Impact of home blood pressure variability on cardiovascular outcome in patients with arterial stiffness: Results of the J‐HOP study

Abstract This study sought to investigate whether the relation between increased blood pressure (BP) variability and increased arterial stiffness confers a risk for cardiovascular disease (CVD) events. We analyzed 2648 patients from a practitioner‐based population (mean ± SD age 64.9 ± 11.4 years: 75.8% taking antihypertensive medication) with at least one cardiovascular risk factor who underwent home BP monitoring in the Japan Morning Surge‐Home Blood Pressure Study. The standard deviation (SDSBP), coefficient of variation (CVSBP), and average real variability (ARVSBP) were assessed as indexes of day‐by‐day home systolic BP (SBP) variability. The authors assessed arterial stiffness by brachial‐ankle pulse wave velocity (baPWV) and divided patients into lower (< 1800 cm/s, n = 1837) and higher (≥1800 cm/s, n = 811) baPWV groups. During a mean follow‐up of 4.4 years, 95 cardiovascular events occurred (8.1 per 1000 person‐years). In Cox proportional hazard models adjusted for traditional cardiovascular risk factors including average home SBP, the highest quartiles of SDSBP (hazard ratio [HR], 2.30; 95% confidence interval [CI], 1.23‐4.32), CVSBP (HR, 2.89; 95%CI, 1.59‐5.26) and ARVSBP (HR, 2.55; 95%CI, 1.37‐4.75) were predictive of CVD events compared to the other quartiles in the higher baPWV group. Moreover, 1SD increases in SDSBP (HR, 1.44; 95%CI, 1.13‐1.82), CVSBP (HR, 1.49; 95%CI, 1.16‐1.90) and ARVSBP (HR, 1.37; 95%CI, 1.09‐1.73) were also predictive of CVD events. These associations remained even after N‐terminal pro‐brain natriuretic peptide was added to the models. However, these associations were not observed in the lower baPWV group. We conclude that arterial stiffness contributes to the association between home BP variability and CVD incidence.


INTRODUCTION
A majority of studies on home blood pressure (BP) readings have reported that BP readings measured at home are associated with an increased risk of cardiovascular disease (CVD) events independent of office BP. [1][2][3][4][5][6] Therefore, recent guidelines recommend the diagnosis and management of hypertension using home BP monitoring. [7][8][9] In addition, BP variability assessed by home BP, which has been recognized as day-by-day home BP variability, has also been reported to be associated with organ damage and CVD events independent of or beyond mean home BP levels. [10][11][12][13][14] The relation between BP variability and CVD events has not been fully elucidated, but arterial stiffness has been studied as a potentially potent factor involved in this relation. Increased BP variability has been reported to be associated with greater arterial stiffness, and vice versa. [15][16][17] As a result, the vicious cycle of increased BP variability and increased arterial stiffness may lead to a risk for CVD events. However, despite the potential importance of such a cycle to the incidence of CVD events, no studies have been performed to verify this proposition.
Using data from the Japan Morning Surge-Home Blood Pressure (J-HOP) Study, which enrolled patients with a history of and/or risk factors for CVD and measured their home BP using a validated device in the morning and evening over a 14-day period, we investigated whether the association between home BP variability and a risk of CVD differs according to increased brachial-ankle pulse wave velocity (baPWV). The baPWV has been noninvasively measured and used as an index of arterial stiffness in clinical settings and is related to the incidence of CVD events. [18][19][20] We hypothesized that the day-by-day home BP variability may be more useful to predict CVD events in patients with increased baPWV than those without.

Study population
We used baseline data from the Japan Morning Surge-Home Blood Pressure (J-HOP) Study. The detailed information of the J-HOP study has already been published 13  All home BP data were automatically stored in the memory of the BP device and were downloaded to a computer by a physician or nurse at the time of the patient's clinic visits.
We calculated and used the standard deviation (SD SBP ), the coef- Healthcare). 20 The baPWV was measured by trained investigators in a quiet and temperature-controlled laboratory after the patient rested for 5 min in the supine position. 23

Outcome ascertainment
Vital status was ascertained through May 2018, with an average follow-up period of 4.4 years (11690 person-years). The primary outcome of this study was a composite of incident CVD events that included fatal or nonfatal coronary heart disease (hospitalization for myocardial infarction and angina pectoris requiring cardiac revascularization) and fatal or non-fatal stroke.

Statistical analyses
Data are expressed as the mean±SD, the median and interquartile range (IQR), or a percentage. Continuous variables were compared using Student's t-test, and categorical variables were compared using the chi-squared test between groups. We used the average of the right and left baPWV values for this analysis, and we divided the patients by baPWV into lower baPWV and higher baPWV groups with a cutoff level of 1800 cm/s according to previous studies. [24][25][26] After dividing baPWV into lower and higher groups, the Kaplan-Meier curve of CVD event-free survival for quartiles of SD SBP , CV SBP , and ARV SBP in the lower and higher baPWV groups was calculated. Using Cox proportional hazards models, the hazard ratios (HRs) and 95% confidence

RESULTS
The patients' clinical characteristics according to the groups which were included or not are summarized in Supple. Table 1. Home BPs of the included population were higher than those of the excluded population, but other characteristics were similar in both populations.
The patients' demographic and clinical characteristics according to the groups with lower (< 1800 cm/s) and higher (≥ 1800 cm/s) baPWV values are summarized in Table 1 of the higher baPWV group. Figure 1 shows the incidence rate of CVD events in the cohort separately for the divisions into quartiles of SD, CV, and ARV. Although a stepwise increase in observed CVD events was seen with increasing quartiles in both the groups with lower and higher baPWV, the trend in the group with higher baPWV was significant (all P for trend < .01), but the trend in the group with lower baPWV was not (the P for trend were .038, .149, and .111 in SD, CV and ARV, respectively). We then performed Kaplan-Meier survival analysis across quartiles of the indexes of day-by-day home BP variability in the lower and higher baPWV groups. The results of the log-rank test indicated that the indexes of day-by-day home BP variability were significantly associated with CVD incidence in the group with higher baPWV ( Figure 2). In the Cox proportional hazards model adjusted for other covariates that included average home SBP, the indexes of dayby-day home BP variability were incrementally associated with CVD incidence from the lowest quartile to the highest quartile in the group with higher baPWV, whereas this association was not found in the group with lower baPWV ( Table 2). Even after adding adjustment for log-transformed NT-proBNP, these associations remained. In the secondary analysis, after we divided patients into the quartiles of SD SBP , CV SBP , and ARV SBP , we divided them into groups of lower and higher baPWV, and then performed a similar analysis. The results of CVD incidence (Supple. Figure 1) and the risk of day-by-day home BP variability in Cox proportional hazard models (Supple. In addition, the highest quartiles of SD SBP , CV SBP , and ARV SBP were significantly associated with CVD incidence in the group with higher baPWV (all P < .05), but not in the group with lower baPWV (Table 3).
There was an interaction between CV SBP and CVD events according to the lower and higher baPWV group (P = .047). When we performed a Cox proportional hazards model for each day-by-day home BP variability per 1SD as a continuous variable for CVD events in the group with lower and higher baPWV adjusted for covariates including average home SBP (model 1) and plus log-transformed NT-proBNP (model 2) (Table 3), the SD SBP , CV SBP , and ARV SBP per 1SD were significantly associated with CVD incidence in the group with higher baPWV (all P < .05), but not in the group with lower baPWV.
In mediation analysis, the proportions of the association between each index of BP variability and CVD events mediated by higher baPWV were −1.0%, −0.8, and −2.4% for SD SBP , CV SBP and ARV SBP , respectively. Data are mean±SD, median (IQR), or percentage. ARV indicates average real variability; baPWV, brachial-ankle pulse wave velocity; CV: coefficient of variation; CVD, cardiovascular disease; DBP, diastolic blood pressure; IQR, interquartile range; NT-proBNP, N-terminal pro-B-type natriuretic peptide; SBP, systolic blood pressure; SD, standard deviation; PR, pulse rate.

DISCUSSION
The results of our present analyses demonstrated that day-by-day home BP variability was associated with CVD events in the group with higher baPWV values, independently of traditional cardiovascular risk factors including office BP and average home BP in this practitionerbased population with more than one cardiovascular risk, whereas this association was not found in the group with lower baPWV values. In the previous reports, increased day-by-day home BP variability was associated with CVD events. We previously reported that dayby-day home BP variability was associated with CVD events in the population of the J-HOP study. 12 Another study also demonstrated TA B L E 2 Hazard ratio of the quartile of BP variability for cardiovascular events in the population with lower or higher baPWV We divided into quartiles by SBP variability after dividing into two groups by baPWV. Adjusted factors for age, sex, body mass index, diabetes, totalcholesterol, high-density lipoprotein cholesterol, smoking, alcohol, pre-existing cardiovascular disease, the use of antihypertensive drug and statin, office SBP, and average home SBP. ARV indicates average real variability; baPWV, brachial-ankle pulse wave velocity; CI, confidence interval; CV, coefficient of variation; HR, hazard ratio; SBP, systolic blood pressure; SD, standard deviation.
that home BP variability assessed by SD SBP was associated with CVD events in 2455 patients without CVD risk from a general Japanese population. 11 Concerning the association between baPWV and CVD incidence, Takashima and colleagues. reported that higher baPWV (≥1800 cm/s) was significantly associated with an increased CVD risk in the general Japanese population. 25 In several guidelines, carotidfemoral pulse wave velocity (c-fPWV) has been recommended as an index of arterial stiffness. 30 The value of c-fPWV as an index of organ damage is more than > 10 m/s. 31 According to previous studies, a baPWV value of ≥1800 cm/s (18 m/s) is equivalent to a c-fPWV value of 10 m/s, [24][25][26]32,33 which was associated with organ damage 31 and all-cause and CVD. 34 Thus, the prognostic impact of increased BP variability and arterial stiffness for CVD incidence has been shown separately. Our findings thus provide the missing link between day-by-day home BP variability and arterial stiffness for CVD incidence. The findings of a previous study support our present findings. In a previous study about the association of BP variability assessed by ambulatory blood pressure monitoring and cardiovascular mortality, a higher ARV of 24-h DBP was significantly predictive of cardiovascular mortality in untreated hypertensive but not normotensive patients. 35 It has been considered that untreated hypertensives have greater progression of arterial stiffness compared with normotensive patients.
In the present study, the association between day-by-day home BP variability and CVD incidence remained after adjustment for NT-proBNP. We previously reported that home BP variability was asso- and NT-proBNP were only measured at baseline. Further studies will be needed to determine whether increased NT-proBNP as a result of the coincidence of higher day-by-day home BP variability and arterial stiffness is directly related to CVD incidence.

Limitations
This study has some limitations. First, because this was an observational study, we were unable to determine causality in the findings.
Second, we assessed a practitioner-based population with at least one cardiovascular risk factor; our findings may not be applicable to populations with lower cardiovascular risk or generalizable to other racial/ethnic and practical groups, and they must be verified in other groups. Third, higher home BP variability did not significantly increase the risk of CVD events in the lower baPWV group, which may have

CONCLUSIONS
Day-by-day home BP variability was associated with the CVD events in a group with higher baPWV values but not in a group with lower baPWV values in a practitioner-based population with at least one cardiovascular risk factor.

PERSPECTIVES
Several hypertension guidelines [7][8][9] indicate that the BP level and cardiovascular risk are appropriate targets of treatment and that BP variability predicts the prognosis of CVD, but the clinical implication of the vascular component for the management of hypertension has not been clear. Our present results demonstrate that the prognostic impact of home BP variability for CVD events differs according to the degree of arterial stiffness. Clinicians may therefore need to consider arte-rial stiffness in addition to BP variability when treating hypertension.
However, there are no randomized control trial data to support this assertion. In the future, additional studies will be needed to determine whether the suppression of both arterial stiffness progression and BP variability provides a synergistic effect in reducing CVD events that exceeds the effects of targeting either arterial stiffness or BP variability alone.

SOURCES OF FUNDING
This study was financially supported in part by a grant from the