Prognostic value of systolic short-term blood pressure variability in systolic heart failure

Background Traditional cardiovascular risk factors in the general population are usually correlated to a better prognosis in patients with chronic heart failure (HF). Most of the studies show that blood pressure variability (BPV) has noxious effect on general population but data are missing for patients with systolic HF. The aim of this study was to assess the prognostic impact of short-term blood pressure variability (BPV) in systolic HF. Methods and results We retrospectively studied 288 patients (60 ± 12 years-old; 79 % male) referred to our tertiary center of HF for the management of their systolic HF (left ventricular ejection fraction was 28 ± 9 %). All patients underwent ambulatory blood pressure monitoring (systolic BP: 110 ± 15; diastolic BP: 68 ± 10 and pulse pressure: 42 ± 11 mmHg) and the prognostic impact of BPV was collected with a mean follow-up of 4.4 ± 3.1 years. Twenty-five (9 %) patients were missing for follow-up. Among the others patients, 70 (27 %) cardiovascular events (cardiac deaths: 24 %; heart transplantation: 2 %) were recorded. By multivariate analysis BPV daytime (OR = 0.963, p = 0.033) and severe NYHA class (OR = 5.2, p < 0.0001) were found as independent predictors of cardiac event. Patients with a systolic daytime BPV under a cut-off value of 19 mmHg had the poorest prognosis with an OR for cumulative events of 1.65 (IC95 % 1.1–2.7; p < 0.04). Conclusion BPV is simple tool and a predictor of cardiac events in patients with systolic HF.


Background
If high blood pressure (BP), body mass index and cholesterolemia represent traditional cardiovascular risk factors in the general population, they are correlated to a better prognosis in patients with chronic heart failure (HF) [1][2][3]. In a meta-analysis, Raphael and al emphasized the paradoxical effect of higher systolic BP on mortality of patients with chronic HF, showing a decrease of 13 % in cardiovascular death for an increase of 10 mmHg in systolic BP [1]. For the last decades, the prognostic impact of each determinant of BP profile such as systolic BP, diastolic BP, pulse pressure (PP), BP variability (BPV) was essentially studied in patients with hypertension [4], but few in chronic HF. Thus, Rothwell et al. showed in a hypertensive population that the daytime BPV was a powerful predictor of stroke and coronary events [5]. The aim of the present study was to assess the prognostic impact of short-term BPV in chronic systolic HF.

Population
Patients were retrospectively extracted from our local database of HF and including patients referred the exploration and the management of systolic HF in of the HF unit of the University Hospital of Toulouse from 1999 to 2006. Inclusion criterions were: age over 18 years old, one systolic HF event in life, systolic dysfunction defined by left ventricular ejection fraction <45 %, ambulatory monitoring of BP in at admission. Exclusion criterions were: patients with low flow or treated with intravenous drugs such as inotropic support, infection, dialysis and incomplete ambulatory monitoring of BP.
The study was approved by our local ethics committee.

Twenty-four-hour ABPM
Twenty-four-hour ambulatory BP monitoring (ABPM) was performed as previous described in chronic HF [6], using the oscillometric method (Spacelabs 90207 device® [7]). Successive BP measures were performed every 15 min during daytime (7 am. to 9:59 pm.) and every 30 min during nighttime (10 pm. to 6:59 am.). BP measures were expressed in millimetres of mercury (mmHg). All devices for ABPM were placed on the right arms by a trained nurse 24-h after admission. Patients were instructed to relax the cuffed arm during the measure and received a diary to record unexpected events. The analysis of ABPM records were performed using Spacelabs software allowing us to extract systolic BP, diastolic BP, PP, BPV and Dip ratio daytime and nighttime BPV was calculated using the average difference between maxima and minima from each systolic BP measure to another. Nighttime BP dipping can be quantified as the ratio of mean nighttime (sleep) BP to mean daytime (awake) BP. The calculation formula was: BPV = (maximum systolic BPminimum systolic BP)/2.

Follow-up
Follow up was conducted using physician, patient or family phone contacts. Patients without news within the last month after the AMBP were considered as missing for follow-up. The composite endpoint was defined by the occurrence of cardiovascular events: cardiac death or heart transplantation.

Statistical analysis
Continuous variables with a normal distribution are expressed as mean ± standard deviation (SD). To compare numerical data between two groups, paired and unpaired Student t-test was used when appropriate. Nominal variables were compared using either the χ 2 or Fischer tests when appropriate. Univariate analysis of the predictive factors of cardiovascular event was performed using respectively log rank and cox methods for qualitative and quantitative variables. Kaplan Meier curve and log mantel cox method were used to illustrate prognosis of high and low level of systolic BP and VBP. Variables from the univariate analysis with P < 0.1 were included in the multivariable stepwise analysis to identify independent predictors of events. Two-tailed P-values <0.05 were considered statistically significant.

Follow-up and events
Twenty-five (9 %) patients were missing for follow-up.

Univariate predictors of cardiac event
Results of univariate analysis are summarized in Table 4. By univariate analysis, discriminative-predictive variables were NYHA stage III/IV (OR = 5.  (Table 4B.) The Fig. 1 illustrates the predictive positive impact of a daytime SBP ≤ 110 mmHg on cumulative survival.

Multivariate analysis
Daytime BPV, daytime SBP, nighttime SBP, daytime and nighttime PP for ABPM variables, and age, body mass index, diabetes and decompensate HF for clinical variables were included in stepwise multivariable cox model.
Forward and backward stepwise allowed to characterize BPV daytime (OR = 0.963, p = 0.033) and severe NYHA class (OR = 5.2, p < 0.0001) as two independent predictors of cardiac events. As shown in Fig. 2, a systolic daytime BPV under a cut-off value of 19 mmHg had a poor prognostic impact with an OR for cumulative events of  1.65 (IC95 % 1.1-2.7; p < 0.04). The association of NYHA and BPV gave a better prognosis (Fig. 3).

Discussion
Low daytime BPV (<19 mmHg) is found as an independent factor of cardiac death and/or cardiac transplantation. The daytime level of BPV, as systolic BP level, is correlated with the prognosis of patients with systolic HF. Previous study of Gibelin et al. [8] showed a worse prognosis of a low BPV in 154 patients with CHF. Our study confirms this result with a biggest recruitment during a mean follow up of 4.4 years. Thus, in comparison to previous studies on the prognostic value of BPV in hypertension, we have demonstrated a paradoxical prognostic impact of BP variability on cardiovascular event occurrence in systolic HF (cardiac death and heart transplantation).

Determinant factors of BPV in CHF
Decrease of BPV in patients with CHF is well described.
Radaelli et al. [9] demonstrated an alteration of BPV in nine patients with congestive CHF during exercise with a BPV of 15 and 41 mmHg 2 in patients with CHF and healthy control subjects respectively [2]. The question that remained is: what mechanism(s) underlie(s) the decrease of BPV in CHF? If components involved in BP as cardiac output or systemic vascular resistance are well identified in patients with CHF, determinant factors of its variability are unclear. Intrinsic factors such as baroreflex sensibility [10], autonomic nervous system integrity [11], pre-load volume, myocardial contractile reserve, circadian hormonal secretion [12] are potential pathways that could participate to BPV. Also, extrinsic factors such as daily activities, stress and smoking must  Paradoxical prognostic of BPV: one more Paradigm in CHF For the last three decades, several studies demonstrated that usual cardiovascular risk factors become good prognosis markers in patients with CHF. Thus, high body mass index [13], hypercholesterolemia or high BP are associated with survival in a CHF population [14]. If BPV is added to lots of complex physiological adaptive system to survive [10], a high level of it worse the prognostic of hypertensive patients [15] whereas a very low level worse the prognosis of CHF. That is one more paradigm in CHF and BPV must be took into account among the reverse epidemiology of conventional risk factors in CHF [16].

Conclusion
BPV is simple tool, easily available, predictor of cardiac death and/or cardiac transplantation in patients with systolic HF. Measure of BPV could help physicians to improve the management of patients with HF by helping to better assess the prognosis and adjust the therapeutics.