Evolution of isolated systolic hypertension with normal central blood pressure in adolescents—prospective study

Spurious hypertension (sHT) is a phenotype of primary hypertension (PH) defined as elevated office and ambulatory systolic blood pressure (SBP), yet maintaining normal central systolic blood pressure (cSBP) and presents clinically as isolated systolic hypertension (ISH). sHT was originally described by O’Rourke et al. in a group of 6 adolescent and young adult males aged 14–23 years who all had office ISH with normal cSBP, none of whom presented with hypertensive target organ damage (TOD) [1]. In most reports sHT was described mainly among males. The more recent large prospective Chicago Heart Association Detection Project in Industry Study with 30 years of follow-up demonstrated that adult males but not females with ISH (cSBP was not assessed) had a low risk of cardiovascular events and cardiovascular death, compared with subjects with high-normal blood pressure (BP) [2].

Because ISH is the dominant hemodynamic phenotype in children with PH and males dominate among hypertensive children, it is important to assess the risk associated with ISH and normal cSBP in youth with PH.

To date, only a few studies have assessed the prevalence of sHT among adults with ISH and its clinical relevance. Studies by Saladini et al. and Palatini et al. showed that sHT in young adult males is a mild condition and patients with sHT are at a low risk of developing sustained hypertension (15.2%), similar to the normotensive population (14.7%). Thus, it was proposed that subjects with sHT and low cardiovascular risk do not require extensive diagnostics and pharmacological treatment [3, 4]. The importance of cSBP was documented in adult studies showing that cSBP better correlated with TOD than with peripheral BP [5‐8].

Although ISH is the predominant hemodynamic phenotype of PH in children and young adults, there are only single reports on the prevalence and association of sHT and TOD among adolescents. Previously we showed that 35% of children with office and ambulatory (ABPM) hypertension had brachial systolic hypertension with normal cSBP, i.e., fulfilled criteria of sHT and had a lower risk of TOD [9]. They also had lower body mass index (BMI) and waist circumference (WC) than those with “true” hypertension, i.e., both with elevated brachial and central blood pressure. It was also found that cSBP positively correlated with left ventricular mass index (LVMi), carotid intima-media thickness (cIMT) and pulse wave velocity (PWV). cSBP also had a greater predictive power than 24-h SBP in predicting LVH [9]. This suggests that adolescents with sHT may be at a lower risk of developing TOD compared with patients with both elevated brachial and central blood pressure. However, there is no prospective data on the evolution of sHT in adolescents and the risk of development of TOD later in life.

The aim of our study was to analyze the longitudinal changes of office, 24-h ABPM, cSBP, hemodynamic and TOD parameters over a 1-year follow-up period in a group of children with sHT.

This study was approved by the Bioethical Commission of the Children’s Memorial Health Institute. All patients and their parents gave informed consent. The study meets the criteria of the 1975 Declaration of Helsinki revised in 2013. The study subjects were recruited from a cross-sectional, previously reported study [9]. All consecutive patients who were referred in years 2015–2019 because of arterial hypertension and in whom ultimately PH was diagnosed were included in this study. All patients underwent full diagnostic evaluation according both to local and the pediatric guidelines of the European Society of Hypertension, including anthropometric measurements and biochemical tests, assessment of office blood pressure, 24-h ABPM, hemodynamics, echocardiography, cIMT, PWV, and pulse wave analysis (PWA); secondary causes of arterial hypertension were excluded [10‐12].

At baseline, out of a total of 294 consecutively referred patients with PH, 127 (43%) patients were diagnosed with white coat hypertension (WCH) and 29 (10%) patients with ambulatory prehypertension (AmbPreHT). In the group of patients with both office and 24-h ambulatory hypertension (n = 138; 31 girls; 22%), 48 (35%) patients (7 girls; 15%) presented with sHT phenotype (office and ambulatory ISH, but normal cSBP) and 90 (65%) patients had true hypertension (tHT; elevated office, ambulatory and cSBP). Out of 48 patients with sHT, 43 (6 girls; 14%) with median age 16.7 (8.5–17) years were followed for 12 ± 3 months during non-pharmacological therapy and were included in the current study; 5 were lost to follow-up (Fig. 1). Non-pharmacological therapy was based on uniform dietary advice and prescription of moderate to vigorous physical activity for at least 60 to 90 min daily.

All the tests were performed at baseline and after 12 ± 3 months.

Calculation of power size to detect change in effect size of 0.5 in longitudinal analysis in paired test showed 89% power in paired test for 43 patients. However, it was lower for comparison 10 vs. 33 patients in unpaired test. Thus, additional tests (Cohen effect size) were done to improve sensitivity of statistical analysis.

cSBP and 24-h ABPM were expressed in absolute values as mmHg. We also calculated the index of the patient’s 24-h SBP and DBP value to the upper limit of normal (95th percentile); an index > 1.0 represents abnormal/elevated values in relation to the 95th percentile of normal values. Twenty-four-hour mean arterial pressure (MAP) values were presented as absolute (mmHg) and standard deviation score (SDS) values. SDS values were calculated using LMS method and were based on normative ABPM data published by Wuehl et al. [25]. The change of measured parameters over time was assessed by delta value expressed as the difference between second and first value. BMI and WC were expressed in absolute values and as SDS based on Polish normative values for age and sex [26]. Similarly, cIMT and PWV values were analyzed as SDS.

Independent measures of two subgroups were compared using an unpaired t test (normally distributed variables); non-parametric variables were presented as medians and IQR (Tables 2 and 3), but their distribution was normalized by using cube root (power to 1/3) transformation (in order to transform negative and positive data values), and all variables were subsequently compared with a parametric unpaired t test. This allowed us to calculate and compare Cohen d effect size (expressed as SD) of all (parametric and non-parametric variables), which, unlike significance tests, is independent of sample size. Cohen d ≤ 0.2 can be considered a “small” effect size, 0.5 represents a “medium” effect size, and 0.8 a “large” effect size [27]. Three variables with the highest Cohen effect size were then included into the multiple regression analysis in order to assess determinants of cSBP change (increase/decrease) over time.

The threshold of a significant p value (usually < 0.05) was adjusted for multiple testing using Bonferroni correction, by dividing the obtained p values by the number of performed tests (indicated below each table). Statistics were performed using Python (version 3.7.0., packages TableOne and Pingouin) [28].

After 12 ± 3 months, patients’ height increased, whereas BMI-SDS decreased. There were no significant changes during the observation period in biochemical and TOD parameters and the 24-h SBP, DBP, and MAP decreased over time (Table 1, Table 1S).

Twenty-seven patients (63%) increased their cSBP over study period, 12 (28%) patients had an elevated cSBP > 1.0 after 12 months of follow-up (Fig. 2), but the median cSBP was still below the upper limit of normal at baseline and after 12 months (Table 1). Patients who increased their cSBP (n = 27) also increased heart rate, cardiac output, PWV SDS, cholesterol and serum uric acid concentrations (Table 2). The change of uric acid concentration was associated with the highest/large effect size (Cohen d = 1.05) on the change of cSBP, the change in following variables had a medium to large effect size (Cohen d = 0.5 to 0.8) on the change of cSBP: heart rate, cardiac output, PWV SDS, PWV, cPP, AugPress, cardiac index, weight, LVMi, BMI, stroke volume, and cholesterol (Fig. 3). Moreover, there was significant correlation between change of uric acid serum levels and change of heart rate (r = 0.030; p = 0.04, Spearman test).

Ten of 43 patients (23.2%) developed tHT with elevated 24-h ABPM and cSBP values after 12 months, 22 evolved to WCH or normotension, and 11 maintained their status of sHT.

Comparison of baseline characteristics of 10 patients who developed tHT with another 33 patients who had maintained their cSBP or evolved to WCH or normotension, showed that patients who developed tHT significantly increased their office and 24-h ABPM values, and CO and UA levels (Table 3). While the statistical significance of change of other parameters relative to the change of sHT status was borderline, the change of CO and UA was associated with a large effect size (Cohen d > 0.8), whereas the effect size of the change of 24-h, peripheral and central pulse pressure, HDL-cholesterol, triglycerides, PWV, and cardiac index was medium to large (Cohen d = 0.5–0.8) (Table 3). Differences in sex distribution between groups did not attain statistical significance. However, numerically there were only 9% of females in the group which maintained or normalized their blood pressure status as compared with 30% in the group which progressed to tHT.

In the multiple regression analysis we did not analyze BP-related parameters (SBP, DBP, MAP), since cSBP is derived from measurements of peripheral BP; i.e., the inherent relationship between peripheral and central BP may represent a potential bias in the prediction of cSBP. The multivariate regression analysis of the top three non-BP-related parameters with the highest effect size on univariate analysis (UA, HR, and CO, Fig. 3), showed that only the change of serum UA concentrations was a significant predictor (β = 0.225, 95% CI = 0.018–0.432, p = 0.03) of the change of cSBP.

The main finding of our study is that 23% of adolescents with ISH and normal cSBP who were prescribed only non-pharmacological therapy developed tHT after 12 months. The other group of 33 (77%) patients maintained (11 subjects) their sHT status or normalized peripheral and central BP (22 subjects). The main determinant of change in BP status was the change in serum UA concentration. These findings suggest that ISH and normal cSBP in adolescents is a heterogenous condition, and in a significant proportion of subjects is associated with the risk of development of sustained hypertension and hypertensive TOD, thus requiring close monitoring. These are novel findings, based on analysis of both peripheral and central blood pressure, and to our knowledge, this is the first prospective pediatric study which includes only adolescents and not adults, and included the greatest number of sHT cases described so far.

Starting from the first report of O’Rourke et al., several studies on ISH with normal cSBP were conducted in so-called young adults and included both adolescents and adults [1]. Mahmud and Feely assessed the prevalence of this phenomenon in a group of 174 medical students (aged 23 ± 0.5 years; 87 women) and 22 young hypertensive male adults aged 24.8 ± 0.88 years. ISH with normal cSBP was diagnosed in only 11 out of 174 (6.3%) of otherwise healthy normotensive students (all males) [29]. This is a lower prevalence of this phenotype of BP than in our previous cross-sectional study, in which we found ISH with normal cSBP in 35% of adolescents from the group of consecutive patients referred because of PH. Moreover, in our study, ISH was confirmed by office and ABPM. Mahmoud et al. provided data on 6/11 patients with sHT after 2 years of follow-up. All 6 patients maintained their status of ISH with normal cSBP, but only 2 patients had echocardiography performed showing no LVH [29].

The finding that 23% of subjects in our study increased their cSBP, i.e., became true hypertensive, is higher than in prospective studies of young adults, in whom the risk of developing sustained hypertension needing treatment was rather low (15.2%) and similar to the control group (14.7%) [3]. In both adolescents in the current study and 33 young adults with mean age of 33 years in the study of Saladini et al. [3], most subjects were males (86% vs. 94%, respectively). Because of the low number of subjects in each group, we did not find any statistically significant differences in sex distribution between those who maintained or normalized their BP status and those who increased cSBP, but numerically there were more females in the group which increased cSBP (9 vs. 30%, respectively). The 3 times higher percentage of females in the group which increased cSBP is in agreement with both our previous study (24.4%) and the overall sex distribution among adolescents with PH [9].

Saladini et al. found that among young adults with ISH and normal cSBP the long-term risk of developing sustained hypertension and needing treatment seems to be directly related to the level of cSBP, as 50% of patients with cSBP above normal at baseline developed sustained hypertension (stage 2) as compared with only 15.2% of patients with a lower cSBP at baseline [3]. Our 43 patients had similar baseline absolute values of cSBP (median = 116 mmHg) compared with patients from the Saladini et al. study (113.8 mmHg) [3]. However, median cSBP at the end of a 1 year follow-up was 120 mmHg (median increase by 4 mmHg per year). More importantly, the increase of cSBP was significantly higher in patients who developed tHT (median change = 11 mmHg/year) compared with patients who developed WCH or ambulatory prehypertension or those who maintained sHT (median change = 1 mmHg/year) (Table 2S). It may be hypothesized that over a longer period of time (10 years as in the study by Saladini et al.), some of our patients, mainly those who increased their cSBPi over 1 year (63%), would develop sustained hypertension. In addition, our patients were younger by almost 18 years, which suggests that earlier elevation of brachial blood pressure even with normal cSBP is associated with earlier development of sustained hypertension. However, only long-term observation time can prove this hypothesis.

Mild TOD, mainly in the form of increased PWV was found in 7 of 43 (16%) patients at study start. As reported previously, patients with ISH and normal cSBP had lower prevalence of LVH and lower values of cIMT and PWV than those with elevated cSBP [9]. The finding of mild forms of TOD in subjects with brachial hypertension and normal cSBP is not surprising because cIMT and LVMi increase with change of blood pressure status from normotension to severe ambulatory hypertension and cSBP was found to be one of the predictors of LVMi [30]. Although subjects included in our study had normal cSBP values, they were above median values [17].

The main determinant of change of cSBP over time was the change of serum UA concentrations. The role of UA in the pathogenesis of PH is well described and elevated serum UA levels are typical for adolescents with PH [31, 32]. Moreover, the decrease of UA during treatment with allopurinol also lowered BP [32]. Previously we found that a change in UA was one of the predictors of regression of hypertensive TOD [33]. There are no reports on the association between serum UA and cSBP in hypertensive children. However, serum UA was found to be associated with the increased expression of MMP-14 in peripheral blood leucocytes of hypertensive children [34]. This may be a potential explanation for the link between an increase in UA and cSBP over time, because MMP-14 is involved in arterial remodeling and enhances MMP-2 activity, which has been found to be the most significant marker of matrix metalloproteinases/tissue inhibitors of the matrix metalloproteinase system activation in hypertensive children [34, 35]. Increased serum UA can also be associated with adrenergic drive [36]. The association of increased serum UA with an increased HR and CO, along with an increase of cSBP, suggests the role of sympathetic drive in the evolution from sHT to tHT. An association between CO and cSBP was also found in a recent study of hemodynamics in adolescents with PH [37].

In contrast to our previous studies we did not find any changes in BMI or WC, both expressed in absolute and standardized values, as predictors of cSBP change. However, it must be underlined that subjects from this study had relatively normal BMI and WC values which were close to the median of the population norm (BMI-SDS and WC-SDS both below 1.0) and had lower BMI and WC than patients with ISH and elevated cSBP, as reported previously [9].

According to O’Rourke’s hypothesis [1], ISH with normal cSBP in adolescent and young adult males is caused by an increased elasticity of middle-sized arteries, such as the brachial artery, when compared with adults. It allows for the accumulation of a backward pulse wave via the brachial artery with a rise in brachial SBP but cSBP and cPP do not change. This phenomenon occurs more often in boys and in taller adolescents. It is in contrast to ISH in elderly, where a rise of SBP is caused by increased stiffness of the arterial tree and faster return of the backward wave to the aorta. In our study, of all the patients studied, 41/48 (85%) were males. There were also more males (91% vs. 70%) in the group of patients who after observation maintained sHT or normalized their BP status. This group was also taller (177 cm vs. 169 cm); however, this difference is not significant and likely due to the limited number of study participants (Table 1).

Our study has several limitations. Firstly, although we prospectively analyzed the greatest number of patients with sHT reported so far, the number of study subjects was relatively low. Moreover, the role of sex was difficult to analyze due to the low number of females. Secondly, the observation period was limited to 1 year, which may not be long enough to develop any significant changes in hemodynamic and TOD parameters. However, as changes in the ABPM blood pressure status in adolescents are described [38], cSBP change over time is also to be expected. The other limitation is that we did not objectively assess adherence to non-pharmacological therapy. However, both mean BMI-SDS and WC-SDS values decreased after 12 months, which suggests adherence to non-pharmacological treatment.

The strength of our study stems from the fact that it is the first prospective pediatric study to include an unselected group of adolescents who were referred for work-up of PH and in whom a comprehensive assessment of hemodynamic parameters including ABPM, cSBP, TOD assessment, and laboratory investigations was done. Second, this is the largest prospective study to date analyzing the evolution of ISH with normal cSBP. Third, all subjects had all tests completed at baseline and after 1 year of follow-up. Fourth, although unselected and consecutively referred, our patients were relatively lean, non-obese, which allowed us to study the impact of factors not related to obesity.

Our findings document that ISH with normal cSBP is not benign and may progress to elevated cSBP in a significant proportion (23%) of patients. The main determinant of change of cSBP over time was the change of serum UA concentration. Thus, our results suggest that adolescents with ISH and normal cSBP are a heterogenous group and that metabolic abnormalities play a significant role in the rise of cSBP, even in the cohort of non-obese hypertensive children. cSBP is not routinely measured in hypertensive children and adolescents, and management depends on hypertension severity and presence of TOD. Although there is a relationship between brachial blood pressure and cSBP—the higher the brachial blood pressure, the greater the probability of an elevated cSBP—almost one third of our patients with severe ambulatory hypertension had normal cSBP [9]. Thus, assessment of cSBP may be a useful tool in selected cases with ISH, without hypertensive TOD and without metabolic cardiovascular risk factors.