Introduction

Hypertension is one of the most important treatable risk factors for chronic kidney disease (CKD).1, 2 However, the mechanism by which hypertension influences renal function and produces renal vascular damage is still incompletely understood. Blood pressure (BP) level is widely recognized to strongly correlate with CKD, as it does with other cardiovascular diseases.3, 4, 5 Clinical guidelines recommend using mean BP, systolic BP (SBP) and diastolic BP (DBP) for diagnosis and management of hypertension to prevent cardiovascular events such as progression of renal dysfunction.

Recently, additional parameters, such as variability in BP, were demonstrated to correlate with cardiovascular events.6 For example, Rothwell et al.7 showed that visit-to-visit variability in SBP is a strong predictor of stroke. There are no data available regarding the relationship between renal function and visit-to-visit variability in BP.

The resistive index (RI: (peak systolic velocity−end diastolic velocity)/peak systolic velocity at segmental arteries in kidney) evaluated by renal Doppler ultrasonography (RDU) is considered a useful index of renal vascular resistance secondary to arteriosclerosis8, 9, 10 and is a good prognostic indicator of renal function.11, 12, 13, 14, 15, 16 We previously reported that the RI might be a more efficacious parameter for the evaluation of very early renal damage than estimated glomerular filtration rate (eGFR).17 It is thought that renal vascular damage caused by atherosclerotic risk factors such as hypertension and variability of BP could be evaluated more precisely by the RI.

In this study, we assessed the hypothesis that visit-to-visit variability in BP correlates significantly with renal function. We investigated the correlation between visit-to-visit variability in BP (expressed as standard deviation (s.d.) and coefficient variant (CV: s.d./mean)) and various renal function parameters such as eGFR, RI and level of proteinuria. We also investigated the correlation between visit-to-visit variability in BP and other biochemical parameters.

Methods

Study subjects

Figure 1 shows a flow chart of the selection process for study participants. In our hospital ward, almost all admitted patients for several internal diseases, such as diabetes mellitus, hypertension, CKD and so on, undergo RDU to evaluate renal arteriosclerotic change. We initially enrolled 281 consecutive patients with and without CKD undergoing RDU in our hospital ward between February 2009 and May 2011. Patients were excluded if they had renal artery stenosis (n=16), renal transplant (n=1) or were on dialysis (n=1). Of the 263 patients, 120 returned for follow-up visit with BP measurement less than six times; therefore, a total of 143 patients were included in this study. In addition, we analyzed a subgroup of 104 patients who had BP measured at outpatient visits 10 or more times before and/or after RDU examination.

Figure 1
figure 1

Flow chart of the selection process of study subjects.

Visit-to-visit variability in BP was defined as the s.d. and CV in BP. Variables were calculated from BP measured at six serial visits before and/or after RDU examination.

Subjects underwent biochemical examination of the blood and urine. Clinical parameters considered in this study were: height, weight, body mass index, eGFR, serum lipid profile, fasting blood glucose, glycosylated hemoglobin, hemoglobin A1c, additional biochemical parameters, proteinuria level, SBP and DBP at the time of RDU, smoking history and drug profile. The Clinical Investigations Ethics committee of Osaka University Hospital approved the study protocol. The study was performed in adherence with the principles of the Declaration of Helsinki and according to Good Clinical Practice standards.

Ultrasonographic determination

RI was calculated as:

RI=(peak systolic velocity−end diastolic velocity)/peak systolic velocity.

Patients were placed in a supine position, and the size of the left and right kidneys and the flow velocity in the aorta and renal arteries were evaluated to detect morphological abnormality or renal artery stenosis. RI was determined in three different segmental arteries of both kidneys, and expressed as the mean of these values; this method was reported to be identical and technically easy to perform so that reproducibility of RI could be improved.18, 19, 20 Previous studies indicated that reliable RI measurements depended on proper measuring techniques performed by experienced operators,21 so in this study Doppler examinations were performed by the three experienced operators (TK, KK and MO) using a XARIO SSA-660A ultrasound machine (TOSHIBA, Tokyo, Japan) with a 2.5-MHz sector transducer.

Renal function

eGFR was calculated using the following equation:

eGFR (ml min−1 per1.73 m2)=194 × creatinine−1.094 × age−0.287 ( × 0.739 if female).22

The level of albuminuria was evaluated according to the American Diabetes Association classification.23 The albumin/creatinine ratio in spot urine was used to classify proteinuria as follows: no proteinuria: <30 mg g−1 creatinine; microalbuminuria: 30–300 mg g−1 creatinine; clinical albuminuria: ≥300 mg g−1 creatinine.

BP measurements

Conventional BP was measured by trained observers with an electronic sphygmomanometer (HEM-705IT or HEM-711; OMRON, Kyoto, Japan). Following the guidelines for the management of hypertension, at every visit (monthly–bimonthly), clinic BP was measured at least two times in sitting position after 5 min rest, and we adopted the average of two readings as office BP if the difference of measured values was <5 mm Hg. When the difference of measured values was more than 5 mm Hg, additional measurements were conducted to obtain stable BP readings and we adopted the average of the two stable readings as office BP.

Statistical analysis

All data are expressed as mean±s.d. Differences between groups were analyzed employing the unpaired Student's t-test and Pearson's χ2 test. Multiple linear regression analysis was performed to determine more related variables for RI. Stepwise multiple regression analysis was used to identify possible determinants of RI. The level of significance was defined as P<0.05. All statistical analyses were performed using JMP (JMP version 8.0.1, Cary, NC, USA).

Results

This study included 143 consecutive patients who underwent RDU in our hospital ward from February 2009 to May 2011 and who had their BP measured at outpatient visits six or more times before and/or after RDU examination (Figure 1).

Table 1 reports baseline patient characteristics. Mean age was 68.1±13.0 years. In all, 79 patients were women and 64 were men. A total of 133 patients (93.0%) were receiving treatment with antihypertensive agents, including angiotensin receptor blockers, angiotensin-converting enzymes inhibitors, calcium channel blockers, diuretics, β-blockers, α-blockers or aldosterone blockers. In all, 95 patients (66.4%) were receiving two or more different antihypertensive agents; 58 patients (40.8%) were taking statins for dyslipidemia; and 51 patients (35.7%) had type 2 diabetes mellitus. Table 2 showed comparison of baseline clinical characteristic of the subjects between lower s.d. in SBP group (N=71; mean±s.e.m.: 8.91±0.26) and higher s.d. in SBP group (N=72; mean±s.e.m.: 17.49±0.58). Higher s.d. in SBP group showed significantly higher RI, lower DBP and higher prevalence rate of type 2 diabetes mellitus.

Table 1 Baseline clinical characteristic of all subjects
Table 2 Comparison of baseline clinical characteristic of the subjects between lower s.d. in SBP group (N=71; mean±s.e.m.: 8.91±0.26) and higher s.d. in SBP group (N=72; mean±s.e.m.: 17.49±0.58)

Firstly, we investigated how multiple clinical parameters correlated with variability in BP. Subjects with diabetes mellitus (DM) showed a significantly higher variability in SBP than those without DM (s.d. in SBP, patients with DM: 14.46±5.97 vs. patients without DM: 12.32±5.68, P<0.05; CV in SBP, patients with DM: 0.10±0.04 vs. patients without DM: 0.09±0.04, P<0.05). There were no significant correlations between variability in BP and usage of each antihypertensive agent (angiotensin receptor blockers, angiotensin-converting enzymes inhibitors, calcium channel blocker, diuretics, β-blockers, α-blockers or aldosterone blockers).

Secondly, we compared RI, eGFR and level of proteinuria between the quartile of s.d. in SBP and CV in SBP to investigate how variability of BP correlated with renal function. Patients with higher variability in SBP showed a significantly higher prevalence rate of clinical albuminuria and microalbuminuria (Pearson's χ2 test, P=0.0014) (Figure 2) and a significantly higher RI (Figure 3). However, there was no significant correlation between variability in SBP and eGFR.

Figure 2
figure 2

Association between s.d. quartile in SBP and proteinuria. Patients with higher s.d. in SBP showed a higher prevalence rate of microalbuminuria and clinical albuminuria (Pearson's χ2 test, P=0.0014).

Figure 3
figure 3

Association between the resistive index and variability in blood pressure analyzed in 143 patients with blood pressure measured at least six times at outpatient visits. Left: association between the resistive index and quartile of s.d. in SBP. Quartile 1: 0.68±0.01; quartile 2: 0.69±0.01; quartile 3: 0.70±0.01; and quartile 4: 0.73±0.01 (mean±s.e.m.). Right: Association between the resistive index and quartile of CV in SBP. Quartile 1: 0.69±0.01; quartile 2: 0.68±0.01; quartile 3: 0.69±0.01; and quartile 4: 0.74±0.01 (mean±s.e.m.). Patients with higher variability in SBP showed a significantly higher RI than patients with lower variability.

As Rothwell et al.7 suggested that risks of stroke and coronary events increased in relation to maximum SBP in outpatient visits, we investigated the correlation between maximum SBP in six outpatients visit and RI, eGFR and proteinuria. RI was significantly correlated with maximum SBP (R=0.179, P<0.05); however, eGFR was not significantly correlated with maximum SBP (R=0.080, P=0.344). Patients with albuminuria or microalbuminuria showed significantly higher maximum SBP (patients with albuminuria or microalbuminuria: 167.5±25.0 vs. patients without albuminuria: 154.6±15.8, P<0.005.)

In the recent article from the NHANES III, BP measured at outpatient visits for three times were analyzed as BP variability, and higher levels of short-term visit-to-visit SBP variability were associated with increased all-cause mortality.24 Therefore, to further investigate, we analyzed a total of 171 patients with BP measured at outpatient visits three or more times before and/or after RDU examination and calculated s.d. and CV similarly. Subjects with higher variability in SBP from three BP measurements also showed a significantly higher RI with lower decision correlation (Figure 4). On the other hand, patients with higher variability in SBP from three BP measurements showed the tendency of higher prevalence rate of clinical albuminuria and microalbuminuria, but there was no significant correlation (data not shown).

Figure 4
figure 4

Association between the resistive index (RI) and variability in blood pressure analyzed with 171 patients with s.d. and CV of three times blood pressure measurement at outpatient visits. Left: association between the RI and quartile of s.d. in SBP. Patients with quartile 1: 0.68±0.01; quartile 2: 0.67±0.01; quartile 3: 0.71±0.01; and quartile 4: 0.72±0.01 (mean±s.e.m.). Right: association between the RI and quartile of CV in SBP. Patients with quartile 1: 0.69±0.01; quartile 2: 0.68±0.01; quartile 3: 0.70±0.01; and quartile 4: 0.72±0.01 (mean±s.e.m.). Patients with higher variability in SBP showed significant higher RI than patients with lower variability.

Multivariate analysis revealed that s.d. in SBP and CV in SBP was significant risk factors for higher RI independent of other renal risk factors. Table 3 showed that correlation between RI and variability in SBP. Without adjustment, s.d. in SBP and CV in SBP were significantly correlated with RI (s.d. in SBP: P=0.0006; CV in SBP: P=0.0020). In model 1, adjusted for age, body mass index, SBP, eGFR and with or without DM, s.d. in SBP and CV in SBP were also significantly correlated with RI (s.d. in SBP: P<0.0001; CV in SBP: P<0.0001). In model 2, adjusted for model 1+with or without smoking, s.d. in SBP and CV in SBP were significantly correlated with RI too (s.d. in SBP: P<0.0001; CV in SBP: P<0.0001). In model 3, adjusted for model 1+with or without albuminuria, s.d. in SBP and CV in SBP were also significantly correlated with RI (s.d. in SBP: P<0.0001; CV in SBP: P=0.0002).

Table 3 Correlation between variability in SBP and RI

Discussion

This is the first report analyzing the impact of variability in BP on renal function. We showed that visit-to-visit variability in BP was significantly correlated with the level of albuminuria and was a risk factor for higher RI independent of other traditional renal risk factors.

Previous studies revealed that the RI was a useful predictor of renal dysfunctions11, 12, 13, 14, 15, 16 and correlated significantly with organ damage. Measurement of RI in addition to low-grade albuminuria is reportedly useful for target organ damage screening in patients with resistant hypertension.10 RI values were independently correlated with carotid intima–media thickness in patients with never-treated essential hypertension9 and metabolic syndrome.8 These results suggest that renal vascular resistance indicated by the RI reflects the degree of systemic atherosclerosis, and that RI can serve as a useful marker to detect and evaluate atherosclerotic diseases due to cardiovascular diseases risk factors, such as hypertension, DM, dyslipidemia and metabolic syndrome.

Previously, a very interesting investigation was reported concerning the relationship between RI and histopathological analysis by renal biopsy, including glomerular sclerosis, interstitial fibrosis/tubular atrophy, interstitial infiltration and arteriolosclerosis. By stepwise multiple regression analysis, only arteriolosclerosis was found to be an independent risk factor for increased RI.25 Therefore, the results of our study showing the correlation between visit-to-visit variability in BP and RI indicate that visit-to-visit variability in SBP is an independent risk factor for renal arteriosclerotic change, and could be a useful predictor of renal dysfunction.

Multivariate analysis revealed that variability in SBP was significant risk factors for higher RI independent of not only other renal risk factors, but also eGFR and albuminuria. It indicated that employing variability in SBP in addition to eGFR and albuminuria to evaluate renal arteriosclerotic change is convenient and efficacious in clinical practice.

Previous reports suggest that DBP variability is independently predictive of nephropathy, but not of retinopathy, in DM patients.26 We found very similar results in hypertensive patients. In this study, variability in BP correlated significantly with the level of proteinuria and RI, but not with eGFR; although eGFR calculated easily from serum creatinine level, age and sex is useful for renal function screening, it is generally difficult to assess the pathogenesis of CKD using only eGFR. We previously reported that the RI might be a more sensitive parameter for the evaluation of very early renal vascular damage than eGFR,17 so we thought that RI could detect early renal damage correlated with variability in BP more sensitively than eGFR, and that this is the reason why higher variability in SBP was significantly correlated with higher RI and higher prevalence rate of clinical albuminuria and microalbuminuria, but not with eGFR.

Based on previous large clinical trials, mean BP is widely recognized to correlate strongly with cardiovascular diseases, including CKD, and clinical guidelines recommend using mean BP for diagnosis and management of hypertension to prevent cardiovascular event. This study suggests that visit-to-visit variability in BP should be considered a new therapeutic target for renal protection independent of mean BP.

Rothwell et al.27 reported that calcium channel blocker and β-blockers had different effects on variability of BP. In this study, however, there were no differences according to the type of antihypertensive agent (angiotensin receptor blockers, angiotensin-converting enzymes inhibitors, calcium channel blocker, diuretics, β-blockers, α-blockers or aldosterone blockers). One explanation for this is that most of our subjects were receiving two or more different antihypertensive agents. Therefore, studies comparing each antihypertensive agent independently are necessary to evaluate their renoprotective effects.

Compared with lower s.d. in SBP group, higher s.d. in SBP group showed significantly higher RI, lower DBP and higher prevalence rate of type 2 DM. With aging or atherosclerotic disease, the arterial wall is generally stiffening, and the elasticity reduces; thus, DBP lowers and the pulse pressure rises.28, 29 The decrease in DBP is thought to reflect vascular damage and increased risk of atherosclerosis. Therefore, the vascular damage induced by cardiovascular risk factors such as DM and hypertension is thought to be correlated with variability in BP.

We showed that subjects with higher variability in SBP from three BP measurements also showed a significantly higher RI with lower decision correlation. In clinical setting, the BP variability calculated from few outpatient visits is thought to be more convenient and easy-to-use index to assess the arteriosclerotic change of patients, although for more accurate assessments we could take more BP measurements in account.

Study limitations

This study has several limitations. First, our study was observational and cross-sectional. Longitudinal, prospective studies are necessary to evaluate the utility of measuring variability in BP for predicting the progression of atherosclerotic diseases and renal prognosis. Second, our sample size was relatively small. Third, because patients were recruited at the university hospital in this study, most patients had already been under medical treatment for hypertension, dyslipidemia and diabetes at the time of investigation; therefore, several parameters such as BP, lipid profile and glycosylated hemoglobin, hemoglobin A1c might have been influenced by medical treatment.

Conclusion

In conclusion, this study documents the impact of variability in BP on renal function. Visit-to-visit variability in BP is correlated significantly with renal function evaluated by the RI and albuminuria. This parameter could be a useful predictor of renal dysfunction and should be considered as a therapeutic target for renal protection.