (23.4) |
| GLS <18% | 146 (50.3) |
| GLS <16% | 68 (23.4) |
| Presence of any SBHF features§ | 169 (58.3) |
Prevalence of SBHF
Baseline LVEF was preserved (≥40%) in the entire cohort; the average EF was 62.9 ± 6.5% and 4 patients had an EF of 40% to 50%. LVH was present in 68 patients (23%). Subclinical LVD was identified in 10% by abnormal E/e′, 35% by LAE, 50% by abnormal GLS (cutoff 18%), and 23% by impaired GLS (cutoff 16%). In the entire cohort, 97 (33%) had only 1 echocardiographic abnormality, 48 (17%) had 2 echocardiographic abnormalities, 21 (7%) had 3, and 3 (1%) had 4; in total 58% had any feature. The distribution of patients with multiple features is shown in Figure 2.
SBHF Features (LVH, LAE, Abnormal E/e′, and Impaired GLS) Among Elderly Asymptomatic Patients With T2DM (n = 169)
The distribution of these findings is shown among patients with only 1 feature (A), any 2 features (B), and >2 features (C). GLS = global longitudinal strain; LAE = left atrial enlargement; LVH = left ventricular hypertrophy; SBHF = stage B heart failure; T2DM = type 2 diabetes mellitus.
Follow-up
Over a median follow-up of 18 months (interquartile range: 12 to 24 months), 45 patients developed new-onset HF and 4 died, giving an event rate of 112/1,000 person-years. Patients who had events were more commonly men and showed greater levels of obesity (especially central obesity), worse glycemic control, higher LVMi and LA volume, and worse GLS. EF and E/e′ ratio showed no difference between the 2 groups (p = 0.096). Of the 49 individuals having events, 82% had any of the echocardiographic features: 47% had LVH, 12% had abnormal E/e′, 61% had LAE, 65% had GLS <18%, and 45% had GLS <16%.
Survival free of the composite endpoint (HF and death) was about 1.5-fold higher in patients with any echocardiographic abnormalities, compared with patients with a normal echocardiogram (Figure 3A). The proportion of cumulative event-free survival was less with increasing numbers of echocardiographic features of SBHF (Figure 3B).
Primary Outcomes Between Patients With Diabetes
Comparison of the outcomes of patients (A) with and without SBHF features, and (B) increasing numbers of SBHF echocardiographic features. Abbreviation as in Figure 2.
Predictors of HF and death
Cox regression analysis was performed to analyze the association between clinical variables and echocardiographic parameters of interest and the time to the primary composite endpoint (Table 2). Multivariable Cox regression models were constructed to determine whether impaired GLS and other echocardiographic SBHF features of interest were associated with the composite endpoint, after accounting for clinical risk variables for HF, summarized by the ARIC risk score and glycemic control. Significant univariable parameters of interest (ARIC risk score, HbA1c >64 mmol/mol, E/e′ >13, LAE, LVH, and GLS <16%) were entered into the model (Table 3). LVH, LAE, and GLS <16% were associated with increased risk of the composite endpoint, independent of ARIC risk score and HbA1c, but abnormal E/e′ was not. The incremental value of adding abnormal LA volume and GLS to the current standard of clinical risk (ARIC score) and LVH is shown in Figure 4A.
| Event (n = 49) | No Event (n = 241) | HR (95% CI) | p Value | |
|---|---|---|---|---|
| Age, yrs | 71.9 ± 4.7 | 70.7 ± 4.2 | 1.06 (0.99–1.12) | 0.070 |
| Male | 33 (67.3) | 130 (53.9) | 1.58 (0.87–2.86) | 0.136 |
| Weight, kg | 95.4 ± 19.7 | 84.1 ± 15.7 | 1.04 (1.02–1.05) | <0.001 |
| Height, cm | 169.2 ± 10.4 | 168.3 ± 9.7 | 1.01 (0.98–1.04) | 0.673 |
| BMI, kg/m2 | 33.6 ± 7.7 | 29.7 ± 5.2 | 1.10 (1.06–1.14) | <0.001 |
| Waist circumference, cm | 110.6 ± 13.5 | 101.9 ± 12.5 | 1.05 (1.03–1.08) | <0.001 |
| HbA1c, mmol/mol | 57.5 ± 11.6 | 52.9 ± 9.9 | 1.03 (1.01–1.05) | 0.003 |
| Poor HbA1c∗ | 14 (28.6) | 24 (10.0) | 2.41 (1.25–4.65) | 0.009 |
| Obesity | 34 (69.4) | 108 (44.8) | 2.77 (1.51–4.09) | 0.001 |
| Heart rate, beats/min | 68 ± 12 | 69 ± 10 | 0.99 (0.97–1.03) | 0.856 |
| Systolic blood pressure, mm Hg | 139 ± 15 | 139 ± 14 | 1.00 (0.98–1.02) | 0.860 |
| Diastolic blood pressure, mm Hg | 81 ± 9 | 81 ± 10 | 1.00 (0.98–1.03) | 0.811 |
| Hypertension | 39 (79.6) | 183 (75.9) | 1.27 (0.63–2.54) | 0.505 |
| Family history of HF | 11 (22.4) | 79 (32.8) | 0.62 (0.32–1.21) | 0.163 |
| Past chemotherapy | 5 (10.2) | 19 (7.9) | 1.34 (0.53–3.39) | 0.536 |
| Past heart disease | 6 (12.2) | 14 (5.8) | 1.94 (0.82–4.56) | 0.456 |
| ARIC HF risk score (3 yrs), % | 11.1 ± 8.9 | 6.8 ± 5.4 | 1.08 (1.05–1.11) | <0.001 |
| Medication | ||||
| Insulin | 16 (32.7) | 53 (22.0) | 1.57 (0.86–2.86) | 0.140 |
| Metformin | 31 (63.3) | 165 (68.5) | 0.89 (0.50–1.59) | 0.688 |
| ACE inhibitor/ARB | 37 (75.5) | 164 (68.4) | 1.37 (0.72–2.63) | 0.344 |
| Beta-blockers | 6 (12.2) | 10 (4.1) | 2.89 (1.23–6.81) | 0.015 |
| Calcium antagonists | 7 (14.3) | 61 (25.3) | 0.50 (0.22–1.11) | 0.089 |
| Diuretics | 6 (12.2) | 27 (11.2) | 1.08 (0.46–2.54) | 0.858 |
| Lipid-lowering medications | 31 (63.3) | 117 (48.5) | 1.48 (0.83–2.65) | 0.186 |
| Echocardiography | ||||
| LV ejection fraction, % | 61.2 ± 7.9 | 63.2 ± 6.1 | 0.95 (0.91–0.99) | 0.026 |
| Mitral early diastolic inflow velocity (E wave), m/s | 0.69 ± 0.21 | 0.65 ± 0.16 | 4.35 (0.88–21.44) | 0.071 |
| Mitral late-diastolic inflow velocity (A wave), m/s | 0.87 ± 0.27 | 0.82 ± 0.18 | 2.67 (0.62–11.55) | 0.190 |
| Transmitral diastolic flow velocity ratio, E/A | 0.79 ± 0.24 | 0.80 ± 0.20 | 0.96 (0.20–4.59) | 0.960 |
| Early diastolic mitral annular velocity (e'), m/s | 0.08 ± 0.02 | 0.08 ± 0.01 | 1.06 (0.89–1.27) | 0.502 |
| Mitral E/e′ septal-lateral ratio, E/e′ | 9.6 ± 3.1 | 9.1 ± 2.6 | 1.06 (0.96–1.17) | 0.251 |
| Deceleration time, s | 247.5 ± 57.5 | 246.1 ± 51.4 | 1.00 (0.99–1.01) | 0.757 |
| LA volume, ml/m2 | 38.0 ± 11.5 | 31.1 ± 9.3 | 1.06 (1.03–1.08) | <0.001 |
| LVMi, g/m2 | 104.0 ± 27.1 | 90.1 ± 22.5 | 1.02 (1.01–1.03) | <0.001 |
| LVMi, g/m1.7 | 87.0 ± 25.2 | 72.2 ± 19.7 | 1.03 (1.02–1.04) | <0.001 |
| LVMi, g/m2.7 | 51.6 ± 14.9 | 42.9 ± 11.6 | 1.05 (1.03–1.07) | <0.001 |
| GLS, % | 16.3 ± 2.9 | 18.0 ± 2.4 | 1.27 (1.15–1.41) | <0.001 |
| Echocardiography categorical variables | ||||
| E/e′ ratio >13 | 6 (12.2) | 24 (10.0) | 1.35 (0.58–3.18) | 0.488 |
| LVH† | 23 (46.9) | 45 (18.7) | 3.52 (2.00–6.18) | <0.001 |
| LAE‡ | 30 (61.2) | 72 (29.9) | 3.29 (1.85–5.86) | <0.001 |
| GLS <18% | 32 (65.3) | 114 (47.3) | 2.13 (1.18–3.84) | 0.012 |
| GLS <16% | 22 (44.9) | 46 (19.1) | 3.24 (1.94–5.69) | <0.001 |
| Any presence of SBHF features§ | 40 (81.6) | 129 (53.5) | 4.09 (1.62–10.33) | 0.003 |
| Unadjusted HR (95% CI) p Value | E/e′ Adjusted HR (95% CI)∗ p Value | LAE Adjusted HR (95% CI)∗ p Value | LVH Adjusted HR (95% CI)∗ p Value | GLS Adjusted HR (95% CI)∗ p Value | All Except LAE Adjusted HR (95% CI)∗ p Value | All Except E/e′ Adjusted HR (95% CI)† p Value | |
|---|---|---|---|---|---|---|---|
| ARIC HF risk score (3 yrs) | 1.08 (1.05–1.11) <0.001 | 1.08 (1.04–1.11) <0.001 | 1.06 (1.03–1.10) <0.001 | 1.06 (1.02–1.09) 0.001 | 1.07 (1.04–1.11) <0.001 | 1.06 (1.02–1.10) 0.002 | 1.05 (1.01–1.09) 0.007 |
| Poor HbA1c | 2.83 (1.52–5.27) 0.001 | 2.72 (1.45–5.07) 0.002 | 2.79 (1.50–5.19) 0.001 | 2.34 (1.24–4.40) 0.008 | 2.26 (1.20–4.26) 0.011 | 1.97 (1.01–3.81) 0.045 | 2.11 (1.10–4.04) 0.025 |
| Abnormal E/e′ | 1.35 (0.58–3.18) 0.488 | 1.11 (0.47–2.66) 0.810 | — | — | — | 0.86 (0.34–2.16) 0.746 | — |
| LAE | 3.29 (1.85–5.86) <0.001 | — | 2.80 (1.55–5.05) 0.001 | — | — | — | 2.34 (1.27–4.32) 0.007 |
| LVH | 3.52 (2.00–6.18) <0.001 | — | — | 2.34 (1.26–4.35) 0.007 | — | 2.04 (1.08–3.84) 0.027 | 1.62 (0.84–3.14) 0.149 |
| Impaired GLS | 3.24 (1.84–5.69) <0.001 | — | — | — | 2.67 (1.51–4.75) 0.001 | 2.46 (1.36–4.45) 0.003 | 2.29 (1.26–4.15) 0.007 |
| C-statistic | 0.679 (0.593–0.766) | 0.753 (0.679–0.827) | 0.732 (0.652–0.811) | 0.737 (0.661–0.814) | 0.760 (0.683–0.837) | 0.792 (0.723–0.861) |
Receiver-Operating Characteristic Curves From Adding GLS and Left Atrial Size to Existing Risk Evaluation Using Clinical Risk (ARIC Score) and LVH
Incremental value of (A) categorical assessment of abnormal GLS and left atrial enlargement, and (B) GLS and LAVi as continuous variables. ARIC = Atherosclerosis Risk In Communities; AUC = area under the curve; HbA1c = glycosylated hemoglobin; LAVi = left atrial volume index; LVMi = left ventricular mass index; other abbreviations as in Figure 2.
Multivariable models were also used to assess the measures as continuous variables; LA volume (adjusted HR: 1.05; p = 0.001) and GLS (adjusted HR: 1.16; p = 0.008) were predictors of the primary endpoint, independent of HbA1c (adjusted HR: 1.03; p = 0.039) (Table 4). The incremental value of adding left atrial volume index and GLS as continuous variables to the current standard of clinical risk (ARIC score) and LV mass is shown in Figure 4B.
| Unadjusted HR (95% CI), p Value | Adjusted HR (95% CI), p Value | |
|---|---|---|
| HbA1c | 1.03 (1.01–1.05), 0.008 | 1.03 (1.00–1.06), 0.039 |
| E/e′ ratio | 1.06 (0.96–1.17), 0.251 | 1.00 (0.91–1.10), 0.960 |
| LA volume | 1.06 (1.03–1.08), <0.001 | 1.05 (1.02–1.07), 0.001 |
| LV mass index | 1.02 (1.01–1.03), <0.001 | 1.01 (0.99–1.02), 0.165 |
| GLS | 1.27 (1.15–1.41), <0.001 | 1.16 (1.04–1.30), 0.008 |
Because EF from <50% and LVH might be used to identify risk according to current guidelines, a subanalysis was performed in patients with EF >50% and without LVH (Online Table 2). This also showed that LAE and GLS (but not E/e′) was predictive of the composite endpoint.
Prediction of HF
A competing risk analysis that controlled for HbA1c was performed to assess whether impaired GLS and other echocardiographic SBHF features of interest were associated with incident HF in elderly asymptomatic patients with T2DM (Figure 5). The baseline model, showing a significant effect of poor glycemic control (p = 0.009) on incident HF, was improved by LVH (p = 0.042). However, the addition of impaired GLS (p = 0.008) had incremental predictive power to biochemical and other echocardiographic SBHF features for the prediction of incident HF (Figure 5).
Prediction of Incident Heart Failure in Elderly Asymptomatic Patients With T2DM (Competing Risk Analysis)
LVH and impaired GLS showed incremental value over glycemic control for incident heart failure. Poor HbA1c, ICFFstd-HbA1c >64 mmol/mol; LVH cutoff >115 g/m2 for males, >95 g/m2 for females; LAE cutoff >34 ml/m2; abnormal E/e′ cutoff = 13; GLS cutoff 16%. CI = confidence interval; HR = hazard ratio; other abbreviations as in Figures 2 and 4.
The cumulative incidence of HF with time among elderly asymptomatic T2DM stratified by impaired GLS status is shown in Figure 6. By the end of the follow-up, considering the competing risk, the cumulative incidence of HF was 0.17 in patients with GLS <16% and was 0.09 in patients with GLS ≥16%. Gray test also showed that the cumulative incidence of HF was significantly greater in patients with GLS <16% than those patients with GLS ≥16% (p < 0.001).
Cumulative Incidence Estimates of Heart Failure in Elderly Asymptomatic Patients With T2DM
This analysis stratifies by patients with and without impaired GLS. Abbreviations as in Figure 2.
Discussion
In this prospectively enrolled community-based cohort, poor glycemic control, LVH, LAE, and GLS <16% were independently associated with new-onset HF and death in elderly asymptomatic patients with T2DM who had no evidence of overt LVD. Although previous studies have shown LVD to predict mortality, to our knowledge, this is the first study to show that SBHF markers are useful for screening for new-onset HF in asymptomatic patients with T2DM that accounts for all-cause of mortality as a competing risk. Moreover, LAE and reduced GLS provided incremental prognostic information to clinical risk and LVH in patients with T2DM and preserved EF.
SBHF screening in T2DM
T2DM is a risk factor of HF, independent of hypertension, coronary artery disease, and other potential risk factors (19,22). The recognition of the preclinical stages of HF may permit the initiation of cardioprotection for these subjects, but the main functional marker for nonischemic SBHF is LVEF. Although readily obtainable, this is not the optimal diagnostic parameter, because HF with preserved EF is the most common manifestation of HF in T2DM. GLS is an effective means of detecting early changes in LV function (23), which is associated with adverse cardiac events over long-term follow-up (10,11).
A recently reported cluster analysis of echocardiographic variables in patients with T2DM identified distinct clinical profiles associated with 3 different phenotypes (24). The phenotype most associated with adverse outcome in this study (LVH with reduced GLS) corresponds with 1 of these groups and differs from the other 2, comprising patients with relatively mild dysfunction, and a group with DD and hypertensive heart disease. In our experience, the negative prognostic effect of impaired GLS was most evident in patients with GLS <16%, the independent effect of which doubled the risk of HF and provided a 2.4-fold increase in risk of the composite endpoint. Because these patients had preserved EF at baseline, these results support the use of GLS as a criterion for SBHF in community-based patients with T2DM.
Functional and structural changes in T2DM as indicators of HF
Patients with T2DM are at a 1.5-fold increment in risk of LVH, independent of various confounders including obesity, age, and sex (25). This finding is independent of hypertension and has been associated with insulin resistance (25). These results were confirmed in our study, which showed LVH in 68 patients (23%), and after adjustment for glycemic control, LVH by LVMi (indexed by BSA) was superior to biochemical variables in predicting incident HF. Although electrocardiography is widely available and less expensive than echocardiography, it is also less sensitive in detecting LVH in T2DM (26,27).
GLS derived from speckle-tracking measures the extent of tissue deformation as a percentage of the baseline at a longitudinal direction, and it could be used to identify subclinical LVD in cardiomyopathies. Diabetic cardiomyopathy is defined as LVD that is independent of coronary artery disease and hypertension (28,29). Impaired GLS is reported to be highly prevalent even in normotensive asymptomatic T2DM (10,30). This longitudinal dysfunction is caused by the complex interaction between metabolic, hemodynamic, and endocrine abnormalities (11). Limited studies have evaluated its predictive value for adverse cardiac events in T2DM.
E/e′ is an important predictor of cardiovascular events in patients with systolic HF or acute coronary syndrome (31,32). However, the role of DD as a manifestation of diabetic heart disease is controversial. Ernande et al. (24) reported that GLS was the primary disturbance of T2DM, and that DD was primarily a consequence of hypertension (7,10). In a study of 247 T2DM patients (mean age 60 years) with no history of cardiovascular complications, Liu et al. (11) showed that the presence of either impaired GLS (<17.9%) or high E/e′ (>13.6) had a predictive value of cardiovascular events beyond clinical data. However, in a prospective study of 406 middle-aged patients with DM, DD (E/e′ ratio >15) was a stronger independent predictor of cardiac death than GLS (33). In contrast, our results showed GLS to be stronger predictor of incident HF than E/e′ ratio. These differences in the reported literature are likely to reflect the different contribution of comorbid diseases in different populations.
Study limitations
First, the study population was self-selected from the community, based on the presence of at least 1 known nonischemic cardiovascular risk factor and excluding patients with a known history of HF or established asymptomatic LV systolic dysfunction. The process of recruitment through newspaper advertising may have resulted in population selection bias. Second, we did not record circumferential and radial strain. Although these provide additional information regarding myocardial mechanics, longitudinal strain is the most robust and reproducible parameter. Third, the main study endpoint was the onset of symptoms rather than acute decompensation with hospital admission. We chose this because symptoms (stage C heart failure) have clear implications for medical therapy, and the Framingham HF criteria have been widely used as a HF outcome. Moreover, differences in hospital admission (which has social and medical determinants) may take longer to emerge. Finally, we did not gather data on biomarkers, such as natriuretic peptides. Although these are potential predictors of HF and adverse outcome in T2DM, they may not be in the diagnostic range in obese patients (the average body mass index in this study was >30 kg/m2) and those with HF with preserved EF.
Conclusions
Asymptomatic LV systolic dysfunction as an expression of SBHF is highly prevalent in elderly asymptomatic patients with T2DM. Impaired GLS (<16%) and LVH (by LVM/BSA) were independently associated with incident HF over 2-year follow-up. Importantly, impaired GLS adds incremental prognostic value to glycemic control and other conventional echocardiographic parameters. The detection of early myocardial dysfunction may allow identification of asymptomatic patients with T2DM who are at risk of developing symptomatic HF.
COMPETENCY IN MEDICAL KNOWLEDGE: The recognition of SBHF is important because it justifies initiation of cardioprotective therapy. However, original definitions of SBHF were based on LVH and impaired EF. They may not capture the preclinical phase of HF with preserved EF, typical in T2DM.
TRANSLATIONAL OUTLOOK: Both impaired GLS and LAE are independent and incremental to LVH in the prediction of cardiac events, and GLS is incremental in the prediction of incident HF.
Appendix
1. : "Importance of risk factor management in diabetic patients and reduction in stage B heart failure". QJM 2015; 108: 307.
2. : "Diabetic cardiomyopathy: an expression of stage B heart failure with preserved ejection fraction". Diab Vasc Dis Res 2015; 12: 234.
3. : "ACC/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult". J Am Coll Cardiol 2005; 46: e1.
4. : "The epidemiology of “asymptomatic” left ventricular systolic dysfunction: implications for screening". Ann Intern Med 2003; 138: 907.
5. : "Stage B heart failure management of asymptomatic left ventricular systolic dysfunction". Circulation 2006; 113: 2851.
6. : "Uncovering heart failure with preserved ejection fraction in patients with type 2 diabetes in primary care: time for a change". Neth Heart J 2016; 24: 237.
7. : "The increasing detection of asymptomatic left ventricular dysfunction in patients with type 2 diabetes mellitus without overt cardiac disease". Diabetes Res Clin Pract 2013; 101: 309.
8. : "Measurement of strain and strain rate by echocardiography: ready for prime time?". J Am Coll Cardiol 2006; 47: 1313.
9. : "Myocardial strain imaging: how useful is it in clinical decision making?". Eur Heart J 2016; 37: 1196.
10. : "Subclinical LV dysfunction and 10-year outcomes in type 2 diabetes mellitus". Heart 2015; 101: 1061.
11. : "Incremental prognostic value of global longitudinal strain in patients with type 2 diabetes mellitus". Cardiovasc Diabetol 2016; 15: 22.
12. : "Diabetes in older adults: a consensus report". J Am Geriatr Soc 2012; 60: 2342.
13. ARIC heart failure risk calculator using clinical factors. Available at: http://aricnews.net/HFCalcs/RiskCalcHFClinFac.html. Accessed May 8, 2018.
14. : "Recommendations for cardiac chamber quantification by echocardiography in adults". Eur Heart J Cardiovasc Imaging 2015; 16: 233.
15. : "Recommendations for the evaluation of left ventricular diastolic function by echocardiography". Eur Heart J Cardiovasc Imaging 2009; 10: 165.
16. : "Recommendations for the evaluation of left ventricular diastolic function by echocardiography". J Am Soc Echocardiogr 2009; 22: 107.
17. : "Normal ranges of left ventricular strain: a meta-analysis". J Am Soc Echocardiogr 2013; 26: 185.
18. : "Myocardial strain measurement with 2-dimensional speckle-tracking echocardiography: definition of normal range". J Am Coll Cardiol Img 2009; 2: 80.
19. : "The epidemiology of heart failure: the Framingham Study". J Am Coll Cardiol 1993; 22: 6A.
20. : "A proportional hazards model for the subdistribution of a competing risk". J Am Stat Assoc 1999; 94: 496.
21. : "A class of K-sample tests for comparing the cumulative incidence of a competing risk". Ann Statist 1988; 16: 1141.
22. : "Heart failure prevalence, incidence, and mortality in the elderly with diabetes". Diabetes Care 2004; 27: 699.
23. : "Prognostic implications of global LV dysfunction: a systematic review and meta-analysis of global longitudinal strain and ejection fraction". Heart 2014; 100: 1673.
24. : "Clinical implications of echocardiographic phenotypes of patients with diabetes mellitus". J Am Coll Cardiol 2017; 70: 1704.
25. : "Association between diabetes mellitus and left ventricular hypertrophy in a multi-ethnic population". Am J Cardiol 2008; 101: 1787.
26. : "How much echo left ventricular hypertrophy would be missed in diabetics by applying the Losartan Intervention For Endpoint Reduction electrocardiogram criteria to select patients for angiotensin receptor blockade?". J Hypertens 2004; 22: 1403.
27. : "Screening for left ventricular hypertrophy in patients with type 2 diabetes mellitus in the community". Cardiovasc Diabetol 2011; 10: 29.
28. : "Diabetic cardiomyopathy". Am Heart J 1989; 118: 1000.
29. : "The prevalence of diabetic cardiomyopathy: a population-based study in Olmsted County, Minnesota". J Card Fail 2014; 20: 304.
30. : "Subendocardial systolic dysfunction in asymptomatic normotensive diabetic patients". Circ J 2015; 79: 1749.
31. : "Prognostic value of pulse-wave tissue Doppler parameters in patients with systolic heart failure". Am J Cardiol 2008; 102: 722.
32. : "Prognostic importance of tissue Doppler-derived diastolic function in patients presenting with acute coronary syndrome: a bedside echocardiographic study". Eur J Echo 2008; 9: 594.
33. : "Left ventricular diastolic function, assessed by echocardiography and tissue Doppler imaging, is a strong predictor of cardiovascular events, superior to global left ventricular longitudinal strain, in patients with type 2 diabetes". Eur Heart J Cardiovasc Imaging 2015; 16: 1000.
Abbreviations and Acronyms
| BSA | body surface area |
| DD | diastolic dysfunction |
| EF | ejection fraction |
| GLS | global longitudinal strain |
| HbA1c | glycosylated hemoglobin |
| HF | heart failure |
| HR | hazard ratio |
| LA | left atrial |
| LAE | left atrial enlargement |
| LV | left ventricular |
| LVD | left ventricular dysfunction |
| LVH | left ventricular hypertrophy |
| LVMi | left ventricular mass index |
| SBHF | stage B heart failure |
| T2DM | type 2 diabetes mellitus |
Footnotes
This study was partially supported by Tasmanian Community Fund and Diabetes Australia Research Trust. Neither of these agencies had any role in design, analysis, or interpretation of this study. The study was approved by the Tasmanian Human Research Ethics Committee. Dr. Marwick has received in kind support from GE Medical Systems for a trial of myocardial strain imaging for the assessment of cardiotoxicity from chemotherapy, unrelated to this study. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Paul Grayburn, MD, served as the Guest Editor for this paper.
