Introduction

Various studies have demonstrated that excessive sodium intake is associated with elevated blood pressure (BP) levels and an increased risk of cardiovascular events.1, 2, 3 Many trials and epidemiological studies have shown that BP and cardiovascular events decrease with reduced sodium intake.4

Traditionally, brachial BP has been used as a surrogate marker for BP-related cardiovascular events. However, recent studies have shown that central BP has better predictive power for future cardiovascular events compared with brachial BP.5 The early return of the reflected wave during the ventricular ejection period augments central systolic BP (SBP). Among the factors associated with the augmentation of central SBP, an increase in arterial stiffness increases aortic pulse wave velocity, with early return of the reflected wave from peripheral arteries.6 Studies have shown that high sodium intake is associated with elevated aortic pulse wave velocity,7, 8, 9 suggesting that high sodium intake elevates central SBP.

Several studies have shown the association between urinary salt excretion and central hemodynamics.10, 11, 12 However, the relationship between measured (not estimated) 24-h urine sodium (24HUNa), potassium (24HUK) and central hemodynamics is unknown in Asian populations.

We evaluated the association between 24-h urinary salt excretion and central hemodynamic in the general population.

Methods

Study population

Random samples of adults from Goyang city, with an approximate adult population of 600 000, and Paju city, with an approximate adult population of 240 000, were selected by list-assisted random digit dialing, and both listed and unlisted telephone households were included. The study was performed in 2011 in Goyang city and in 2012 in Paju city. The design of the study population selection and enrollment has been described in detail elsewhere.13 We used the same study design in both cities. Adults between 20 and 65 years of age within the selected households were chosen through telephone interviews.

Study protocol

The participants visited the clinical trial center before 0830 hours. Urine collection began after voiding and discarding the first urine sample between 0900 and 0930 hours. All urine voided during the next 24 h was collected. The 24-h ambulatory BP was measured in parallel. The next day, the participants visited the clinical trial center before 0900 hours. The participants were asked to void the last urine sample 24 h after the first voiding. The last voided urine sample was added to the urine collection bag. After completing the 24-h urine collection, casual BP and central BP were measured. The concentrations of sodium and potassium were measured using an ion-selective electrode method (Modular DPE chemistry; Roche Diagnostics, Mannheim, Baden-Württemberg, Germany). Creatinine was measured using the Jaffe reaction (Kinetic colorimetric assay; Roche Diagnostics). For biochemical analysis, a blood sample was obtained after at least 8 h of overnight fasting. The study protocol was reviewed and approved by the Institutional Review Board of Dongguk University Ilsan Hospital. Each participant provided written informed consent.

Completeness of urine collection

The validity of the 24-h urine collection was assessed using a combination of self-reported urine loss and the 24-h urine creatinine-based determination, assuming that the urinary creatinine excretion rate was constant. If the self-reported urine sample loss was more than 100 ml, if the loss occurred more than once, or if either the urinary creatinine level was <6 mmol per day with a total urine volume of <1000 ml per day or the urinary creatinine level was <5 mmol per day, then the collected urine sample was considered to be insufficient for a 24-h collection.14

BP measurement

Casual BP levels were measured from both arms (with the participant in a sitting position) using a validated semiautomatic device (Watch BP Office; Microlife AG, Widnau, Switzerland) by a trained nurse.15 BP was measured three times at 1-min intervals after 5 min of rest, and three SBP and diastolic BP (DBP) measurements from each arm were averaged. Ambulatory BP was measured for 24 h in 30-min intervals (Mobil-O-Graph; I.E.M. GmbH, Stolberg, Germany) in conjunction with the 24-h urine collection.16 More than 70% of the expected measurements, 14 measurements during the day and 7 measurements at night, were considered to be valid measurements.17 Hypertension was defined as an averaged 24-h SBP ⩾130 mm Hg, an averaged 24-h DBP ⩾80 mm Hg or the current use of antihypertensive medication with a previous diagnosis of hypertension.18

Aortic BP was measured using a validated non-invasive device (Sphygmocor; AtCor Medical, Sydney, Australia).19, 20 The participants rested for 5 min in a sitting position before the measurements were performed. Aortic BP was calculated from pressure waveforms obtained by applanation tonometry from the radial artery at the wrist and brachial BP, which was taken before the pressure waveform recordings using the generalized transfer function. A valid measurement was an operator quality index >80%. Aortic pulse pressure was calculated as the difference between the aortic systolic and diastolic BP (AoSBP and AoDBP, respectively). Aortic augmentation pressure is the difference between the AoSBP and the inflection point at the end of the first systolic shoulder. The augmentation index is the ratio of the aortic augmentation pressure to the aortic pulse pressure. Given that the augmentation index is influenced by patient heart rate, an augmentation index normalized for a heart rate of 75 beats per minute (AI75) was derived.

Statistical analysis

The study population was divided into high 24HUNa (⩾151 mmol/24-h, mean 210.5±52.0 mmol per day, range 151.0–432.0 mmol per day) and low 24HUNa (<151 mmol/24-h, mean 115.7±25.0 mmol per day, range 45.6–150.0 mmol per day), according to the level of 24HUNa excretion, and high 24HUNa/24HUK (⩾2.93, mean 4.03±1.00, range 2.93–7.96) and low 24HUNa/24HUK (<2.93, mean 2.13±0.54, range 0.53–2.91), according to the 24HUNa and 24HUK ratio.

All statistical tests were two-sided, and the level of significance was set at 0.05. Continuous variables were expressed as the means±s.d. Categorical variables were described as numbers and percentages in parenthesis. The continuous data were analyzed using Student’s t-test for two-group comparisons, and the proportions between groups were compared using the χ2 test. In the group comparisons of high and low 24HUNa and high and low 24HUNa/24HUK, the effects of age, gender, body mass index and antihypertensive medication were included. Multivariate linear regression analysis was performed to predict the independent effect of 24HUNa and 24HUNa/24HUK on the hemodynamic variables, including the effects of age, gender, body mass index and use of antihypertensive medications. All analyses were implemented using SPSS ver. 20 (SPSS, Chicago, IL, USA).

Results

Among 707 individuals, 524 participants who had valid collections of 24-h urine, valid measurements of 24-h ambulatory BP and valid central BP measurements were included in the analysis. The demographic and clinical characteristics of the analyzed population are shown in Table 1. In total, 219 of the participants had hypertension (135 non-medicated, 84 medicated). The hypertensive individuals had a higher mean age, body mass index, prevalence of men, smokers and diabetes, 24HUNa and 24HUK. The 24HUNa/24HUK did not differ between the hypertensive and normotensive individuals. The serum creatinine, fasting blood glucose, cholesterol and triglyceride levels were higher in the hypertensive individuals than in the normotensive individuals.

Table 1 Characteristics of study population
Full size table

Table 2 shows the hemodynamics comparison between the high-24HUNa and low-24HUNa groups. Among the entire study population, there was no difference in hemodynamics between the high- and low-24HUNa groups. In the hypertensive individuals, the high-24HUNa group had higher aortic systolic BP (AoSBP), 24-h SBP and DBP than the low-24HUNa group, with the adjusted analysis controlled for age, gender, body mass index and use of antihypertensive medication. In contrast, among the normotensive individuals, the low-24HUNa group showed higher casual DBP, 24-h DBP, AoSBP and AoDBP than the high-24HUNa group, after controlling for confounders.

Table 2 Difference of hemodynamic values between high and low 24-hour urine sodium excretion
Full size table

As shown in Table 3, the high 24HUNa/24HUK group had higher 24-h DBP and AoSBP. Among the hypertensive individuals, the high 24HUNa/24HUK group had higher casual SBP and DBP, 24-h SBP and DBP, and AoSBP and AoDBP than the low-24HUNa/24HUK group. However, casual DBP and AoSBP showed a persistent difference, with a multivariate analysis controlled for age, gender, body mass index and use of antihypertensive medication. There was no difference in hemodynamics between the low and high 24HUNa/24HUK groups among the normotensive individuals.

Table 3 Difference of hemodynamic values between high and low Na/K ratio
Full size table

Multivariate regression analysis, controlled for age, gender, body mass index, use of antihypertensive medication and 24HUNa, revealed an independent association between 24HUNa/24HUK and AoSBP (Table 4). The 24HUNa was not independently associated with AoSBP. In contrast to the aortic SBP, the 24-h SBP and DBP were not associated with 24HUNa (P=0.420 and 0.860, respectively) and 24HUNa/24HUK (P=0.181 and 0.121, respectively) in a multivariate regression analysis (table not shown).

Table 4 Independent relationship between aortic systolic blood pressure and 24-h urinary sodium/potassium with multivariate regression analysis adjusted for determinants
Full size table

Discussion

We evaluated the association between urinary Na and K excretions and hemodynamics. This study is the first to reveal the independent association between a high ratio of urinary sodium/potassium and elevated AoSBP in Asian hypertension patients.

In our study, in hypertensive participants, there was a significant association between high 24HUNa and higher 24-h SBP, 24-h DBP and AoSBP compared with low-24HUNa. Previous studies have demonstrated an association between high sodium intake and elevated aortic pulse wave velocity.8, 9 Elevated aortic pulse wave velocity leads to elevated AoSBP through the early return of backward pressure. In addition, high 24HUNa/24HUK was associated with higher AoSBP but not with 24-h SBP and 24-h DBP. These findings indicate that AoSBP is influenced not only by high sodium intake but also by high sodium intake and/or low potassium intake. Multivariate regression analysis showed an independent and positive association between 24HUNa/24HUK and AoSBP, indicating a stronger effect of balanced sodium and potassium intake on AoSBP than sodium intake alone. Given our study results, low sodium and/or high potassium intake is an important strategy to avoid the elevation of central aortic BP and eventually cardiovascular events, because elevated central aortic BP is strongly associated with future cardiovascular events.5

Along with high dietary sodium intake, a well-known environmental factor, low potassium intake has been shown to be related to hypertension.21, 22, 23 However, in the present study, 24HUK levels were higher in hypertensive individuals than that in normotensive individuals. Moreover, there was no relationship between the 24HUK and BP parameters (data not shown). The lack of such a relationship contrasts with the results of a previous study of a Korean population.24 Although the reason is unclear, the lack of a relationship between 24HUK and BP may be explained by the different sodium and potassium intake estimation methods (24-h urine collection vs dietary survey) and the exclusion of individuals older than 65 years of age. The difference in methods may yield bias in terms of the sodium and potassium intake levels at the population and individual levels. Using the 24-h collection method, a discrepancy in sodium and potassium intake levels has been reported when these levels were measured using the dietary survey,25 spot urine26, 27 and overnight urine collection methods.28 Another reason may be the blunted effect of potassium in lowering BP when sodium intake is low29, 30 because the sodium intake in our study population was relatively low compared with that in the previous study.24

In addition to sodium or potassium intake alone, the INTERSALT study revealed a strong negative association between urinary sodium/potassium ratio and BP.31 One of the suggested mechanisms underlying the elevation of BP through long-term potassium intake depletion is a cellular potassium deficit and sodium retention by the kidneys, leading to hypertension.32 Another explanation is sodium sensitivity, which is highly prevalent in blacks.33, 34 Although there is no direct evidence, relatively low urinary potassium excretion (i.e., high 24HUNa/K) suggests the presence of sodium sensitivity, considering the high prevalence of sodium sensitivity and low urinary potassium excretion in blacks.35 The prevalence of sodium sensitivity in Asian hypertension patients is expected to be approximately 50%.36, 37 In our study, the positive association between AoSBP and 24HUNa/24HUK existed only in hypertensive individuals, and it was not observed in normotensive individuals. Thus, the combination of relatively low potassium intake and sodium sensitivity may contribute to the elevation of central BP.

In the literature, three studies previous have evaluated the association between sodium and potassium intake with central hemodynamics.10, 11, 12 Among these studies, only one showed a significant correlation between AoSBP and urinary sodium/potassium ratio. The other two studies did not demonstrate an independent association between AoSBP and urinary sodium/potassium ratio or sodium excretion. There were several differences between the previous studies and our own. First, we diagnosed hypertension using a 24-h ambulatory BP measurement. Unlike our study, Park et al.11 did not exclude white coat hypertension because they used conventionally measured BP. Polonia et al.12 did not provide the criteria they used to diagnose hypertension. Second, we collected 24-h urine samples to measure urinary sodium and potassium excretion. Park et al.11 estimated 24HUNa using spot urine samples from Korean hypertension patients. The accuracy of the equations estimating 24HUNa from spot urine has been questioned,26 although it is widely used. Thus, it is unclear whether the results of an association between aortic hemodynamics and estimated 24HUNa could be extrapolated to the measured24HUNa. Third, the mean 24HUNa in our study was 159.8 mmol/24-h, which was between the measured 24HUNa in other studies (209 mmol/24-h and 105 mmol/24-h).10, 12 The positive association between 24HUNa and central hemodynamics in populations with different means of 24HUNa suggests a significant influence of high sodium intake on central hemodynamics without the distinction of sodium intake levels. The estimated 24HUNa from Park et al.11 is difficult to compare because they estimated 24HUNa from spot urine samples using an equation previously developed by Tanaka et al.;38 this equation showed a significant difference between the mean estimated and measured 24HUNa values in the validation of the equation in the external population. We also showed that estimating 24HUNa from spot urine samples, including with Tanaka’s equation, tends to under- or overestimate 24HUNa according to the amount of the 24-h urine Na, differing significantly from the measured 24HUNa.26, 39 Fourth, the ethnicity of the population in our study differed from that of the Redelinghuys et al. study.10 Blacks have a higher prevalence of sodium sensitivity compared with other ethnicities. The high prevalence of sodium sensitivity may contribute to a positive association of casual and ambulatory SBP with 24HUNa/24HUK in the Redelinghuys et al. study,10 which was not observed in our study. Nevertheless, the linear association between AoSBP and 24HUNa/24HUK in hypertensive individuals but not in normotensive individuals in our study may be explained by sodium sensitivity, which is more prevalent in hypertensives than in normotensives.33, 37 Sodium-resistant individuals, who have no response or a negative response to high sodium intake, are more likely to be normotensive than hypertensive. Thus, the lack of an association between AoSBP and 24HUNa/24HUK in normotensive individuals may be explained by the high prevalence of sodium-resistant individuals.

One limitation of our study is its cross-sectional nature, resulting in the inability to speculate on the causality between the urinary sodium/potassium ratio and central aortic BP. The following limitations may have resulted in bias in our study results. The sodium intake was measured using a single 24-h urine collection. A single 24-h urine collection does not accurately reflect dietary sodium intake, although it is considered to be the most reliable method in population surveys.40, 41 However, multiple collections of 24-h urine samples (more than seven times) are impractical in population surveys. Moreover, an indirect method (using a generalized transfer function) was used to measure central hemodynamics.42

In conclusion, 24HUNa/24HUNK was positively related to AoSBP in patients with hypertension, demonstrating a stronger effect compared with 24HUNa alone. The result indicates that high sodium and low potassium intakes may increase the subsequent risk of cardiovascular disease through the elevation of central aortic BP.