1 INTRODUCTION

Management of nocturnal blood pressure (BP) is one of the important issue for preventing the progress of hypertensive target organ damage (TOD) and the onset of cardiovascular disease.1-3 Recent studies have demonstrated that the nocturnal BP measured by ambulatory BP monitoring (ABPM) is a stronger predictor of cardiovascular events (CVE) than daytime BP.4-9 Although ABPM has been the gold standard for assessing nocturnal BP, nocturnal BP measured by home BP monitoring (HBPM) also has high risk of future CVE10-14 and correlates with TOD as with ABPM.15 A new wrist-type nocturnal HBPM device that automatically measures BP in the supine position without reducing sleep quality has been developed, and it could be used to facilitate real-world sleeping studies.16, 17 In addition, the concern of an association between short-term BP variability (BPV) and cardiovascular disease has been growing.18 Nocturnal BPV, defined as the standard deviation (SD) of nocturnal BP measured by ABPM, is associated with risks of CVE.19

It is assumed that some pathological factors such as obstructive sleep apnea (SA), rapid eye movement (REM) sleep, and microarousal increase nocturnal BPV.20-22 In particular, obstructive SA (OSA) has demonstrated a critical impact on nocturnal BP level and BPV.23 We have developed an oxygen-triggered BP monitor and demonstrated that the maximum value of systolic BP (SBP) measured by the oxygen-trigger function is higher than mean nocturnal SBP measured by the intermittent oscillometric method.24-28 However, as this BP monitor is based on the oscillometric method, the peak of short-term BPV might be underestimated.

Consequently, we developed a continuous Beat-by-Beat (BbB) BP monitoring device using a tonometry method to detect the peak of the short-term BPV in detail29. Using this device, we have already observed “BP surge in seconds (sec-surge)”, which characterized as an acute transient BP elevation over several tens of seconds in OSA patients.2, 30 We have developed the tools for studies on sec-surges such as the device and an automatic sec-surge detection algorithm from BbB BP recordings overnight.31, 32 However, it is not clear whether the severity of sec-surge is different between sec-surges induced by SA (apnea/hypopnea detected by polysomnography (PSG) or oxygen desaturation) and those induced by non-SA factors (sympathetic nerve activity such as REM sleep, micro arousal), and the CVE risks of sec-surges.

Thus, this study aimed to assess the severity of the sec-surge between inducing factors and to assess the association between left ventricular hypertrophy (LVH) and sec-surge measured in a sleep laboratory setting.

2 METHODS 2.1 Study design and patients

In total, 48 outpatients were recruited for this study at the Washiya Hospital, Tochigi, Japan, from July, 2017 to February, 2019. Patients who met at least one of the following criteria were temporarily registered: (1) hypertensive patients, and (2) patients who had subjective symptoms of sleep apnea syndrome. Next, patients whose mean nocturnal SBP was ≥120 mm Hg, measured by ABPM or using a home BP monitor measuring BP automatically and intermittently, before 1 month of formal registration for the study were enrolled in this study. All registered patients underwent overnight full PSG, BP measurement using a BbB BP monitor worn on patient's left wrist and cuff-oscillometric BP monitor worn on the patient's right arm for two nights within a month. After wearing these devices given them by study staff members, they were requested to sleep overnight at the sleep laboratory. A total of seven patients who had atrial fibrillation, missing data of PSG/BbB BP measurement in both nights, did not have adequate length of BbB BP records in both nights, or did not have sec-surges in both nights were excluded from the study. In the end, 41 patients were included as study patients for assessing the difference in sec-surge severity between sec-surges induced by SA (apnea/hypopnea detected by polysomnography (PSG) or oxygen desaturation) and those induced by non-SA factors (sympathetic nerve activity such as REM sleep, micro arousal). Eighteen patients who had inspection results of cardiac magnetic resonance imaging (MRI) within a year out of the 41 were included as study patients for assessing the association between sec-surges and LVH. The detailed flow of study patients was shown in Figure S1.

The PSG was recorded by PSG-1100 (Nihon-Kohden) and analyzed by Polysmith (Nihon-Kohden). The cardiac MRI was performed by OPTIMA MR450w Expert 1.5T (GE Healthcare). The measurement of left ventricular mass (LVM) was analyzed by cardiacVX (GE Healthcare) and Vitrea (Canon medical systems Corp) following our recent study.33 The LVM index (LVMI) was calculated as LVM/body surface area using the Fujimoto formula. The patients were recruited from Washiya Hospital, and this study was approved by the institutional review board of the Jichi Medical University School of Medicine, where this study was performed. Written informed consent was obtained from all patients upon their recruitment into the study.

2.2 Development of beat-by-beat BP monitoring device

In the present study, overnight BbB BP was recorded by using the BbB BP monitoring device, based on the tonometry method34 we recently developed. Figure 1(A) shows a block diagram of the device. Pulse wave signals were obtained by 46 sensors in the “tonometry sensor unit35” (the appearance is shown in Figure 1(B)) directly placed on the skin above a radial artery, and were transmitted to the processor. Meanwhile, BP for calibration was measured when the “cuff-oscillometric BP measurement unit” received a triggering signal from the “calibration control” function in the processor. The timing of calibration was automatically judged by the function when contact between the tonometry sensor and the skin was significantly changed due to body motion. Once the processor received BP for calibration, the “calculation of calibrated BbB BP” function transforms an amplitude of pulse wave signal obtained at an active sensor into the calibrated BbB BPs using the values of BP for calibration. Calibrated BbB BPs were recorded to the memory of the device. The active sensor was automatically selected based on the maximum amplitude of the pulse wave signal among 46 sensors at each moment in the “active sensor selection” function. The details are shown in Figure 1(C). The heatmap indicates the amplitude of pulse wave signals obtained by 46 sensors as a colormap in timeseries. The moment of (a), (b), and (c) show the signal amplitudes obtained by 46 sensors before the sec-surge, peak of sec-surge, and after changing the active sensor, respectively. The selected active sensor was kept as 26 in the stable section including both (a) and (b), although the signal amplitudes were increased at (b) due to sec-surge. Even though the active sensor was changed from 26 to 16 by body motion, the maximum of signal amplitude and BbB BPs at (c) were maintained as (a). By selecting the active sensor automatically at each moment, the device could continuously monitor BbB BP without BP calibration, even though the contact between the tonometry sensor and the skin was slightly changed.

Overview of the BbB BP monitoring device. (A) A block diagram of the BbB BP monitoring device. (B) Tonometry sensor unit in the BbB BP monitoring device. (C) A typical case of overnight BbB BP measurement. BbB BPs were derived by the pulse wave signals at an active sensor and calibrated using cuff-oscillometric BP. The active sensor is automatically selected based on the maximum amplitude of the pulse wave signals among 46 sensors at each moment. The heatmap indicates the amplitude of pulse wave signals obtained by 46 sensors as a colormap in a time series. (a), (b), and (c) show the amplitudes of pulse signals obtained by 46 sensors at each moment before the sec-surge, peak of sec-surge, and after changing the active sensor, respectively. BbB indicates beat-by-beat; BP, blood pressure; and sec-surge, blood pressure surge in seconds

2.3 Definition of oscillometric BP variables

The office BP of the study patients was measured when they were recruited into this study. Conventional nocturnal BPs were measured using the arm-cuff-oscillometric BP monitor (HEM-7220; Omron Healthcare Co., Ltd), which measures both intermittent BPs (30 min intervals) and oxygen-triggered BPs. Oxygen-triggered BPs were measured when the oxygen saturation (continuously monitored by pulse oximetry) falls below a variable threshold.24-27 For further details of the definition of oscillometric BP variables, see Expanded Methods in the data supplement.

2.4 Definition of nocturnal beat-by-beat blood pressure surge in seconds (sec-surge)

The sec-surges were detected from overnight BbB BP recordings by an automatic sec-surge detection algorithm32 we recently developed. For the detailed flow of detecting sec-surge, see Expanded Methods in the data supplement.

We defined the sec-surge feature as shown in Figure 2. All sec-surge variables were calculated from the BbB SBPs between the start and end points of sec-surges (duration of sec-surge). The peak, start, and end point of the sec-surge were detected by the above-mentioned algorithm. The threshold of the amplitude of sec-surge (difference between the peak SBP and the start SBP) was ≥20 mm Hg. The peak point was detected as the local maximum of BbB SBPs by using a sliding window. The start point was detected as the final point during stable BbB SBPs in backward-searching ranges set before the peak point. The end point was detected as the point that SBP decreased by 75% of the amplitude of sec-surge. We defined the duration from start to peak as upward duration and from peak to end as downward duration. The peak of sec-surge and mean of sec-surge were defined as the maximum value of BbB SBP and the mean value of BbB SBP, respectively, during duration of sec-surge. The integrated values were calculated in upward, downward, and sec-surge durations as the sum of BbB SBPs in each duration. The dp/dt of the sec-surge was calculated in upward and downward durations as amplitude divided by the duration. Sec-surge index was defined as number of sec-surge events per effective analysis time. Each sec-surge variable was taken as an average of the detected sec-surges during the night. Furthermore, the mean, maximum, SD, coefficient of variation (CV), and average real variability (ARV) of nocturnal BbB SBPs were calculated from reliable BbB SBPs during the night. The maximum nocturnal BbB SBP was defined as the 95th percentile of BbB SBP (instead of the maximum value) to avoid noises. The mean and maximum value of cuff-oscillometric BPs used for BbB BP calibration (it was different from the abovementioned conventional nocturnal BP variables) were also calculated.

Definition of nocturnal blood pressure surge in seconds (sec-surge) variables. BbB indicates beat-by-beat; SBP, systolic blood pressure; and DBP, diastolic blood pressure

2.5 Labeling of sleep apnea and sleep stages to each sec-surge

To investigate the sec-surge features between sleep stages and between the inducing factors of sec-surges (SA-related sec-surges and non-SA-related sec-surges), labeling of sleep stages and sleep apnea for each sec-surge was conducted. Sleep stages, apnea, and hypopnea were automatically determined by the Polysmith. For further details of the labeling, see Expanded Methods in the data supplement.

2.6 Statistical analysis

BP variables and PSG-derived variables were calculated for each night. The average values of the two nights were used in the analysis. Continuous variables were expressed as mean±SD, and categorical variables were summarized as frequencies and percentages. In the analysis of sec-surge severity between inducing factors of sec-surges, Student's t-test was used to compare SA-related sec-surges and non-SA-related sec-surges in each sleep stage, and to compare nocturnal BP variables in severe OSA patients (AHI ≥ 30) and non-severe OSA patients (AHI < 30). In the analysis of the associations between sec-surge and LVH, correlations between LVMI and BP variables both sec-surge and conventional one were analyzed using Pearson's correlation coefficient. Furthermore, to investigate the contribution of sec-surge variables, multiple regression analyses were performed using a sec-surge variable and a conventional BP variable as independent variables. The partial regression coefficient β were compared after calculating z-score. Values of p < .05 in all statistical analyses were considered statistically significant. All statistical analyses were performed with R version 3.6.0.

3 RESULTS

The clinical characteristics of the study patients are shown in Table 1. In the analysis of the difference in sec-surge severity between sec-surges induced by SA (apnea or hypopnea) and those induced by non-SA factors (sympathetic nerve activity), two of the 41 patients were normal in the perspective of apnea severity (AHI < 5), eight had mild OSA (5≤AHI < 15), seven had moderate OSA (15≤AHI < 30), and 24 had severe OSA (AHI ≥ 30). Table 2 shows intermittent BPs and oxygen-triggered BPs measured by the cuff-oscillometric method, and sec-surge variables in each sleep stage. The peak of sec-surge was ≥ 20 mm Hg higher than the mean of nocturnal SBPs measured by intermittent oscillometric method (127.2 ± 14.7 mm Hg and 148.2 ± 18.5 mm Hg). A typical case of sec-surges is shown in Figure S4. Three sec-surges were induced repeatedly, and the peak of sec-surge reached almost 200 mm Hg from the baseline of 150 mm Hg. The distribution of the number of sec-surges per patient is shown in Figure S5. In the analysis of associations between sec-surges and LVH, the mean ± SD of LVMI in 18 patients who had LVM data was 68.9 ± 13.2 g/m2, and non-severe LVH.

TABLE 1. Clinical characteristics of the study patients Study patients for analyzing sec-surge severity between inducing factors (SA and non-SA) All (n = 41) AHI < 30 (n = 17) AHI ≥ 30 (n = 24) Study patients for analyzing the association between sec-surges and LVH (n = 18) Age (years) 63.2 ± 12.6 60.5 ± 14.9 65.2 ± 10.6 65.0 ± 12.5 Female, n (%) 12 (29.3) 7 (41.2) 5 (20.8) 6 (33.3) Body mass index (kg/m2) 27.3 ± 4.4 26.7 ± 4.8 27.7 ± 4.2 27.0 ± 4.6 Hypertension treatment, n (%) 33 (80.5) 14 (82.4) 19 (79.2) 15 (83.3) Diabetes, n (%) 6 (14.6) 4 (23.5) 2 (8.3) 2 (11.1) History of angina, n (%) 2 (4.9) 1 (5.9) 1 (4.2) 2 (11.1) History of myocardial infarction, n (%) 1 (2.5) 0 (0.0) 1 (4.2) 1 (5.6) History of stroke, n (%) 3 (7.3) 2 (11.8) 1 (4.2) 1 (5.6) AHI, events/h 33.8 ± 21.0 14.0 ± 8.0 47.8 ± 15.0* 30.9 ± 24.4 Arousal index, events/h 22.9 ± 16.3 14.2 ± 4.2 29.2 ± 18.8* 18.1 ± 10.4 SpO2 < 90 % 19.6 ± 17.4 10.0 ± 11.9 26.3 ± 17.7* 16.3 ± 19.2 Lowest SpO2, % 77.7 ± 8.8 81.7 ± 7.0 74.9 ± 9.0* 79.3 ± 9.4 Total sleep time (min) 423.6 ± 93.7 446.2 ± 83.8 407.6 ± 98.7 419.1 ± 99.4 Sleep efficacy, % 68.5 ± 16.0 71.9 ± 14.5 66.1 ± 16.9 68.6 ± 17.4 REM, % 11.4 ± 6.0 11.3 ± 5.7 11.5 ± 6.3 12.1 ± 6.1 Non-REM1, % 25.4 ± 11.6 17.9 ± 9.0 30.7 ± 10.3* 20.5 ± 9.7 Non-REM2, % 57.1 ± 11.7 64.1 ± 9.0 52.1 ± 10.9* 60.7 ± 11.9 SWS, % 6.1 ± 7.5 6.8 ± 7.7 5.7 ± 7.5 6.7 ± 7.1 Data are expressed as mean ± SD or frequency and percentage. Abbreviations: AHI, apnea hypopnea index; SpO2, oxygen saturation; REM, rapid eye movement; SWS, slow wave sleep. TABLE 2. Results of each nocturnal BP variable in each sleep stage (n = 41) Nocturnal BP variables Whole sleep period Wake REM Non-REM1 Non-REM2 SWS Intermittent oscillometric BP Number of BP measurements 16.7 ± 5.8 5.4 ± 3.9 1.2 ± 1.4 2.6 ± 1.7 6.9 ± 3.7 0.6 ± 0.8 SBP (mm Hg) 127.2 ± 14.7 129.9 ± 16.4 130.1 ± 18.7 127.5 ± 14.3 125.7 ± 15.4 116.1 ± 16.7 DBP (mm Hg) 77.6 ± 10.3 80.4 ± 12.0 79.6 ± 11.9 76.7 ± 11.1 76.8 ± 10.9 70.2 ± 9.7 PR (beats/min) 60.7 ± 8.9 63.3 ± 10.0 60.5 ± 9.9 60.6 ± 10.0 59.6 ± 8.2 59.2 ± 9.2 Oxygen triggered oscillometric BP Number of BP measurements 11.5 ± 14.2 2.2 ± 3.4 1.9 ± 3.4 2.9 ± 4.2 4.1 ± 6.6 0.3 ± 1.6 Hypoxia-peak SBP (mm Hg)a 147.6 ± 22.7 – – – – – Hypoxia-mean SBP (mm Hg) 130.4 ± 18.5 128.6 ± 16.8 129.3 ± 19.9 127.5 ± 21.0 127.1 ± 20.7 110.6 ± 9.3 Hypoxia-mean DBP (mm Hg) 79.7 ± 16.0 79.6 ± 16.6 78.6 ± 14.7 78.4 ± 12.3 78.4 ± 13.5 69.1 ± 10.9 Hypoxia-mean PR (beats / min) 63.4 ± 9.1 64.5 ± 9.5 61.8 ± 11.1 61.9 ± 9.6 62.1 ± 9.9 61.7 ± 4.7 Sec-surgeb Number of sec-surges All, events 36.0 ± 40.2 5.4 ± 8.2 2.8 ± 4.0 6.0 ± 7.6 19.5 ± 23.3 2.2 ± 6.6 Induced by SA, events 19.5 ± 26.0 3.3 ± 6.4 1.6 ± 2.4 4.1 ± 6.4* 10.1 ± 14.7 0.4 ± 0.7 Induced by non-SA factors, events 16.4 ± 29.8 2.1 ± 3.7 1.2 ± 2.7 1.9 ± 2.5 9.4 ± 17.7 1.9 ± 6.4 Peak of sec-surge All (mm Hg) 148.2 ± 18.5 149.5 ± 19.6 151.2 ± 19.4 147.3 ± 19.7 147.4 ± 18.9 140.1 ± 16.7 Induced by SA (mm Hg) 148.2 ± 18.5 147.6 ± 20.0 151.3 ± 22.2 147.7 ± 21.4 147.3 ± 19.1 139.8 ± 15.5 Induced by non-SA factors (mm Hg) 149.3 ± 19.2 151.1 ± 19.7 150.5 ± 24.4 145.8 ± 19.8 148.3 ± 20.0 139.7 ± 18.3 Amplitude of sec-surge All (mm Hg) 25.8 ± 4.8 26.2 ± 6.0 26.0 ± 4.4 25.6 ± 3.5 25.7 ± 5.5 24.0 ± 3.6 Induced by SA (mm Hg) 26.0 ± 4.3 25.4 ± 7.0 27.0 ± 7.2 26.7 ± 5.0* 25.9 ± 4.4 25.3 ± 3.2 Induced by non-SA factors (mm Hg) 25.8 ± 5.7 26.8 ± 6.5 25.5 ± 5.6 24.1 ± 3.0 25.6 ± 6.1 23.8 ± 4.1 Data are expressed as mean ± SD. SBP indicates systolic blood pressure; DBP, diastolic blood pressure. Abbreviations: PR, pulse rate; REM, rapid eye movement; SWS, slow wave sleep; sec-surge, blood pressure surge in seconds; SA, sleep apnea. 3.1 Comparison of sec-surge features between inducing factors

There was no significant difference in the peak of sec-surges between SA-related sec-surges and non-SA-related sec-surges in the whole sleep period (148.2±18.5 vs. 149.3±19.2 mm Hg) and each sleep stage (Table 2). Similarly, there were no significant differences in the number and amplitude of sec-surges between SA-related sec-surges and non-SA-related sec-surges (19.5±26.0 vs. 16.4±29.8 events/night and 26.0±4.3 vs. 25.8±5.7 mm Hg, respectively), except for the non-REM1 stage. The associations between SA-related sec-surges and non-SA-related sec-surges are shown in Figure S6. Both the peak and amplitude of SA-related sec-surge were significantly and strongly associated with those of non-SA-related sec-surges (r = 0.874, p < .01, n = 31 and r = 0.473, p < .01, n = 31, respectively). Table 3 shows the comparison of each nocturnal BP variable in the whole sleep period between severe OSA (AHI ≥ 30) patients and non-severe OSA (AHI < 30) patients. The number of sec-surges induced by the SA factor was significantly higher among severe OSA patients than among non-severe OSA patients (10.7 ± 10.3 vs. 25.8 ± 31.6 events/night). In contrast, that induced by non-SA factors was significantly higher among non-severe OSA patients than among severe OSA patients (29.9 ± 42.7 vs. 6.9 ± 6.8 events/night). There were no significant differences in the peak and amplitude of sec-surge between severe OSA patients and non-severe OSA patients (148.8±14.6 vs. 147.8±21.1 mm Hg and 26.3±6.0 vs. 25.4±3.9 mm Hg, respectively).

TABLE 3. Comparison of each nocturnal BP variable between severe and non-severe OSA patients (n = 41) Whole sleep period Nocturnal BP variables AHI < 30 (n = 17) AHI ≥ 30 (n = 24) Intermittent oscillometric BP N 17.5 ± 6.1 16.1 ± 5.7 SBP (mm Hg) 127.2 ± 12.1 127.2 ± 16.4 DBP (mm Hg) 78.3 ± 7.0 77.2 ± 12.1 PR (beats / min) 58.4 ± 8.3 62.2 ± 9.1 Oxygen triggered oscillometric BP N 5.3 ± 7.6 15.9 ± 16.2** SBP (mm Hg) 128.9 ± 14.1 131.2 ± 20.4 DBP (mm Hg) 76.0 ± 10.1 81.4 ± 18.0 PR (beats / min) 62.0 ± 8.7 64.0 ± 9.4 Sec-surge Number of sec-surges All, events 40.6 ± 47.7 32.7 ± 34.7 Induced by SA, events 10.7 ± 10.3 25.8 ± 31.6*,**, *,** Induced by non-SA factors, events 29.9 ± 42.7 6.9 ± 6.8** Peak of sec-surge All (mm Hg) 148.8 ± 14.6 147.8 ± 21.1 Induced by SA (mm Hg) 149.4 ± 16.7 147.6 ± 19.7 Induced by non-SA factors (mm Hg) 149.8 ± 14.4 149.0 ± 22.7 Amplitude of sec-surge All (mm Hg) 26.3 ± 6.0 25.4 ± 3.9 Induced by SA (mm Hg) 26.4 ± 3.9 25.8 ± 4.6 Induced by non-SA factors (mm Hg) 26.4 ± 6.6 25.3 ± 5.0 Data are expressed as mean ± SD. Abbreviations: SBP, systolic blood pressure; DBP, diastolic blood pressure; PR, pulse rate; REM, rapid eye movement; SWS, slow wave sleep; sec-surge, blood pressure surge in seconds; SA, sleep apnea; AHI, apnea hypopnea index. 3.2 Association between nocturnal BP variables and LVH

The BP measurements and correlations between LVMI and each BP variable were shown in Table 4. The mean nocturnal SBP measured using the oscillometric method and the mean nocturnal BbB SBP were almost comparable, and were distributed around the threshold for sleep SBP (120 mm Hg).36, 37 The mean of the oscillometric SBP for BbB BP calibration, the maximum of that, and the mean of nocturnal BbB SBP were significantly and strongly correlated with LVMI (r = 0.614, p < .01, n = 18; r = 0.635, p < .01, n = 18; and r = 0.492, p = .038, n = 18, respectively). The maximum of peak of sec-surge and the mean of that were also significantly and strongly correlated with LVMI (r = 0.579, p = .012, n = 18; r = 0.607, p < .01, n = 18). The mean of peak of sec-surge was correlated with the mean of oscillometric SBP for BbB BP calibration (r = 0.870, p < .01, n = 41). Even though the sec-surges were classified as SA-related sec-surges and non-SA-related sec-surges, significant and strong correlations were observed between peak of sec-surge and LVMI in both SA-related and non-SA-related sec-surges (r = 0.551, p = .041, n = 14 and r = 0.606, p = .017, n = 15, respectively). The integrated values calculated in sec-surge duration and downward duration had a marginally significant correlation with LVMI (r = 0.401, p = .099, n = 18 and r = 0.407, p = .094, n = 18, respectively). Although hypoxia-peak SBP measured by the oxygen-triggered oscillometric method was also significantly correlated with LVMI (r = 0.602, p = .038, n = 12), six out of the 18 patients did not have oxygen-triggered BP measurements. There were no significant correlations between other conventional BP variables and LVMI.

TABLE 4. Simple Pearson's correlations between LVMI and blood pressure variables in study patients for analyzing the association between sec-surges and LVH (n = 18) BP variables No. of BP measurements per one patient (Mean ± SD) Measurement (Mean ± SD) r p value Conventional oscillometric BP (n = 18) Office SBP (mm Hg) 2.0 ± 0.0 136.1 ± 15.7 0.256