The circadian blood pressure variability: There is a signal in the noise

1.

For centuries, we acknowledge the existence of physiologic rhythms and their close relationship with health and life.1, 2 In particular, the circulatory system shows a complex and mutable rhythmic pattern that ultimately continuously perfuse all body tissues.1, 3 Systolic and diastolic blood pressures oscillates over short‐ (eg beat‐to‐beat, day‐to‐night) to long‐term (eg visit‐to‐visit, seasonal) periods due to mechanical (eg abdominal and thoracic pressures during ventilation), neural central or peripheral (eg sympathetic and parasympathetic activity), and humoral (eg renin‐angiotensin, insulin‐glucagon, melatonin secretion) quasi‐periodic physiologic events, among others.1, 2, 3, 4, 5 Those oscillations are superimposed to non‐periodic events that have effects on the blood pressure (eg behavioral, environmental) or are even unrelated to the circulatory function itself (eg measurement artifact, device precision).2 Collectively, the blood pressure variability (BPV) has both deterministic and indeterministic information6 about the functioning of the circulatory system and its interaction with other systems and the environment, as well as unwanted information—the challenge remains distinguishing them. Clinically, BPV shows the extent to which the body adapts to stressful conditions such that those rhythmic patterns are linked to the pathophysiology of cardiovascular diseases (CVD).1, 2, 3, 4, 5

Ambulatory blood pressure monitoring (ABPM) has been used since the late 60s7 for estimating BPV and advances in biomedical instrumentation and digital signal processing allied to the continued clinical interest in this subject helped increasing its use.8 The recent study by Tadic et al9 highlights this by analyzing ABPM data of the Pressioni Arteriose Monitorate E Loro Associazioni (PAMELA) study10 to investigate the relationship of cognitive function with blood pressure and BPV in the general population. The cognitive function was assessed using the Mini‐Mental State Examination,11 with a cut‐off value of <24 points to classify the population as showing worse cognitive function. ABPM collected each 20 minutes over 24 hours was used to estimate traditional parameters of BPV, namely the standard deviation (SD) and the coefficient of variation of systolic and diastolic blood pressures. They also estimated the individual residual variability 12, 13 as the sum of the squared differences between the individual ABPM recording and the sum of frequency components that accounts for ≥95% of the systolic and diastolic pressures SD obtained from the population‐averaged ABPM recordings. Their analyses revealed, among other results, that patients with worse cognitive function have higher individual residual variability that cannot be explained by those cyclic patterns, except for subjects aged >75 years. Their findings support an inverse relationship between blood pressure and cognitive function in the general population and that the individual residual variability would be an interesting variable to follow‐up in the clinical setting.

Motivated by the aforementioned study,9 we herein promote an in‐depth discussion of the methodological aspects9, 12, 13 for estimating the individual residual variability. First, the population‐average of ABPM—aka the coherent average6—increases the signal‐to‐noise ratio of time‐series that are synchronized, time‐invariant, and corrupted with uncorrelated white noise. Second, the coherent average performs a filtering operation such that high‐frequency components are attenuated. Third, the linear curve‐fitting of the Fast Fourier Transform frequency components leads to a perfect oscillatory pattern that just simplifies the actual ABPM. Finally, the individual residual variability is estimated after removal of only two frequency components ascribed to day‐night and preprandial‐postprandial oscillations,12, 13 accounting for ~50% of the variance of the systolic and diastolic blood pressures SD in Tadic et al’s study.9

Such discussion provides new insights into the clinical utility of ABPM as follows. First, the between‐subjects synchronization assumption cannot be met in an absolute time frame but can be fairly made provided a strict data acquisition protocol is adopted. The small variations in the actual timestamp though lead to further low‐pass effect due to phase‐shifting of the signals. It would be interesting to investigate whether pre‐processing techniques for time‐regularization of the sampled data would affect the residual variability. Second, the time‐invariant assumption also holds due to the quasi‐periodic nature of most physiologic rhythms, which leads to the interpretation that the two frequencies removed are the signal and the residual is the noise regarding the circulatory rhythms. However, because “ruling out” those two quasi‐periodic events explains only ~50% of the blood pressure SD, the remaining frequency components conveys other events that could change the BPV. Therefore, there is a signal in the noise that is clinically correlated to cognitive function. Nonetheless, actual noise leading to higher BPV can be identified from at least two sources: the manual edition that yields a variable number of blood pressure measurements being averaged at each time stamp; and the current device's precision14 as well as its decay15 with time.

BPV can be estimated by a variety of parameters, apparently, not all parameters hold clinical relevance for predicting morbidity and mortality from CVD.4, 5, 16 Tadic et al9 discussed that “fast” blood pressure changes might explain the individual residual variability that is neither too low (day‐night and preprandial‐postprandial oscillations) or too fast (if manual edition and device's precision are considered the true uncorrelated white noise). In this sense, their study9 provides very interesting supporting evidence on the use of the individual residual variability as a marker of cognitive function in the general population. Herein we stress that further research is warranted to unveil other physiologic, behavioral, and/or environmental events that play an important role in explaining the relationship between cognitive function and BPV.

DISCLOSURE

The authors report no specific funding in relation to this research and no conflict of interests to disclose.