N-terminal probrain natriuretic peptide predicts 1-year mortality following acute stroke: possible evidence of occult cardiac dysfunction among patients with acute stroke

SIR—Brain natriuretic peptide (BNP) is a vasoactive peptide hormone synthesised in the cardiac atrium [1] and ventricular myocardium [2]. Increased levels of BNP are associated with various cardiovascular insults [3, 4]. In response to fluid overload, BNP is cleaved from its precursor proBNP resulting in the release of N-terminal (NT)-proBNP. NT-proBNP is an established diagnostic marker of sub-clinical heart failure [5] and predicts cardiovascular events and mortality among asymptomatic patients as well as in patients with coronary artery disease (CAD) [6] independent of conventional cardiovascular risk factors. In view of the high mortality and morbidity associated with acute stroke, recent studies have focused on non-neurological variables that may affect stroke outcomes [7–9]. Plasma BNP has been shown to predict the risk of stroke and atrial fibrillation (AF) [10]. In the acute phase of stroke, plasma levels of natriuretic peptide have been reported to be elevated [11] and correlate with risks of mortality [12–14]. Conversely, another study has concluded that NT-proBNP concentrations did not confer additional prognostic advantage over conventional cardiovascular risk factors in acute stroke patients [15]. The aim of this study was to investigate the role of NT-proBNP in predicting 1-year mortality of acute stroke in relation to various cardiovascular and biochemical parameters. We recruited consecutive patients of all ages with an acute ischaemic or haemorrhagic stroke, who were able to provide an informed written consent with mild and moderate level of dependency pre-admission on modified Rankin Scale [16] of <5. All patients were treated with similar protocol and care pathways by the stroke team and underwent chest radiography, electrocardiography (ECG) and computed tomography of brain and were examined for clinical stroke severity using Scandanavian Stroke Scale (SSS) and Oxford Community Stroke Project (OCSP) classification. Echocardiogram was not routinely performed as this was not within the study protocol. Patients were assessed for cardiovascular disease: existing heart failure on standard clinical and radiography features, history of CAD, current AF based on ECG findings and use of antihypertensive drugs. Assessors were blinded for diagnosis of stroke and past medical history. Serum creatinine was measured at the same time as blood samples were taken for NT-proBNP, on the same day as the detailed clinical assessment was performed. Blood samples were obtained during 9 AM to 12 PM within a mean of 3 days of stroke admission. NT-proBNP was measured using an automated assay on the Modular Analytical E170 supplied by Roche Diagnostics. This assay has an analytical sensitivity (zero signal + 2SD) of 5 ng/l, a functional sensitivity (CV < 20%) of <50 ng/l and a CV ranging from 3.2 to 2.3% at levels from 175 to 5000 ng/l. All samples were separated within 2 h of venepuncture, immediately frozen and analysed in batches and in duplicates for each sample. Outcome was measured as mortality at 1-year, death was verified from hospital notes or the patient administration system of the hospital. The Nottingham Research Ethics Committee approved the study. Data were analysed using software (SPSS, Version 13, SPSS Inc., Chicago, IL, USA). Univariate analysis using cross-tabulations for χ2 test was performed for categorical variables to assess the significance of association with death. Non-parametric tests were used to investigate the significance of the differences in median values. Relationship of log NT-proBNP was assessed against measures of clinical severity by a bivariate Pearson's correlation with SSS and the median and ranks of NT-proBNP in various OCSP stroke syndromes. Mann–Whitney U test was used to assess significance of difference for means and ranks where appropriate. Log NT-proBNP was used because of the skewed distribution of NT-proBNP. Cox regression was used to examine the effect of the biochemical variables and log proBNP for the primary outcome of death at 1 year. This is a pilot observational study, from which power calculation will be derived for future definite study. One hundred and twenty-five patients (63 women) in the age range 40–95 years (mean 73 ± 12) were recruited. At 1 year, 20 patients had died. Patients who died had a significantly higher concentration of NT-proBNP 980.2 ± 1249.9 versus 125.4 ± 244.9 pmol/l, P < 0.001, and NT proBNP log 6.0 ± 1.7 versus 3.5 ± 1.7 pmol/l, P < 0.001. Similarly, patients who died had a significant difference in the levels of urea 10.1 ± 4.6 versus 7.5 ± 4.4 mmol/l, P = 0.002; alkaline phosphatase 140 ± 116 versus 91.6 ± 45 U/l, P = 0.002; Barthel Index 2.2 ± 2.5 versus 7.1 ± 4.9, P < 0.001 and SSS 22 ± 14 versus 35 ± 12, P < 0.001. No difference was noted in mean creatinine, blood pressure, heart rate, estimated glomeruler filtration rate (eGFR), oxygen saturation, HbA1c, glucose, serum osmolality, cholesterol, LDL cholesterol, triglyceride and HDL cholesterol. There was no difference in the distribution of sex, AF, overt heart failure or ischaemic ECG. A total of 32 patients had AF at admission. Cox regression analysis in a forward stepwise method was performed for 1-year mortality using the variables of age, sex, creatinine, urea, ALT, alkaline phosphatase, SSS (at day 1) and log NT-proBNP. The most significant variable for mortality was the log NT-proBNP (Wald 17.9, P < 0.0001) followed by SSS (Wald 5.7, P = 0.02). Survival curve was plotted for patients surviving or dying at 1 year with NT-proBNP dichotomised at the median value of 42 pmol/l (Figure 1). (log-rank Mantel–Cox P < 0.001). Of the 57 patients with NT-proBNP above 42 pmol/l, 15 patients had died at 1 year when compared with only one of the 57 patients dying in the group of patients with the value less than 42 pmol/l, P < 0.001. Kaplan–Meier plot of patients according to NT-proBNP dichotmised at a median value of 42 pmol/l, censored at1 year. The present findings indicate that in the acute phase of stroke, elevated values of NT-proBNP are the most important independent predictor of mortality at 1-year after an acute stroke among previously independent patients when adjusted for other conventional biochemical and cardiovascular risk factors. This finding raises the possible likelihood of the presence of occult and/or acute cardiac impairment after an acute stroke. Our findings are in concordance with previous studies reporting increased short and longer-term mortality in patients with higher concentrations of NT-proBNP concentrations [12–14, 17–19]. The latest study to show this measured BNP at 6 months after index admission rather than during acute admission. The mechanism of increased plasma concentrations of NT-proBNP with mortality following an acute stroke is unclear. Increased concentration of NT-proBNP in the setting of an acute stroke may indicate myocardial injury and functional cardiac impairment as a consequence of cerebral damage due to activation of renin–angiotensin–aldosterone system-induced vasoconstriction and fluid retention. Since heart failure is associated with higher mortality after a stroke [7], high NT-proBNP may simply be an index of occult cardiac impairment. Importantly, evidence of cardiac damage in acute stroke patients have been supported by studies showing increased levels of troponin [8, 9] associated with acute stroke and that the levels of NT-proBNP in patients with acute stroke is similar to levels in patients with acute myocardial infarction [12]. Another plausible explanation is that NT-proBNP may be released directly from the cerebral tissue. Immunohistochemical studies suggest that cerebral ischaemia induces natriuretic secretion in brain tissue [20] and gradually increasing BNP levels have been reported with subarachnoid haemorrhage [21]. High concentrations of plasma BNP have been associated with the development of cerebral ischaemia and neurological deficits [23]. In the presence of pre-existing occult cardiac dysfunction, these patients are predisposed to more severe strokes and increased mortality. Several limitations of this study must be acknowledged. We have not objectively measured left ventricular function with more sensitive modes of imaging. Unlike previous studies in acute stroke, we have included patients with both ischaemic and haemorrhagic stroke. The number of the latter, however, was very small and there was no significant difference in the mean values of NT-proBNP and the outcomes of both types of stroke. Lastly, despite the higher number of patients in this study compared with a previous similar study, the relatively small sample size may still limit the generalisation of the results of this study. However, the extensive list of clinical and biochemical variables that were added to the multivariate analysis, unlike in any previously reported study have aided to enhance the independent prognostic power of NT-proBNP to predict mortality as a biochemical measure in acute-phase stroke. In summary, this study showed that elevated plasma levels of NT-proBNP during the acute phase is the most important independent predictor of mortality at 1 year. Since NT-proBNP is a sensitive marker of occult heart failure, it is tempting to hypothesise the rationale for the use of cardio-protective drugs during acute stroke to improve clinical outcomes. Further adequately powered interventional randomised controlled trial is required to answer this research hypothesis to protect myocardium from strain in acute stroke patients. BNP predicts mortality in stroke. Possible presence of subclinical heart failure in patients with stroke. Potential use of cardioprotective drugs acutely during stroke to reduce mortality. None.