Positional circulatory control in the sleeping infant and toddler: role of the inner ear and arterial pulse pressure

Abstract

Heart rate (HR) and arterial blood pressure (BP) are rapidly and reflexively adjusted as body position and the force/direction of gravity alters. Anomalies in these mechanisms may predispose to circulatory failure during sleep. We analysed the development of two key reflexes involved by undertaking a longitudinal (birth–1 year) comparison of instantaneous HR and BP changes evoked by abrupt upright, sideways or horizontal repositioning. Each manoeuvre triggered an identical rise in HR (tachycardia) followed by a slower rise in diastolic blood pressure (DBP)/systolic blood pressure (SBP) and variable pulse pressure (PP) change. We show that tachycardia is triggered by acceleration (vestibular) sensors located in the inner ear and slight changes in the pulsatile component of BP then signal to the arterial baroreceptors to reinforce or oppose these actions as needed. We also identified a PP anomaly in sleeping 1-year-olds of smokers that prematurely slows HR and is associated with mild positional hypotension. We conclude that positional circulatory compensation is initiated pre-emptively in a feed-forward manner and that feedback changes in vago-sympathetic drive to the heart (and perhaps blood vessels) by PP exert a slower but powerful modulating effect. An anomaly in either or both mechanisms may weaken positional compensation in some sleeping infants.

Key points

• Fast-acting reflexes fine-tune heart rate and blood pressure as the push and pull of gravity changes from moment to moment.

• Circulatory failure may occur if these mechanisms fail to develop normally.

• Reactions to routine movements such as head-up and sideways tilt show how two key reflexes involved develop during sleep, in early infancy.

• Sensors that detect motion and arterial pulsations, located in the inner ear and arteries, respectively, produce carefully coordinated changes in heart rate and blood pressure as body position and the force/direction of gravity alters.

• Both mechanisms are intact in infants exposed to tobacco, but a mild pulse pressure anomaly reflexively slows their heart and lowers their blood pressure on tilt to upright.

• Circulatory dysfunction need not necessarily reflect abnormal development of key reflexes per se, but other factors that inappropriately trigger or inhibit them.

Introduction

Acceleration is a fundamental organizing force of life on earth, not least because it alters blood pressure and flow. The static gravitational acceleration on standing, for instance, regularly forces blood away from the brain and heart towards the legs and feet. The fact we successfully spend much of our lives upright and in motion is due to strategies we have developed to counteract such effects. The baroreflex is widely regarded as the principal mechanism involved in this counteracting action, via stretch receptors that instantly sense changes in arterial calibre and tension as blood passively shifts along the axes of acceleration (Smyth et al. 1969). But there are other mechanisms triggered specifically by motion and gravity about which much less is known. The vestibular system may initiate pre-emptive changes in heart rate (HR) and blood pressure (BP) before the baroreceptors are engaged, via sensors in the inner ear exclusively devoted to sensing the body's motion and orientation in space (Radtke et al. 2000; Yates & Bronstein, 2005; Carter & Ray, 2008; Viskari-Lahdeoja et al. 2008). It comprises a gyroscope to detect angular or turning motion (three paired semicircular canals) and an accelerometer to measure translational (inertial) accelerations and changes in the position of the head relative to gravity (the two otoliths). These five suborgans act in concert to monitor instantaneous changes in velocity in three linear (up–down, left–right, fore–aft) and three rotational directions (pitch – nod your head ‘yes’; yaw – shake your head ‘no’; and roll – move your head back and forth from the left to right shoulder) (Furman & Baloh, 1992; Angelaki & Cullen, 2008). Neural pathways connect the vestibular and cardiovascular systems in the brainstem, so not surprisingly vestibular activation can elicit rapid increases in HR and BP under experimental conditions (Yates, 1996; Jauregui-Renaud et al. 2006). But whether vestibular reflexes are important in routine human positional compensation is difficult to discern (Yates, 1996; Kaufmann et al. 2002; Carter & Ray, 2008). One aspect of our study addresses this issue.

The second issue addressed is developmental: when and how do we acquire circulatory countermeasures to gravity and acceleration? For 9 months the fetus is immersed in a fluid and compressed by a capsule that exerts a counter pressure not unlike that delivered by the anti-gravity suits worn by fighter pilots; this diminishes hydrostatic gradients, turbulence, shear stress, and the direction and force of gravity. The heart and blood vessels are highly responsive to loads so how well prepared are we for these at birth? Are compensating mechanisms already functional and effective and how are they shaped subsequently? These questions prompted us to undertake a longitudinal (birth–1 year) comparison of the circulatory effects of passive repositioning, simulating natural body movements. We focused on the roles of the vestibular and baro-reflexes in human circulatory control and how these evolve after birth. Both reflexes help resist (‘buffer’) sudden rises/falls in BP and both play a key role in normal – and aberrant – positional compensation (Farquhar et al. 2000; Harper et al. 2000; Yates & Bronstein, 2005). We addressed three questions in particular: (1) can discrete vestibular effects be discerned in infancy? (2) How effective is positional compensation at birth and how does it change subsequently? (3) Could an anomaly in either or both reflexes weaken compensation in certain orientations whilst asleep? We studied infants and toddlers because their small size and weight makes it relatively easy to perform controlled reorienting manoeuvres, and because their relatively rapid anatomical and physiological maturation may provide unique insights into functional cardiovascular reflex development. As the infant grows and begins to stand and walk, he/she must not only learn to navigate and map the surrounding environment (Derdikman & Moser, 2010), but also adapt and strengthen reflex circulatory countermeasures to gravity. Inadequate or inappropriate reflex engagement could accentuate undesirable positional effects, such as uncompensated hypotension during sleep (Harper et al. 2000; Harrington et al. 2001; Cohen et al. 2010). Follow-up studies such as this are the only reliable way to determine whether reflex or other anomalies emerge early on and track into childhood, evaluate their likely impact and – ultimately – develop strategies to reduce adverse consequences and improve outcomes.

Methods

Subjects

Our cohort comprised healthy infants born at term to non-smokers (controls) or mothers who smoked moderately throughout pregnancy. All were appropriately grown (Table 1). Controls were tested as soon as practicable after birth and again at 3 months and 1 year, but because of the difficulty of recruiting infants of smokers in Sweden we recalled and studied a previous cohort using this new protocol at 1 year only (Cohen et al. 2010). All 1-year-olds (‘Toddlers’) had achieved the general milestone for their age, i.e. were able to pull themselves up to a standing position, walk unsteadily with a wide based gait holding onto people or furniture, and fall easily. During a routine daytime nap the subject slept on his or her back on a purpose-built bed which consisted of two pivoted, independently rotating frames that permitted sideways (the inner frame) or head-over-heels (the outer frame) rotation to 60 deg left/right or upright, respectively. The infant was positioned within the inner frame, comfortably immobilized with padding and straps to prevent limb/head movements. The head was tilted forward 30 deg so the horizontal semicircular canals and utricle (which lie in the same plane) were perpendicular to the earth horizontal plane and parallel to gravity (Furman & Baloh, 1992). Beat-to-beat heart rate and blood pressure were recorded from a wrist or finger cuff kept at heart level (Finometer, FMS, Amsterdam, The Netherlands) (Cohen et al. 2007, 2008, 2010). We also recorded a three-lead electrocardiogram and breathing movements via inductance plethysmography (infants <3 months only). Testing was confined to behavioural quiet sleep (eyes closed, absences of eye and gross body movements except for occasional jerks/starts/sighs, occasional rhythmic mouthing, regular HR, BP and breathing if discernible) (Anders & Keener, 1985). Once HR and BP was steady for 30 s, one of four manoeuvres was performed: head-over-heels rotation from supine to 60 deg semi-upright (pitch); rotation from supine to 60 deg left or right lateral decubitus (roll); or supine head/feet-first acceleration 2.5 m (translation). Pitch and roll activates mainly the vertical and horizontal semicircular canals, respectively; both manoeuvres stimulate the otoliths since rotations about the body's lateral and rostro-caudal axes reorient the utricle and saccule relative to gravity. Translation is exclusively an otolith stimulus that activates mainly the saccule when supine (the organ most sensitive to acceleration in the earth-horizontal plane) (Furman & Baloh, 1992). Movements were executed in 2–3 s, and the infant returned to the starting position after 30 s; we allowed 2–3 min for recovery between manoeuvres. If arousal or a change in state occurred we waited for sleep to resume before continuing. Starting position/order of manoeuvres was counterbalanced across infants/sleep epochs in order to obtain two to three responses in each direction. The procedure took 2–3 h, complied with the Declaration of Helsinki, and was approved by the Ethics Review committees of Royal Prince Alfred Hospital and Karolinska University Hospital. Parents provided written, informed consent.

Table 1.

The population

Age group	Control ≤1 week	Control 2–3 weeks	Control 3 months	Control 1 year	Tobacco 1 year
n	10	7	17	13	8
GA (weeks)	40 ± 1	40 ± 1	39 ± 1	39 ± 1	40 ± 1
PNA (days)	4 ± 3	19 ± 6	93 ± 18	462 ± 53	379 ± 27
Birth wt (g)	3300 ± 187	3475 ± 477	3433 ± 380	3225 ± 505	3115 ± 332
Study wt (g)	3238 ± 180	3728 ± 539	5881 ± 805	10118 ± 3476	10142 ± 1142
MBP (torr)	80 ± 13	91 ± 18	74 ± 12	67 ± 4	66 ± 5
SBP (torr)	93 ± 15	107 ± 20	91 ± 14	87 ± 7	85 ± 6
DBP (torr)	69 ± 12	77 ± 16	58 ± 11	48 ± 4	47 ± 5
PP (torr)	24 ± 4	30 ± 4	33 ± 8	37 ± 7	38 ± 4
HR (bpm)	113 ± 12	138±12	121 ± 8	107 ± 11	100 ± 14 *
Pitch	71 (11)	18 (1)	49 (4)	33 (1)	21 (1)
Roll (L/R)	24/21 (3)	11/10 (2)	26/24 (1)	18/16 (0)	15/16 (0)
Translation	7 (1)	14 (1)	29 (0)	21 (0)	19 (0)

Infants of non-smokers (controls) and mothers who smoked >10 cigarettes/day were studied during a nap (the latter only at 1 year). GA, gestation at birth; PNA, postnatal age. Mean (MBP) systolic (SBP) diastolic (DBP) and pulse pressures (PP) and heart rate (HR, beats min⁻¹) pre-test; the total number of movements analysed together with the number excluded due to arousal (brackets) is shown in the final 3 rows. Means ± SD; *P = 0.01 vs. controls at the same age (ANOVA).

Analysis

Tests interrupted by a sigh or head/limb movement were discarded. We combined fore and aft translations, and roll to either side from (and back to) supine since cardiac and BP responses did not differ significantly between these manoeuvres. For pitch we present data for rotation to semi-upright (but not the return to supine). The 10 beats before and 40 beats from the onset of movement were used to generate beat-to-beat responses to pitch, roll and translation. Data from replicate tests were firstly averaged for each subject, then pooled to generate mean beat-to-beat responses at each age. We calculated systolic (SBP) diastolic (DBP) and pulse (PP) pressure, and the time interval between successive systolic peaks (inter-beat interval, IBI, assigned to the 2nd peak); DBP was the minimum preceding each SBP peak, and PP the amplitude of the systolic upstroke (Fig. 1).

Figure 1. Effects of motion during sleep.

Blood pressure (BP) and breathing during left rotation (arrow in A). Systolic (SBP), diastolic (DBP) and inter-heart beat interval (IBI) were as shown (B). Translation, pitch and roll caused tachycardia (C) and increased PP (D), DBP and SBP (E). Tachycardia always coincided with acceleration–deceleration (C). When horizontal it raised PP and SBP (D and F) and was followed by bradycardia (labelled 1 in C). Pitch triggered only tachycardia and a narrowing of the SBP-DBP gap/fall in PP (labelled 2 in D; G). Data are means ± SD for 1-year-olds of non-smokers (qualitatively representative of all ages).

To analyse the correlation between BP components and IBI following translation and roll we used the 8 s (10 beats) immediately after the nadir in IBI, assuming the increase in IBI was linked cause-and-effect to the rise in BP via the baroreflex. We used a fixed delay of one beat between PP–IBI and three beats between SBP–IBI; these delays were determined empirically by plotting SBP/PP against the coincident and succeeding IBIs and gave the best linear correlations (r² > 0.8) between IBI and SBP/PP during pressor sequences. By convention, the line relating IBI to BP measures cardiac baroreflex sensitivity (BRS; ms torr⁻¹); regression coefficients (slope, correlation and P value) were estimated using least squares regression. This is essentially a non-pharmacological version of the Oxford technique (Smyth et al. 1969). The same approach was used to analyse SBP/PP–IBI changes following pitch.

Cardiovascular autonomic modulation at rest was evaluated using auto-regressive cross-correlation algorithm to compute low frequency (LF; 0.04–0.15 Hz) power density of HR and BP variability (Pagani et al. 1986; McKinley et al. 2003). We used 5 min BP recordings for all 1-year-olds made during undisturbed QS once passive movements were complete; the Finometer self-calibration function was disabled at his time. The algorithm supplied the number, centre frequency, and power of each oscillatory time series. Cross-spectral analysis gave the frequency-related coherence, phase shift, and transfer function between each component (SBP/DBP/PP) and IBI. We took the points with the highest squared coherence >0.5 within the LF band to maximize meaningful correlation between signals. The LF transfer function estimated cardiac BRS, phase relationships, and delays between the various time series. The BRS measures spontaneous baroreflex-mediated modulation of cardiac rhythm. Phase estimates the angular difference (time delay) between two waveforms: a negative phase indicates delay (i.e. that a change in BP precedes a change in IBI) and a positive phase the converse. A negative phase is conventionally taken to indicate baroreflex-mediated feedback modulation of HR.

Statistics

We used repeated measure two-factor analysis of variance (ANOVA) or Kruskal–Wallis non-parametric tests as appropriate; when the omnibus ANOVA was significant, modified t tests (Bonferroni correction) were applied. Error is presented as standard deviation; P≤ 0.05 was considered significant (95% confidence level).

Results

The study cohort and number of manoeuvres are summarized (Table 1). Baseline HR/BP asleep increased during the first month then fell; because this may influence reactivity to motion, the cohort was stratified into four groups: ≤1 week, 2–3 weeks, 3 months and 1 year (Cohen et al. 2010). Longitudinal data from the control cohort are presented in Figs 1–3; for clarity general effects seen at all ages are illustrated for one age only (Fig. 1 and 3) and differences between ages highlighted (Fig. 2). The 1-year-olds of smokers and non-smokers are compared in Figs 4 and 5.

Figure 3. Circulatory transients are reversed by a widening pulse wave.

During translation and roll diastolic pressure (DBP) rises first as heart beat interval (IBI) shortens; pulse pressure (PP) increases last and systolic pressure (SBP) changes are intermediate. The response to translation (A) shows the reversal in IBI due to deceleration (labelled 1 in A), presumably reflecting baroreflex engagement as BP rises (shaded traces in A overlaid in B); Slope ( = cardiac baroreflex sensitivity) doubled if PP was the stimulus, for infants of non-smokers (C) and smokers (D). Data in A and B are means ± SD from 1-year-olds of non-smokers (representative of all ages; transients were similar for translation and roll).

Figure 2. Compensation changes with age.

Heart rate (HR), pulse pressure (PP) and systolic pressure (SBP) during translation (A–C) roll (D–F) and pitch (G–I) at 4 ages. Translation and roll always caused tachycardia (labelled 1 in A and D) then rebound bradycardia (labelled 2 in A and D) but pitch elicited only tachycardia (labelled 1 in G). Translation boosted PP and SBP at all ages (B and C) and roll after 3 months (labelled 3 in E and F) but the same responses to pitch were subdued at all ages (H and I). A rise in BP was associated with bradycardia if SBP and PP increased (labelled 2 in A), but not if SBP alone increased (roll before 3 months; labelled 3 in D, E and F). Peaks (or nadirs during pitch) in HR, PP and SBP are compared in the lower panels (J–L; 3-beat averages at the approximate times indicated by the arrows in A–I); significant differences between the same movement at all other ages (*) and all other movements at the same age (¶) are indicated (ANOVA). Data are means ± SD for infants followed longitudinally from birth.

Figure 4. Orthostatic compensation is reinforced by a narrowing pulse wave.

Changes in systolic (SBP), diastolic (DBP) and pulse pressure (PP) and inter-heartbeat interval (IBI) during pitch for infants of non-smokers (A) and smokers (B). The heart usually continued to speed-up after reaching a semi-upright position (labelled 1 in A) but slowed dramatically in tobacco-exposed infants (labelled 2 in B); re-plotting the highlighted sections showed that IBI always tracked PP (C) but not SBP (D). Data are means ± SD for 1-year-olds.

Figure 5. Pulse pressure affects heart rhythm at rest.

Spectral analysis of pulse pressure (PP), systolic pressure (SBP) and inter-heartbeat interval (IBI) of all infants at 1 year during undisturbed sleep revealed the fastest coupling was between PP and IBI. Baroreflex sensitivity (BRS, the modulus of the low frequency transfer function; A) effectively doubled if PP was the stimulus, indicating spontaneous PP oscillations have the strongest/fastest effect on the heart at rest. Data are means ± SD; significance levels (Kruskal–Wallis test) for control vs. tobacco-exposed infants are shown; *all values significantly different from the respective PP–IBI values for the same group (P < 0.001).

A vestibular circulatory reflex can be discerned

We looked for similarities between beat-to-beat transients triggered by different manoeuvres for evidence of a vestibular role in routine adjustments, e.g. to upright repositioning (‘orthostasis’). Movement was brisk and gentle, typically lasting 2–3 s (Fig. 1A and B). Translation and roll to either side evoked an initial rise in HR lasting 5–6 beats (tachycardia) then a fall below baseline (rebound bradycardia), but pitch elicited only tachycardia. The initial rise in HR was identical for all movements in all directions and coincided precisely with motion per se (Fig. 1C). Translation and roll consistently raised all BP components but pitch raised only SBP and DBP whilst PP fell (Fig. 1D–G). Thus, the tachycardia that accompanies a sudden change in body position is primarily an otolith-mediated (inner ear) response to motion; it is not intrinsically a baroreflex response to a secondary fall in SBP, which may actually rise in parallel with DBP.

Positional compensation is partially mature at birth

Longitudinal comparisons of healthy infants reveal how vestibular and baroreflex-mediated effects change with age. All movements evoked brisk tachycardia at all ages. The infant possessed a strong BP (pressor) response to linear acceleration (a weak circulatory load) at birth (Fig. 2A–C), and to roll (a moderate transverse load) from 3 months (Fig. 2D–F). The pressor response to pitch – which imposes the greatest load on the heart and blood column – was slowest to develop (Fig. 2G–I). Thus, the newborn possess a well-developed vestibulo-cardiac reflex but full positional BP control develops incrementally after birth (Fig. 2J–L).

Positional compensation is influenced by pulse pressure

To clarify how equilibrium is normally restored after a disturbance we analysed transients induced by translation and roll; both manoeuvres keep the body horizontal so blood remains relatively evenly distributed and circulatory perturbations brief. Each typically caused all arterial BP components to rise with slightly different latencies: the fastest rise was always in DBP and the slowest was in pulse pressure (PP), with SBP intermediate in amplitude and phase (Fig. 3A). The pressor ramps were associated with a rebound fall in HR consistent with baroreflex feedback. Comparisons of cardiac baroreflex sensitivity (BRS) for each arterial pressure component (Fig. 3B-D) revealed that PP had the strongest (and fastest) effect on HR; infants of different ages and those of non-smokers and smokers did not differ in either respect. A sudden rise in PP consequently triggers powerful baroreflex slowing of HR to lower BP and restore equilibrium, i.e. circulatory changes are promptly reversed as the pulse wave widens.

In view of this finding we wondered why the rise in HR (and DBP/SBP) is sustained when the body is repositioned semi-upright, and not promptly reversed by baroreflex engagement as occurs after horizontal motion (Fig. 4A); for clues we compared 1-year-olds of non-smokers whom we have shown previously exhibit mild orthostatic hypotension (Cohen et al. 2010). The HR of these infants does in fact slow abnormally after pitch (Fig. 4B). We noticed that changes in IBI always tracked changes in PP (Fig. 4C), but the correlation with SBP was poor and inconsistent (Fig. 4D). Baroreflex sensitivity estimated from the short pressor ramps induced by translation and roll was comparable for both groups, regardless of which component of BP was used (Fig. 3C and D). Thus, a fall in PP normally prolongs tachycardia and helps maintain DBP (hence SBP) elevated during orthostasis, but a rising PP prematurely slows the heart leading to an early/accelerated decline in BP. In brief, circulatory changes are promptly reinforced as the pulse wave narrows.

Finally, we used spectral analysis of spontaneous HR/BP oscillations around the set point to verify whether PP effects described above also occur during undisturbed sleep. The PP–IBI time delay at rest was comparable to that measured by the sequence method (1 beat) but the SBP–IBI delay was nearly doubled (6–7 beats vs. 3 beats). As shown elsewhere (Yiallourou et al. 2010), absolute BRS values from spectral analysis were consistently (about twofold) greater compared with sequence analysis, but the same trend was evident: using PP instead of SBP effectively doubled BRS (Fig. 5). Infants of smokers and non-smokers did not differ significantly in either respect. These findings were consistent with data from evoked pressor sequences. In brief, arterial pulsations also exert, via the baroreflex, the quickest and strongest influence on the heart's rhythm at rest.

Discussion

What tells the heart and blood vessels to work harder or relax as the pull and push of gravity changes, often from moment-to-moment? Our study highlights two key neglected mechanisms. The first is an exceptionally fast, vestibular feed-forward reflex that pre-emptively elevates HR (hence BP). The second is a slower feedback (baro) reflex that reinforces or reverses these actions according to need. How is this need assessed? Surprisingly, it is via small changes in the pulsatile (not the systolic) component of the arterial pressure wave, to which the baroreflex is exceptionally sensitive. Thus the pulse wave – which is responsible for maintaining blood flow between heartbeats – may be much more influential in routinely regulating the heart's natural rhythm than we suspect!

A powerful vestibulo-cardiac reflex is a useful strategy to enable supply to be adjusted before gravity's pull on the blood actually changes. Its greatest value is likely to be when – as our studies performed during sleep perhaps illustrate – other (e.g. visual, proprioceptive, and behavioural) cues are weakest, i.e. when vigilance is low and acceleration rapid and unexpected (Yates et al. 1999). Much like mechanical ‘pump priming’, speeding-up the heart suddenly ejects blood into the arteries at a faster rate than it flows out, which immediately boosts DBP and SBP even when upright. Even the modest momentary linear acceleration we achieved (maximum <0.2 g; compare a limb movement, cough or sneeze, which produces ∼3 g) (Allen et al. 1994), still raised HR by ∼8% and BP by ∼7 torr, comparable to the effect of linear acceleration on the intact (but not vestibular deficient) adult (Yates et al. 1999; Radtke et al. 2000; Yates & Bronstein, 2005). The baroreflex is rather inefficient by comparison: it is engaged after BP begins to fall and may take 5–10 s to be fully effective (Korner, 1971). Could such fast, feed-forward adjustments sometimes over-compensate? We found if this happens it is short-lived and promptly reversed by the baroreflex (which can also prolong feed-forward effects if circumstances demand). Positional compensation is consequently best achieved by interplay between both mechanisms. Because feedback mechanisms are by nature reactive, feed-forward mechanisms may take the lead in shaping HR and BP at rest (Silvani et al. 2008; Kamiya et al. 2011; Tan & Taylor, 2011). The vestibular system may be particularly well suited for such a role because its nuclei routinely receive input from so many sensors – motion, somatosensory, proprioceptive and graviceptive (Yates et al. 2000). Since chronic exposure to tobacco reduces the frequency of spontaneous arousals during sleep (Richardson et al. 2009), it may reduce the effectiveness of vestibular triggering of feed-forward positional compensation.

Because the fetus experiences weak linear, centrifugal and angular accelerations that stimulate and refine otolith and semicircular canal function and connectivity, vestibular effects on heart rate may develop early in gestation (Fritzsch et al. 2001; Ronca et al. 2008). Functional development of the heart and blood vessels, on the other hand, is influenced by mechanical stress, which is low before birth due to counter-pressure exerted by the uterus but rises after birth (Avery, 1965). Our findings are consistent with these observations: the vestibulo-cardiac reflex is indeed strong at birth whilst BP compensation – particularly to orthostatic loads – is achieved slowly, incrementally afterwards. A newborn infant consequently resembles an adult whose fitness and cardiovascular reserve are reduced by inactivity or microgravity: both must adapt to full activity and gravity at various levels (heart, blood vessels and blood volume) to optimize positional BP (Levine et al. 1997; Spaak et al. 2001).

Baroreflex slowing or speeding of HR is typically elicited by short pressor/depressor ramps produced with drugs or pressure manoeuvres (Smyth et al. 1969; Blaber et al. 1995). By convention, its sensitivity is measured by plotting HR against SBP because the established techniques produce large (15–40 torr) increases in SBP and DBP but little or no change in PP (Smyth et al. 1969). But could PP be influential in other settings? After all, the baroreceptors respond to intermittent deformation of the blood vessel wall and their discharge rate varies rhythmically with the pulse wave (Angell James, 1971; Hajduczok et al. 1988; Chapleau & Abboud, 1989). Our findings made in infancy indicate this is indeed so. Firstly, during small naturally evoked pressor sequences vagal effects (on HR and DBP) were faster than sympathetic effects (on PP); the resulting phase differences revealed the heart slows sooner after a rise in PP then SBP, a discrepancy confirmed by spectral analysis of undisturbed sleep (Fig. 5). A long and variable SBP–HR delay is peculiar a feature of infancy (Yiallourou et al. 2010) that cannot be reconciled either with the known properties of the baroreflex pathways or data from adults (Pickering & Davies, 1973). Secondly, small pressor sequences do not cause true bradycardia unless PP rises; HR simply decays back to baseline once motion ceases if only SBP rises – something that is not consistent with active baroreflex engagement (Fig. 2D–F). Finally, in contrast to the adult (Borst et al. 1984) prolongation of orthostatic tachycardia in infancy was associated with a decline in PP but not in SBP (Fig. 4A and B).

There can be no question that a large rise in SBP, representing the combined effect of DBP and PP, engages the baroreflex. The issue our work raises is whether routine patterns of baroreflex engagement, especially the role played by the various components of arterial pressure, are better delineated by the small pressor sequences produced by natural challenges (Bertinieri et al. 1988) or even from a detailed spectral analysis of the stationary periods that follow, since both approaches agree in relative terms (spectral estimates of BRS are consistently greater). Our conclusion – that the baroreflex is strongly coupled to PP during sleep – makes sense because a fall in PP correlates directly with a fall in stroke volume, often the fastest and most dramatic consequence of gravity's pull. The pulse wave always conveys this information but SBP may not since it depends on DBP. However, limiting the rise in PP must obviously be as important (for different reasons) as allowing it to fall too low. Perhaps this is why the heart is so exceptionally sensitive (BRS is steepest) when the stimulus is dynamic/pulsatile. Static stretch of the artery by a rise in DBP, on the other hand, may stiffen the vessel wall and dampen pressure transmission to the baroreceptors, effectively ‘attenuating’ BRS. If DBP rises as PP falls – as normally occurs during orthostasis – both events presumably unload the baroreceptors to reinforce tachycardia and elevate BP (Fig. 4A). If, however, DBP and PP rise in tandem these actions are likely to be unequal and opposite and favour re-loading/re-excitation of the baroreceptors by a widening PP. This (rather than altered baroreflex gain) may be why the heart of the sleeping, tobacco-exposed infant slows prematurely and a relative positional hypotension sometimes develops (Fig. 4B) (Cohen et al. 2010).

It seems unlikely that if the cardiac arm of the baroreflex responds to PP it does so only in infancy (Charkoudian et al. 2005). In fact it may simply be easier to observe at this age because resting HR is relatively high. Some manoeuvres, though, passively boost blood flow to the heart (e.g. reverse pitch) causing stroke volume and all BP components to rise more-or-less simultaneously. If this happens, or if indeed PP fails to rise at all, it may be erroneously concluded that PP does not influence the heart. It remains to be determined whether the pulse wave also influences vascular tone via the sympathetic arm of the baroreflex; this is more important for regulating BP but also far more difficult to study routinely (Parati & Bilo, 2011).

Could tachycardia be triggered by changes in breathing, since HR typically accelerates during inspiration and slows during expiration? These effects can only be estimated but we did exclude sighs, which trigger this mechanism (Dykes et al. 1986). We also found (like others) that the respiratory component of HR variability was weak at birth, when the reflex was particularly strong. Nevertheless, vestibular cardiovascular actions may well have a respiratory component (Jauregui-Renaud et al. 2006). Could translation cause blood to shift towards the head or feet triggering non-vestibular circulatory reflexes (Yates et al. 1999)? If so, fore and aft acceleration should have had reciprocal effects on HR and BP, which was not the case. Finally, PP measured in the wrist or finger may over-estimate the pressure the baroreceptors ‘see’ because the wave is amplified as it travels away from the aorta; this seems to be less problematic in infancy (Andriessen et al. 2008).

Conclusion

An anomaly in pulse wave generation or transmission can arise for many reasons (Dornhorst et al. 1952; Alfie et al. 1999; Safar et al. 2003; Michard, 2005). Apart from well known side effects (stress-induced cardiac and vascular remodelling) we add another: it can also send powerful signals via baroreflex telling the heart (and perhaps the blood vessels) to prematurely slow down and relax. Effective circulatory compensation also depends on feed-forward mechanisms triggered by motion. Our data suggest that it may not necessarily be an anomaly in the sensitivity of either or both these mechanisms per se that progressively weakens positional compensation during sleep in early infancy but perhaps the way they are (or are not) routinely triggered that may be the real problem.

Glossary

BP

blood pressure

BRS

baroreflex sensitivity

DBP

diastolic blood pressure

HR

heart rate

IBI

inter-beat interval

SBP

systolic blood pressure

PP

pulse pressure

Author contributions

G.C. designed the study; G.C., S.V., H.J., H.L. and M.K-S. performed research; H.J., H.L. and M.K-S. contributed analytical tools; G.C and M.K-S analysed the data; G.C. wrote the paper. All authors reviewed and approved the final published version of the manuscript. This work was undertaken jointly at the Neonatal units at Royal Prince Alfred Hospital (Sydney) and the Karolinska University Hospital (Stockholm). The authors declare no conflicts of interest.