Hyponatremia is a common complication in heart failure (HF), especially when the disease is advanced. Studies in various patient populations, ranging from acutely decompensated hospitalized patients to stable outpatients, have shown an association between low serum sodium concentrations and poor prognosis.1–3 Equally, risk models developed in HF cohorts of outpatients4 and hospitalized patients5,6 have suggested that hyponatremia is one of the most powerful predictors of mortality. Blood pressure (BP) is another predictor of survival,7 and lower sodium concentrations are often associated with low BP. However, this association has not been thoroughly studied and is poorly understood, due in part to the lack of studies that provide dynamic 24-hour BP measurement.

The influence of hyponatremia on BP in HF patients may be the result of changes in both plasma sodium levels and extracellular volume (ECV), but may also be secondary to antihypertensive drugs used to treat HF. It may therefore be useful to know whether BP remains low at night or whether these patients’ circadian profile is disrupted.

Our aim in this study was to determine whether there are differences in BP behavior on dynamic assessment by ambulatory blood pressure monitoring (ABPM) in stable HF patients with different serum sodium levels.

A total of 186 patients (44.57% male, mean age 76.88±8.16 years) were included. Their baseline characteristics according to group of serum sodium levels are shown in Table 1. eGFR and blood glucose levels were slightly higher in Group 1, but not significantly (p=0.78 and p=0.26, respectively). Group 2 had a significantly higher percentage of patients with preserved ejection fraction (86.42% vs. 70.83% in Group 1 and 68.89% in Group 3, p=0.03). Brain natriuretic peptide (BNP) was only reported in 39 patients (13 in Group 1, 18 in Group 2 and 8 in Group 3). Although a tendency for higher BP across groups was found, this was not significant (p=0.37).

Regarding treatment (Table 1), no differences were found in antihypertensive drugs or diuretics, including thiazides. The proportion of patients in Group 1 taking beta-blockers was slightly higher (70.8% vs. 50.62% in Group 2 and 60.87% in Group 3, p=0.07), although not significantly. Dosages in the total sample were (median and interquartile range): 40 mg (0) for diuretics, 5 mg (7.5) for betablockers, 75 mg (128) for angiotensin receptor blockers, 10 mg (15) for angiotensin-converting enzyme inhibitors, 10 mg (15) for calcium antagonists, 25 mg (0) for aldosterone antagonists and 12.5 mg (12.5) for thiazides.

Data from 24-h ABPM (Table 2) showed firstly a predominance of anomalous circadian profiles (absence of normal night-time BP fall) in all three groups. In total we found 37.14% non-dippers, 38.29% risers and 4% extreme dippers, without significant differences between the groups. However, Group 1 had a significantly higher percentage with a riser pattern than the other two groups (p=0.05). Secondly, 24-h SBP and dSBP were significantly higher in Groups 2 and 3 (24-h SBP: 118.82±18.8 mmHg in Group 1, 126.39±16.8 mmHg in Group 2 and 123.8±20.1 mmHg in Group 3, p=0.05, and dSBP: 119.37±20.7 mmHg in Group 1, 128.06±16.68 mmHg in Group 2 and 127.18±19.72 mmHg in Group 3, p=0.008). Consequently, significant differences were also found in falls in both SBP and DBP between waking and sleeping times, the smallest fall being seen in Group 1 (Table 2 and Figure 2).

Figure 2.

Boxplots showing differences in 24-h SBP (mean 24-hour systolic blood pressure), dSBP (mean daytime systolic blood pressure), fall in SBP (systolic blood pressure) and fall in DBP (diastolic blood pressure), by groups of serum sodium level.

(0.24MB).

In multivariate adjusted regression analysis (Table 3), a significant relationship was found between sodium levels and both 24-h SBP (odds ratio [OR] 0.97, 95% confidence interval [CI] 0.95-0.99, p=0.01) and dSBP (OR 0.96, 95% CI 0.94-0.99, p=0.004), but no significant differences in the other variables.

The main finding in our study is that lower serum sodium levels in outpatients with HF are related to lower SBP, principally during waking hours.

Hyponatremia often coexists with advanced HF,10 and data from registries and clinical trials have shown an association between hyponatremia and increased mortality and morbidity even in outpatients.11 In most of these cases, low BP was linked with low serum sodium, but before the present study, data from a complete 24-h period of BP monitoring had not been published, and previous data were probably based on a single reading per patient, likely to have been taken in the daytime.

In the general population sodium balance is almost entirely controlled by the kidneys through urinary sodium excretion, and the immediate effects of dietary sodium intake are to modify blood sodium and consequently ECV. These changes largely account for later alterations that influence BP, but it is also known that acute rises in sodium intake can trigger a rise in BP by either increasing blood sodium levels even when ECV falls,12 or by increasing ECV when plasma sodium levels decrease.13 This would indicate that acute changes in blood sodium and ECV are both likely able to influence BP independently, which in cases of hyponatremia would lead to lower BP. Although in HF this situation could persist, the reasons are different, particularly in the case of hyponatremia due to upregulation of maladaptive neurohormonal mechanisms, including increased secretion of arginine vasopressin and increased activity of the renin-angiotensin and sympathetic nervous systems, which may underlie the link between development of hyponatremia and HF severity.14 Thus, in HF patients, neurohormonal control may not affect BP as much as in other populations, and reduced cardiac output leading to low SBP may play a more important role; thus, in our study only the differences in SBP were significant, DBP being similar across groups, even though no relationship was found between serum sodium concentrations and LVEF either in our analysis or in most published studies.15–18 In addition, drugs used to treat HF can also influence both BP and hyponatremia. In our study no differences were found between groups regarding the use of these drugs, but it should be noted that in all groups the normal circadian BP profile was absent, with a lower night-time fall in BP, as described previously.19

This lower BP fall while sleeping can be explained by several factors. Firstly, most drugs that affect BP are taken early in the morning, their effects persisting during waking hours but then weakening during the night. Secondly, it could result from a nocturnal increase in central venous pressure as a consequence of dispersion of retained fluids when body position changes during sleep. Thirdly, in healthy individuals the cardiopulmonary baroreflex inhibits sympathetic activation and helps decrease BP, but some studies suggest that impaired baroreflex in HF may cause a reduction in night-time BP decline,20,21 a finding that may also be related to the increased sympathetic stimulation observed in HF, especially in patients with hyponatremia.22 The latter hypothesis could explain the lower nocturnal BP dip in our Group 1, as well as the higher proportion of riser profiles in this group. Reduction in the circadian amplitude of BP rhythm has been related to severity of HF,23 similarly to hyponatremia, and could partly explain the worse outcomes in these patients. It also suggests avenues for further research, focusing on changes in scheduling of medication, administering antihypertensive drugs at the beginning of sleep instead of on waking, aiming to increase night-time BP fall.

This study has several limitations. Firstly, its results are based on cross-sectional analysis, and so causal relationships cannot be determined. Secondly, because of the small sample size, it was divided by quartiles of sodium levels, instead of comparing a true hyponatremic group (defined as sodium level <135 mEq/l) with other patients; however, some studies have shown worse outcomes when sodium levels drop below 140 mEq/l.1 In addition, the number of patients per group was low, which probably diminished the significance of the results, and so we used robust statistical methods to improve them. On the other hand, the pathogenesis of hyponatremia in HF is multifactorial. Data on osmolarity were not available in all our patients, and thus it is not known whether patients’ hyponatremia was hyper- or hypovolemic. In this regard, diuretic doses were not available by group, but the rate of thiazide use was low, and it therefore seems unlikely that diuretic overuse was the cause of hyponatremia in a significant proportion of cases. Finally, the doses of drugs taken by participants were not assessed, and the lack of this information – which could have influenced BP values – could have biased our results.

On the other hand, our study included the largest number to date of HF patients to undergo ABPM.

Our findings show a relationship between lower sodium levels in patients with HF and low SBP, principally during waking hours, and with a smaller fall in BP between daytime and night-time, which could be partly responsible for the worse outcome in these patients. Further studies will be needed to elucidate this subject.

The authors have no conflicts of interest to declare.