Opportunistic screening for asymptomatic left ventricular dysfunction in type 2 diabetes mellitus

Background International guidelines recommend natriuretic peptide biomarker-based screening for patients at high heart failure (HF) risk to allow early detection. There have been few reports about the incorporation of screening procedure to existing clinical practice. Objective To implement screening of left ventricular dysfunction in patients with type 2 diabetes mellitus (DM). Method A prospective screening study at the DM complication screening centre was performed. Results Between 2018 and 2019, 1043 patients (age: 63.7±12.4 years; male: 56.3%) with mean glycated haemoglobin of 7.25%±1.34% were recruited. 81.8% patients had concomitant hypertension, 31.1% had coronary artery disease, 8.0% had previous stroke, 5.5% had peripheral artery disease and 30.7% had chronic kidney disease (CKD) stages 3–5. 43 patients (4.1%) had an elevated N-terminal prohormone of brain natriuretic peptide (NT-proBNP) concentration above the age-specific diagnostic thresholds for HF, and 43 patients (4.1%) had newly detected atrial fibrillation (AF). The prevalence of elevated NT-proBNP increased with age from 0.85% in patients aged <50 years to 7.14% in those aged 70–79 years and worsening kidney function from 0.43% in patients with CKD stage 1 to 42.86% in CKD stage 5. In multivariate logistic regression, male gender (OR: 3.67 (1.47–9.16), p=0.005*), prior stroke (OR: 3.26 (1.38–7.69), p=0.007*), CKD (p<0.001*) and newly detected AF (OR: 7.02 (2.65–18.57), p<0.001*) were significantly associated with elevated NT-proBNP. Among patients with elevated NT-proBNP, their mean left ventricular ejection fraction (LVEF) was 51.4%±14.7%, and 45% patients had an LVEF <50%. Conclusion NT-proBNP and ECG screening could be implemented with relative ease to facilitate early detection of cardiovascular complication and improve long-term outcomes.


INTRODUCTION
Heart failure (HF) is a cardiovascular disease epidemic affecting 26 million people worldwide. 1 The incidence and prevalence of HF appear to be on a rising trend, particularly in developing countries due to the rapidly ageing population and the high prevalence of cardiovascular risk factors. In Asian countries, the prevalence of HF reportedly ranges between 1.3% and 6.7%. 2 In China, there are 4.2 million people living with HF. 3 4 Over the past few decades, advances in pharmacological therapy, including ACE inhibitors, 5 6 angiotensin II receptor blockers, beta-adrenergic blockers, [7][8][9] mineralocorticoid receptor antagonists, 10 11 ivabradine 12 and more recently sacubitril/valsartan, 13 as well as various device therapies, have been shown in randomised controlled trials to significantly improve patient outcomes particularly in those with reduced left ventricular ejection fraction (LVEF). Nonetheless, due to the progressive nature of the condition, the prognosis of HF in the real-world practice remains poor. In a recent European study, the 1-year mortality rate in patients with newly diagnosed HF was as high as 16.4% 14 ; likewise, the reported 1-year mortality in Asia ranged between 8.9% and 19.5%. [15][16][17] Therefore, prevention of HF has turned into the priority of HF management at the public health level.
Diabetes mellitus (DM) is a well-recognised risk factor of cardiovascular diseases and HF. 18 Patients with DM have a twofold-fourfold higher risk of HF compared with those without, 19 and up to 27.7% of patients with DM had concomitant HF. [20][21][22][23][24] In 2001, the American College of Cardiology (ACC) and the American Heart Association (AHA) introduced a new HF classification system, including individuals who have not had clinical HF but are at high risk of developing HF due to either concomitant comorbidities such as DM (stage A), or underlying cardiac structural abnormalities (stage B). 25 The initial intention to include these so-called patients with 'pre-HF' into the classification is merely to help healthcare providers with the early identification of patients who are at risk of developing HF. Recently, sodium-glucose cotransporter-2 (SGLT2) inhibitors have been shown in pivotal studies to substantially reduce hospitalisation of HF and/or cardiovascular mortality in patients with DM, even among those without pre-existing HF or established cardiovascular disease, 26 27 providing a genuine therapeutic option to prevent HF. Early identification of patients with DM at high risk of developing HF before the occurrence of clinical HF, followed by aggressive preventive measure, appears to be a logical strategy. Nonetheless, among patients with newly diagnosed HF, the prevalence of DM was as high as 36%, 17 precluding any form of primary preventive measure. In fact, the latest ACC/AHA/HFSA guideline for the management of HF has recommended natriuretic peptide biomarker-based screening for patients at Original research risk of developing HF (class IIa, B-R), 28 which could help early identification of individuals at high risk to prevent the development of HF. Despite this, clinical literatures describing the incorporation of N-terminal prohormone of brain natriuretic peptide (NT-proBNP) measurement as a screening tool to routine DM complication screening programme are scarce. In the present study, we explored the possibility of incorporating NT-proBNP measurement into the DM complication screening programme in order to identify individuals at high risk of developing HF.

Study design and patients
This prospective screening study was coordinated by the Division of Cardiology, and the Division of Endocrinology and Metabolism, the Department of Medicine, The University of Hong Kong. All participants were recruited from the Diabetic Complication Screening Centre, Queen Mary Hospital. Patients with type 2 DM scheduled for their regular diabetic complication screening were invited to participate in the study. Written informed consent was obtained from all recruited participants. Patients were excluded if they were under 18 years of age, had history of HF and/or atrial fibrillation (AF). Prior HF was defined as HF stages B, C or D as defined in the 2001 ACC/AHA guidelines, including patients with underlying structural abnormality with or without symptoms. 5 Patients attended their regular diabetic complication screening visit every 1.5-2 years. At the first visit, demographic data and detailed medical and drug histories were obtained using a standardised questionnaire and the territory-wide computerised medical system. During each visit, anthropometric parameters, including body weight, height, body mass index and blood pressure, were measured. After an overnight fast of at least 8 hours, fasting blood was drawn for plasma glucose, lipids and glycated haemoglobin (HbA1c). Serum creatinine was measured and estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation as described previously. 29 Albuminuria status was assessed using at least two random urine samples. All patients, except those who were being followed up regularly by ophthalmologists at the public sector, had retinal photographs captured and graded systemically by trained optometrists and ophthalmologists, according to the United Kingdom National Screening Committee guidelines. 30 In this study, individuals who agreed to participate, after their written informed consent, would have serum NT-proBNP assayed (Roche Diagnostics, GmbH), and a frontal plane (6-lead) 10-second ECG performed at a seated position for the detection of AF. Patients with an elevated serum NT-proBNP concentration above the age-specific diagnostic threshold of ≥450 pg/mL in age <50 years, ≥900 pg/mL in age 50-75 years and ≥1800 pg/ mL in age >75 years were invited for a transthoracic echocardiography within 1 week. Echocardiography was performed using Vivid i Cardiovascular Ultrasound Systems (GE Medical Systems, Israel) and included two-dimensional, M-mode, and colour Doppler images according to the guidelines of American Society of Echocardiography. 16 17 The average of three cardiac cycles was used for measurements of cardiac dimensions and LVEF. 31 32 Two independent cardiologists reviewed all the ECG tracings recorded to provide a rhythm diagnosis using standard criteria. 33

Statistical analysis
Continuous and discrete variables were expressed as mean±SD and percentages, respectively. Chi-square test or Fisher's exact test was used to compare categorical variables between groups. Student's t-test was performed to compare continuous variables. ORs with corresponding 95% CIs of each variable to predict elevated NT-proBNP concentration and AF were determined by univariate logistic regression and with multivariate logistic regression model for variable with p value of <0.1. All statistical analyses were performed using the SPSS software V.21.0 for Mac.

RESULTS
Between November 2018 and October 2019, 1132 patients who fulfilled the inclusion criteria of the present study were invited to participate, of which 30 patients declined. The response rate was 97.3%. In addition, 59 patients consented for the study (5.2%) did not complete all screening procedures and were excluded from the final analysis. As a result, 1043 patients with type 2 DM were included in this final analysis (figure 1). Table 1 summarises the characteristics of the study population. The mean age was 63.7±12.4 years; 587 patients (56.3%) were male. The mean body mass index was 26.0±4.6 kg/m 2 , and the mean HbA1c was 7.25%±1.34%. 846 patients (81.8%) had concomitant hypertension, and 337 patients (32.3%) had positive smoking history. In addition, 477 patients (45.7%) had diabetic retinopathy, 350 patients (33.6%) had peripheral neuropathy and 320 patients (30.7%) had CKD stages 3-5. For macrovascular complications, there were 324 patients (31.1%) with coronary artery disease, 83 patients (8.0%) with previous stroke and 57 patients (5.5%) with documented peripheral artery disease.

NT-proBNP and ECG screening
Forty-three patients (4.1%) had an elevated NT-proBNP concentration above the age-specific diagnostic threshold for HF during the screening procedure. At the same time, 43 patients (4.1%) were newly detected to have AF. Ten patients (1.0%) had both an elevated NT-proBNP concentration and newly detected AF. The prevalence of elevated NT-proBNP concentration increased with age from 0.85% in patients aged <50 years to 7.14% in those aged 70-79 years and with worsening kidney function from 0.43% in patients with CKD stage 1 to 42.86% in CKD  Table 2 summarises the prevalence of elevated NT-proBNP concentration in these 1043 patients stratified according to their age and CKD stage. Likewise, the prevalence of newly detected AF increased with age, as well as worsening kidney function (figure 3). Table 3 summarises factors predictive of an elevated NT-proBNP concentration above the age-specific diagnostic threshold of HF in patients with type 2 DM, together with their corresponding ORs based on logistic regression and 95% CIs. On univariate analysis, increasing age, male gender, hypertension, coronary artery disease, prior stroke, peripheral arterial disease, proteinuria, CKD, peripheral neuropathy and newly detected AF were associated the increasing risk of elevated NT-proBNP concentration. On multivariate analysis, only male gender (OR: 3.67 (1.47-9.16), p=0.005*), prior stroke (OR: 3.26 (1.38-7.69), p=0.007*), CKD (p<0.001*) and newly detected AF (OR: 7.02 (2.65-18.57), p<0.001*) remained significantly associated with elevated NT-proBNP concentration (table 3). All patients with an elevated NT-proBNP concentration above the diagnostic threshold of HF were invited for transthoracic echocardiogram, of which 42 patients participated and 1 patient refused (table 4). The mean LVEF was 51.4%±14.7%; 19 out of 42 patients (45%) had an LVEF below 50%, 13 patients with LVEF >50% (31.0%) had left ventricular hypertrophy and diastolic dysfunction, and 1 patient had moderate-to-severe mitral regurgitation. Twelve patients with LVEF below 50% were in sinus rhythm and 7 were in AF. Table 5 summarises factors predictive of newly detected AF in patients with type 2 DM, together with their corresponding ORs based on logistic regression and 95% CIs. On univariate analysis, increasing age, male gender, coronary artery disease, prior BMI, body mass index; CAD, coronary artery disease; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; Hb, haemoglobin; HbA1c, glycated haemoglobin; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; PAD, peripheral artery disease.

Figure 2
Prevalence of elevated NT-proBNP concentration above the diagnostic threshold for HF across (1) age groups and (2) chronic kidney disease stages. HF, heart failure; NT-proBNP, N-terminal prohormone of brain natriuretic peptide. CKD, chronic kidney disease; DM, diabetes mellitus; NT-proBNP, N-terminal prohormone of brain natriuretic peptide.

Original research
stroke and CKD were associated with newly detected AF. On multivariate analysis, only increasing age, male gender and prior stroke remained significantly associated with newly detected AF (table 4). Non-vitamin K oral anticoagulants and warfarin were initiated in 39 patients (90.7%) and 4 patients (9.3%), respectively.

DISCUSSION
In the present study, we incorporated NT-proBNP measurement together with ECG screening into the DM complication screening programme. We identified the prevalence and predictors of elevated NT-proBNP concentration, which is indicative of a higher risk of developing HF and, therefore, could inform further clinical management. First, in our cohort of 1043 consecutive patients with DM without prior history of HF or AF, we showed that the prevalence of elevated NT-proBNP concentration, defined as above the age-specific diagnostic threshold  were associated with elevated NT-proBNP concentration. Third, among patients with DM with elevated NT-proBNP concentration, 45% of them had an LVEF below 50%, and 31% of them had diastolic dysfunction with left ventricular hypertrophy. Last but not least, AF was newly detected in 4.1% of the current study population. DM is associated with macrovascular and microvascular complications that can go unnoticed and be left untreated until irreversible damage ensues. The American Diabetes Association has recommended a comprehensive list of medical evaluations to detect complications, which has been widely incorporated into DM complication screening programmes worldwide. 34 On the other hand, despite the high prevalence of HF (~28%) among patients with DM, 20-24 early detection or screening of HF and/or asymptomatic left ventricular dysfunction has not been recommended or routinely practiced. This may be partly related to the difficulty to detect HF particularly non-acute HF in primary care setting. 35 Plasma natriuretic peptides, including NT-proBNP and BNP, which have high positive predictive value for HF, are established tests for HF diagnosis, [36][37][38] as well as prognostication. [39][40][41] More recently, the use of plasma natriuretic peptides has been extended for screening of early HF and/or asymptomatic left ventricular dysfunction among patients at risk of developing HF. 28 42-44 The screening yield depends on the prevalence of HF and/or asymptomatic left ventricular dysfunction in specific screened population. In the present study involving patients with DM but without prior history of HF and AF, the overall prevalence of elevated NT-proBNP concentration was relatively high at 4.1%, compared with ~1% prevalence of HF in the general population. Although NT-proBNP assay has a high

Original research
positive predictive value to detect early HF and/or asymptomatic left ventricular dysfunction, it is not widely available due to cost. Nonetheless, the prevalence increased disproportionally with age and CKD stage, two independent predictors of elevated NT-proBNP concentration and asymptomatic left ventricular dysfunction in patients with DM. For instance, while the prevalence of elevated NT-proBNP concentration in patients with DM aged <70 years and with CKD stage 1 is negligible and may not justify screening, for those with CKD stages 3-5, the prevalence of elevated NT-proBNP concentration could range from 9.5% to 14.3%, giving a very high screening yield. Therefore, an age-based and kidney function-based screening would be a more efficient strategy to identify patients with early HF or asymptomatic left ventricular dysfunction. Newly detected AF was associated with elevated NT-proBNP concentration. Although the ventricular myocardium is the predominant source of NT-proBNP, the atrial myocardium also expressed NT-proBNP and mRNA expression level is increased in patients with AF. 45 It has been hypothesised that in patients without HF, the increased NT-proBNP expression in atrial myocardium could contribute to the increased plasma concentration. 46 As the purpose to detect early HF and/or asymptomatic left ventricular dysfunction is to prevent HF, there should be a therapeutic intervention effective at this early phase to modify the disease course. In the STOP-HF trial involving 1374 patients aged 40 years or older with at least one high-risk factor for HF but without HF and/or asymptomatic left ventricular dysfunction, 42 a BNP-based screening programme, together with the collaborative care between primary care physician and cardiovascular specialist, reduced the combined rates of left ventricular dysfunction and HF, which confirmed the role of plasma natriuretic peptides as a HF risk identifier. Furthermore, in the PONTIAC study, 43 accelerated up-titration of renin-angiotensin system antagonists and beta-adrenergic blockers among patients with DM with an elevated NT-proBNP concentration reduced hospitalisation or death due to cardiac disease by 64% in 2 years. More recently, in EMPA-REG, 26 and CANVAS trials that randomised patients with type 2 DM and history of cardiovascular disease to SGLT2 inhibitors (empagliflozin and canagliflozin, respectively) or placebo, SGLT2 inhibitors substantially reduced cardiovascular mortality and hospitalisation for HF. Furthermore, subgroup analyses of these landmark randomised controlled trials revealed that SGLT2 inhibitors reduced HF hospitalisation regardless of the presence of documented HF at baseline, suggesting the possibility of using SGLT2 inhibitors to improve HF outcome in patients with DM with subclinical HF. 47 48 These further strengthens the armamentarium to prevent DM-related HF.

Limitations
There are several limitations in our study. First, echocardiography was performed only in patients with elevated NT-proBNP concentration instead of all patients. Therefore, the prevalence of asymptomatic left ventricular dysfunction with normal NT-proBNP concentration could not be determined as a small proportion of patients with HF may have NT-proBNP concentration below the age-specific diagnostic threshold. 49 Second, screening for AF was performed using a single 12-lead ECG instead of Holter monitoring. Use of the latter strategy will increase sensitivity of detecting asymptomatic paroxysmal AF. Third, since the current study is a single-arm observational study instead of a randomised controlled trial, it remains uncertain whether incorporation of NT-proBNP measurement and ECG to DM complication screening programme could reduce the risks of incident HF, hospitalisation for HF and/or mortality. Nonetheless, identification of these at-risk patients by systematic screening is the critical first step that would allow further prospective assessment of whether early intervention is associated with improved clinical outcome. Fourth, the number of patients with elevated NT-proBNP concentration was relatively small in this study, which could affect generalisability of the derived results.

CONCLUSION
In a tertiary centre of patients with type 2 DM, the prevalence of elevated NT-proBNP concentration and AF were both 4.1%. As the prevalence of raised NT-proBNP increased with age and CKD stage, targeted screening in these patient subgroups may result in a higher yield than universal screening. Identification of patients with DM at risk of developing HF by systematic screening may allow early intervention to improve clinical outcome.

Current research questions
► Is incorporating serum N-terminal prohormone of brain natriuretic peptide measurement into the diabetes mellitus complication screening programme effective in identifying patients with early heart failure?
What is already known on the subject? ► Patients with diabetes mellitus (DM) have a higher risk of developing heart failure (HF) than those without DM. ► Plasma natriuretic peptides are useful in screening early HF and/or asymptomatic left ventricular dysfunction among patients at risk of developing HF.

Main messages
► In a tertiary centre of patients with type 2 DM, the prevalence of elevated NT-proBNP concentration was 4.1%. ► Among patients with elevated N-terminal prohormone of brain natriuretic peptide (NT-proBNP) concentration, their mean left ventricular ejection fraction (LVEF) was 51.4%±14.7%, and 45% patients had an LVEF below 50%. ► Prevalence of raised NT-proBNP increased with age and chronic kidney disease (CKD) stage. Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Consent obtained directly from patient(s).
Ethics approval This study involves human participants and was the study protocol was approved by the institutional review board of The University of Hong Kong/ Hospital Authority Hong Kong West Cluster (UW14-380/07-378). Participants gave informed consent to participate in the study before taking part.
Provenance and peer review Not commissioned; internally peer reviewed.