Cognitive Impairment in Asian Patients with Heart Failure: Prevalence, Biomarkers, Clinical Correlates, and Outcomes

Cognitive impairment (CI) is prevalent in heart failure (HF) patients in western populations (25–74%) but such information among Asian populations is scant.1 HF patients with impairment in multiple cognitive domains have higher risk of developing dementia.2 CI links to poor self-care, leading to poorer prognosis and increased mortality.3 The 2016 European Society of Cardiology (ESC) guidelines highlight the importance of customized management for cognitively impaired HF patients to improve self-care.4 Amino-terminal pro-brain natriuretic peptide (NT-proBNP) and high-sensitivity cardiac troponin T (hs-cTnT) are diagnostic markers for HF and acute myocardial infarction, respectively. They predict mortality in HF patients5 and associate with CI in older adults.6 Given the limited data on CI in Asian HF patients, we examined the prevalence of CI in Singaporean HF patients and explored correlations between CI and biomarkers, clinical factors, quality of life (QoL), and outcomes. One hundred patients (> 18 years old) with diagnosis of HF (according to ESC criteria) were prospectively and consecutively recruited in compensated chronic HF status from Singapore hospitals.7 All patients were assessed by trained research personnel using the Mini-Mental State Examination (MMSE), Montreal Cognitive Assessment (MoCA), and a locally validated formal neuropsychological test battery covering seven domains (Attention, Language, Verbal and Visual Memory, Visuoconstruction, Visuomotor Speed, Executive Function). Since brief screening tests are unable to diagnose CI, a comprehensive formal neuropsychological test battery was used for diagnosis. Education-adjusted cut-offs of 1.5 standard deviation below the established norms were used on individual tests. Failure in at least half of the tests in a domain constituted failure. CI is defined by impairment in at least one domain. Clinical factors included left ventricular ejection fraction (LVEF) and cardiovascular risk factors such as diabetes mellitus (DM), hypertension, coronary artery disease (CAD), atrial fibrillation (AF), smoking, stroke, and chronic kidney disease (CKD). Plasma for NT-proBNP and hs-cTnT assays was collected at baseline. Health outcomes and QoL (Minnesota Living with Heart Failure Questionnaire) were collected at 1 year. Continuous variables were expressed as mean (± standard deviation) and categorical variables as percentages. Structural equation modelling (SEM) analysis adjusted for demographics examined the association between CI, biomarkers, clinical factors, QoL, hospitalization for HF, and death. Receiver operating characteristic (ROC) analysis examined the discriminatory ability of MMSE and MoCA for CI determined by formal neuropsychological tests. Chi-squared automatic interaction detector algorithm identified the optimal cut-points for quantitative predictors. This study complied with the Declaration of Helsinki, received ethics approval, and informed consent was obtained from all participants. Among 100 Asian patients with chronic HF [age: 58.68 ± 10.53; female: 15%; hypertension: 65%, DM: 54%, CAD: 62%, AF: 20%; CKD: 42%; prior stroke: 14%; reduced LVEF (< 50%): 65%], 44 of 100 HF patients had undiagnosed CI (50% with multiple domain impairment). Four participants were excluded due to outliers in NT-proBNP or hs-cTnT level (n = 2) or missing biomarker data (n = 2). HF patients with CI were older and less educated than those without CI (age: 64.3 ± 8.3 vs. 54.2 ± 10.1, P < 0.0001; education: 7.98 ± 4.2 vs. 10.0 ± 3.4 years, P = 0.010) (Table 1). Cognitively impaired patients had normal MMSE (25.2 ± 3.4) but low MoCA scores (21.3 ± 4.6). Patients with CI more frequently had ischaemic HF and more cardiovascular risk factors than those without CI (ischaemic aetiology: 71.4% vs. 44.4%, P = 0.004; total cardiovascular risk factors: 3.48 ± 1.7 vs. 2.76 ± 1.5, P = 0.030). Their neuropsychological impairment was characteristic of vascular pathology with frequently impaired Visuomotor Speed (information processing speed) (60%), Visuoconstruction (construction of geometric figures) (48%), and Visual Memory (recall of visual information) (43%). SEM analysis showed that age strongly predicts CI (β = 1.113, P = 0.002). Older patients (> 63 years old) had six times greater risk of having CI per additional year (β = 6.086, P = 0.003). In HF patients with stroke (n = 14), 64% (n = 9) had CI (six with multiple domain impairment). Since stroke strongly associates with CI, those with documented stroke were excluded to elucidate cognition correlations with HF. Excluding age and patients with prior stroke (n = 14), NT-proBNP was independently associated with CI, hospitalization, and QoL (CI: β = 1.660, P = 0.026; hospitalization: β = 2.538, P = 0.010; QoL: β = 8.412, P < 0.001). By ROC analyses in this subset, MMSE and MoCA were similar in detecting CI [area under the curve (95% confidence interval): 0.740 (0.641–0.840)/0.770 (0.675–0.866), P = 0.481], with optimal cut-off points of < 28 for MMSE and < 25 for MoCA (sensitivity 0.79/0.71; specificity 0.63/0.61; positive predictive value 0.62/0.59; negative predictive value 0.79/0.73; classification accuracy 69.8%/65.6%, respectively). In this pilot study, the prevalence of undiagnosed CI in Asian HF patients was high (44%). Older patients have greater risk of CI. Excluding age and documented stroke, NT-proBNP predicted CI, hospitalization, and QoL. The high rate of undiagnosed CI is concerning as CI will affect HF self-care given medication complexity, numerous lifestyle modifications, and recognition of symptoms.3 Using published MMSE cut-point (< 24), 74% of cognitively impaired patients would be undetected without formal neuropsychological evaluation. Optimal cut-points of MMSE < 28 and MoCA < 25 in our study are consistent with previous study.8 NT-proBNP is associated with CI in cerebrovascular diseases.6 Here, NT-proBNP was associated with CI even without documented stroke, suggesting that HF per se may be related to CI. Since there is no guideline for cognitive screening in HF,1 we propose that HF patients without prior stroke > 63 years old, with MMSE < 28 or MoCA < 25, or high NT-proBNP, may be considered high risk and be formally evaluated for CI. Several limitations require acknowledgement. First, a small sample size limits the generalizability of our findings. A larger study is needed to validate the suggested cut-offs for detecting CI in HF patients. Secondly, the lack of a healthy control group limits our conclusion on whether CI is associated with HF per se. In conclusion, the prevalence of undiagnosed CI in Asian HF patients is high. HF patients with high levels of NT-proBNP may be at risk of developing CI. Therefore, HF patients without prior stroke > 63 years old, with MMSE < 28 or MoCA < 25, or high NT-proBNP, may warrant formal evaluation for CI. This work was supported by National Medical Research Council (NMRC) Research Training Fellowship Seed Funding [NMRC/SEEDFUNDING/014/2015] and NMRC Transition Award [NMRC/TA/0060/2017]. Conflict of interest: none declared.