Caveat utilitor: take measure of your marker

Vascular sympathetic activity is critical to cardiovascular homeostasis, and chronic elevations in sympathetic activity are associated with deleterious cardiovascular outcomes (Malpas et al. 2010). Thus, the ability to measure sympathetic outflow is essential to understand human physiology and to determine the severity of pathophysiology. Although muscle sympathetic nerve activity (MSNA) measured via microneurography is a ‘gold standard’ for assessment of sympathetic outflow in humans, the intensive process required to procure fidelious recordings limits its utility for larger research studies and in clinical settings. Hence, for years researchers have sought to identify non-invasive markers of sympathetic activity in humans.

In the late 1800s, Mayer observed the development of slow blood pressure fluctuations consequent to hypovolaemia (and presumed sympathetic activation) in rabbits. Over the succeeding years, with the advent of rapid quantification techniques, low frequency (∼0.1 Hz) fluctuations in pressure have been used presumptively as a non-invasive measure to quantify sympathetic vasomotor activity. More recently, blood pressure oscillations have been used to signify the integrity of autonomic regulation in patients populations. However, some researchers have seriously challenged the utility and validity of low frequency blood pressure oscillations as a surrogate marker of sympathetic activity in individuals and patient populations (Taylor et al. 1998)

Surrogate markers can be important tools for a broad range of research – from laboratory studies to large scale clinical trials. However, validation of a candidate marker is critical to ensure its reliability for application and its appropriate interpretation. Surrogate markers may be used (and misused) to reflect underlying physiological processes within individuals or used to predict outcomes on a population level; however, predictive markers do not necessarily reflect an underlying physiology. For instance, heart rate variability measures provide limited information on parasympathetic activity at an individual level, but do retain prognostic significance at a group level. Thus heart rate variability may be best described as a statistical tool that predicts outcomes within a population, but that does not provide clear insight to the underlying physiology. Conversely, MSNA is frequently used as a measure of generalized sympathetic activity to the vasculature. Though it does display interindividual variability, it remains stable in individuals over time, correlates with other physiological measures of sympathetic activity (e.g. noradrenaline spillover) and responds predictably to physiological stimuli (e.g. nitroprusside challenge), making it a useful tool reflective of sympathetic outflow within subjects and across groups (Vallbo et al. 2004).

This issue reports findings by Ryan et al. (2011), who sought to inform the controversy surrounding the utility of low frequency blood pressure oscillations as a surrogate marker for sympathetic activity. The authors analysed changes in low frequency (0.04–0.15 Hz) blood pressure oscillations in response to application of progressive lower negative body pressure (LBNP) – a stimulus that increases sympathetic activity as measured by MSNA. As expected, MSNA increased proportionally with progressive LBNP. Low frequency blood pressure fluctuations also increased with LBNP, but only at the highest levels. The authors found that the relationship of changes in low frequency pressure oscillations to changes in MSNA was curvilinear when expressed at the group level, yet highly inconsistent when examined on a subject by subject basis. Hence do these data support use of low frequency blood pressure oscillations as a surrogate marker?

The authors rightly conclude that low frequency oscillations of systolic arterial pressure (SAP_LF) should not be used as a measure of sympathetic activity: ‘Because of the wide inter-individual variability and the very weak SAP_LF–MSNA association in a substantial proportion of the population, the use of SAP_LF as a non-invasive surrogate of MSNA is not reliable and therefore not warranted.’ This highlights the way group level data can be misleading. One could interpret the findings as evidence that low frequency blood pressure oscillations have some usefulness as a marker, with the caveat being it might not reflect the physiology in a strictly linear fashion. But, this population based relationship is analogous to the prediction of an outcome. That is, if sympathetic activity is high enough, we can expect, on average, to see greater low frequency blood pressure oscillations. But, it does not allow for the use of low frequency blood pressure oscillations as a measure for sympathetic activity. Indeed, even within the group averaged data, the relationship fails; low frequency blood pressure oscillations remain essentially unchanged with sympathetic outflow from resting up to three times resting. On the other hand, despite its inability to provide clear insight to the underlying physiology, the amplitude of low frequency blood pressure oscillations may be analogous to heart rate variability and future work may show it can predict outcomes within a population.

This article highlights important issues regarding the derivation, application and interpretation of physiological signal analyses for the measurement of autonomic activity. Our limited ability to directly record central autonomic activity in humans necessitates use of surrogate markers. While current technology has allowed for and assisted in the collection and analyses of quick, cheap and non-invasive ‘physiological’ measures, careful attention needs to be paid to the methods and analyses used to validate these measures – particularly in an age with prepackaged software that has enabled the proliferation of signal analyses.