Should baroreflex sensitivity be corrected for heart rate?

Introduction Abrahamsson et al. [1] propose to standardize a patients baroreflex sensitivity (BRS) to a fixed heart rate of 60 b.p.m. by regressing the logarithm of BRS on heart rate and reading BRS at 60 b.p.m. from the regression line. The data for the regression are obtained in response to bicycle ergometry. If training or treatment is going to improve BRS, but at the same time change heart rate, the situation is potentially confusing. Is BRS improved as a result of the heart rate change? Standardization at 60 b.p.m. might be the answer. We set this proposal against the paradoxical behaviour of the baro-reflex under stress and wonder if it is not too early for standardization. The baroreflex Arterial blood pressure is controlled in many ways, of which the effect of baroreflex on heart rate is the fastest acting [2]. Upon a change in blood pressure, the baroreflex performs various actions, and the effect on the sinus node of the heart is almost immediate. All actions are aimed at stabilizing blood pressure. The gain in the principally vagal heart rate control loop from blood pressure change to heart rate change is termed BRS, and was investigated by Abrahamsson et al. [1]. BRS is the amount of interbeat interval change in immediate reflex response to a blood pressure change. To test the effect of the baroreflex on heart rate and to measure its sensitivity, the stimulus response technique has been used. Examples include the pioneering ‘Oxford’ technique [3] in which a bolus injection of phenylephrine is given and the simultaneously rising blood pressure and interbeat are observed. Their ratio is BRS, which is taken from a regression slope of interval on pressure. Another ingenious and widely used technique, proposed by Di Rienzo et al. [4], uses spontaneously occurring pressure and interval variability to calculate the regression slope when pressure and interval change concordantly over a number of consecutive beats. In this case, a sufficiently clear covariant sequence is required. The outcome of these tests is quite variable. The spontaneous test, more than the phenylephrine test, produces results that show substantial variability between the various sequences occurring in an observation period [1]. This leads to the question: are the measurement techniques inadequate while BRS itself is a stable quantity, or is BRS not constant? It has been known for many years that BRS differs between sleep [3], rest and stress [5] but we cannot readily believe that BRS shows 10 : 1 variability on nearby readings in the same subject state. When subject state varies, such as with increasing work load on an ergometer, the variability includes a systematic component which should be recovered in the noise. The experiment repeated In the scientific tradition that experiments should produce repeatable results, we used the data available from Idema et al. [6] and calculated a form of sequential BRS on the recorded intrabrachial artery pressure wave. The results on six male subjects, aged 22–40 years, are shown in Figure 1. In this case, the bicycle exercise started at 20 W and was stepped up at a rate of 20 W/min, until exhaustion, near 300 W, in less than 20 min. The individual BRS estimates with their associated interbeat intervals were averaged over 100-ms wide intervals. Mean intervals were converted to mean heart rate. Clearly, a curvilinear relationship on semi-log paper is demonstrated in the form of a backwards-looking ‘S’ shape. Linearity is guaranteed only between heart rates of 80 and 120 b.p.m. Thus, the experiments previously performed by Abrahamsson et al. [1] are repeatable, although the linear relationship between log-BRS and heart rate appears to have a limited span.Fig. 1: Log-baroreflex sensitivity (BRS) versus heart rate, with sequential BRS in ms/mmHg, and heart rate in b.p.m. The vertical scale is logarithmic with the numbers indicating powers of ‘e'. Thus, 0 indicates a BRS of 1, and −1 to 4 indicate 0.37, 1, 2.7, 7.4, 20 and 55 ms/mmHg, approximately.Bristow et al. [5] performed an elaborate study exploring the decrease in BRS with increasing work load using the phenylephrine bolus injection method. They found that BRS decreased to near zero at heart rates approaching 120 b.p.m, beyond which it levelled off. Vagal control of heart rate ends at 100–120 b.p.m. Thus, the authors also observed a curvi-linear response of BRS in exercise for intervals less than 500 ms and heart rates over 120 b.p.m. The phenylephrine challenge As illustrated in the study by Abrahamsson et al. [1], we see that the almost linear relationship between log-BRS and heart rate is also valid for mental stress and cold stress, but not for controlled breathing when the BRS response is opposite. Let us look at another opposite response. In a study by Gizdulich et al. [7], the authors provide the systolic pressure and interbeat interval responses to graded phenylephrine infusion, together with differential BRS values calculated for the steady states. The subject group is different and consists of eight mild-to-moderate essential hypertensive subjects aged 18–60 years. The BRS is seen to rise with heart rate (1.2 ms/mmHg at 47.7 b.p.m, 5.2 ms/mmHg at 50.2 b.p.m and 12.6 ms/mmHg at 54.2 b.p.m) instead of falling. What is going on? In the exercise and the mental and cold stress tests, we observe a behaviour that goes against what we would expect of the baroreflex. In these stress tests, blood pressure rises, but the interbeat interval, instead of also rising to stabilize blood pressure, decreases as if to facilitate the blood pressure rise. Only in the phenylephrine experiment is the expected baroreflex behaviour observed (i.e. blood pressure up, then heart rate down). We call the phenomenon that probably occurs in the stress tests baromodulation. It was measured in rats under clonidine stimulus by Korner et al. [8], described by Sleight [9] and has been the subject of model studies [10]. Bristow et al. [5] state ‘With the onset of muscular exercise, however, both heart rate and arterial pressure rise, indicating that the relation changes. The change could be resetting of the reflex at a new level appropriate to the exercise, or a suppression of the reflex, or both'. It has been shown [10] that a ten-fold reduction in BRS is sufficient to raise blood pressure by 50%. Discussion To establish the negative regression of log BRS on heart rate, the range should be limited to where it would most likely be linear (80–120 b.p.m.). Clearly, over this short range, taking the individual values for BRS and heart rate into the regression would produce a more horizontal slope due to the scatter in the data. Thus, averages should be calculated within the linear range, and the regression slope from the averages. For BRS, this should be the averages of the logarithm of the values. Ordinarily in such regression analysis, one best specifies a typical value for BRS in the middle of the heart rate range because it is there that the uncertainty in the regression is least. That would lead to a BRS of 100 b.p.m. as the mid position between a range starting at 80 and ending at 120 b.p.m. A regression is incomplete without its slope, and because slopes differ by almost a factor of 2 in Figure 1, the slope should also be given. Note that 60 b.p.m. is not within this linear range, and estimating it from the regression would require extrapolation. This would produce too high values in view of the downwards curvature with decreasing heart rate. Why should we be interested in the suppressed value of BRS in exercise? If treatment or training is going to affect BRS, would it not be best to take the highest, probably fully uninhibited, value when the subject is asleep at night [3]? If that is too cumbersome, the recording of a triplet of values in quiet resting condition might suffice: (i) the spontaneous BRS value averaged over at least 15 min; with (ii) average heart rate; and (iii) average systolic pressure, assuming subject age is also recorded. Furthermore, when BRS is increased, pharmacologically or otherwise, it is indeed increased. It may not be clinically important that it is associated with changed interval or pressure. Alternatively, this may prove advantageous and a standardization inopportune. Only clinical research can guide us in correctly determining which specification of BRS is significant for a patient, be it the highest value at sleep, or the slope of the regression with heart rate in exercise, or a specific value of BRS, or some parameter not yet known. Presently, BRS itself is of clinical significance, but Abrahamsson et al. [1] have raised an important point in their desire for standardization (e.g. in drugs testing). We also agree on another point, namely the importance of the logarithm and heart rate in BRS research. However, even after more than 30 years of investigations, it may still be too early for really simple BRS characterizations in the light of the bewildering complexity of short-term blood pressure control [2].