The quest to prove or disprove the theory of ‘autoregulation' has stimulated much neonatal research using a variety of techniques and methods. The study presented by Boylan et al. in this issue of Pediatric Research is another such study. They employ sophisticated techniques (coherent averaging and linear regression) devised by the medical physicists in the team (1, 2). Essentially the methods analyze cerebral blood flow velocity (CBFV) changes, captured by Doppler ultrasound, in response to spontaneous and transient changes in systemic blood pressure. The crux of their findings is that, within the limitations of the numbers of infants studied, the pattern of CBFV responses seen in preterm infants and neurologically sick term infants are similar but are different to the responses of the neurologically normal term infants. The inference is that the groups have differing cerebral hemodynamic controls.

The second thrust of the paper is to place their findings into the concepts of autoregulation. They use the terms ‘static' to be synonymous with ‘classical' and equate their study to what they describe as ‘dynamic' control. In the adult literature the term ‘static' refers to CBF(V) responses to pharmacologically altered blood pressure and ‘dynamic' autoregulation refers to responses to maneuvers such as cuff inflation and deflation around a limb (3, 4), as described by Aaslid et al. (5). If one scrutinizes the literature on autoregulation, it becomes evident that the above uses of the terms static, dynamic, and classical are somewhat misleading.

It is clear that the previous concept of ‘autoregulation' is very limited, and we need to consider a much broader scheme of events. Just as standing upright is not dependent on the integrity of one system alone. A patient who fells over may have pathology, for example, of prorioception, a cerebral tumor or a cardiac arrhythmia. A failure of ‘autoregulation,' perceived as the development of a cerebral hemorrhage or an area of ischemia, may also be the result of a failure at one of several levels. The patient may not fall when standing still but does on walking or running. Similarly in cerebral hemodynamics there is evidence of differing processes in play, there are base-line, on-going fine tuning of CBF(V), which we could call static, versus controls of responses to external stimuli, which we could call ‘dynamic.' It is this latter scenario that was originally studied and perceived as autoregulation.

The definition of autoregulation produced by Lessen in 1959 (6) was ‘the ability to maintain cerebral blood flow in the face of a changing cerebral perfusion pressure.' This definition was based upon the classical experiments of Mogens Fog in the 1930s (7–9). He used a technique previously described by Forbes in 1928 (10). Fog created a window in the skull of cats to directly observe changes in the pial vessels' diameters in response to various stimuli. By what ever means the systemic blood pressure was decreased; stimulation of the carotid sinus and depressor nerves; peripheral stimulation of the vagus; or by arterial hemorrhage, he noted an immediate vasoconstriction of the pial vessels followed within 1.5 to 2 min by a secondary vasodilatation to restore the blood flow. On increasing the systemic blood pressure, by an injection of epinephrine or by splanchnic stimulation or by clamping the abdominal aorta, the reverse changes were seen in the pial vessels. He noted too that the compensatory mechanism only persisted over a range of blood pressures (110–140). It is this ability of cerebral vessels to constrict or dilate in response to a sustained change in perfusion pressure, to maintain cerebral blood flow, over the time-frame described, which should rightly be called ‘classical autoregulation.'

A number of animal studies followed using a variety of direct, but highly invasive techniques of measurement of blood pressure and blood flow to study the characteristics in different species of this type of response. They studied its time course, site of control, and possible controlling or confounding variables (11–17). The time frame, with some minor species variability, always demonstrated an immediate, passive response followed by a second response between 10 s to 1–2 min later to restore the blood flow to a steady state.

Studies in humans have many ethical constraints; however, Nomes et al. (18, 19) demonstrated in neurosurgical patients a similar time course for such responses in humans. Gotoh et al. (20) demonstrated in patients with Shy-Drager Syndrome, an autonomic neuropathy, that they were unable to maintain their CBF in response to a change in posture but they could to changes in carbon dioxide. This illustrates an important point that different controlling mechanisms exist for different stimuli.

The ethical considerations in neonates are even more taxing than those in adults. However, many valiant ethical attempts have been made to study this population. The paper by Lou et al. (21), referred to by Boylan et al. is often quoted as the evidence of a failure of autoregulation in preterm infants. However, in this study 19 preterm (29–33 weeks gestation) infants had only one measurement each of CBF, measured by the Xe-133 technique, along with a simultaneous recording of the blood pressure. The data points for each baby were compared with the rest in the group to create cross-sectional data of CBF and blood pressure. Sixteen of the infants had low pHs, 12 had high Paco2, and 7 were described as asphyxiated. This cross-sectional data does not represent autoregulatory changes as described by Fog.

Further neonatal studies have used Xe-133 (22, 23), jugular occlusive plethysmography (24), and Doppler ultrasound (25–28) All of these studies have their specific difficulties. Often cross-sectional data were gathered. Xe-133 cannot be used to make the rapidly repeating measurements of CBF which are required to capture the classical autoregulatory responses described by Fog, nor can one guarantee that outer confounding variables known to alter CBF will not have changed between the repeated measurements. Boylan et al. imply that the problem with some studies was the time it took to alter the blood pressure that resulted in the measurements being taken some time apart. The Xe-133 studies of Pryds et al. (22, 23) looked at spontaneous blood pressure changes and the difficulty was the time-lag of the Xe-133 technique. Further, the confounding variables known to change CBF(V) are typically, Paco2, and hematocrit. However, the main variable, which confused many earlier studies, was to take measurements of blood pressure and CBF(V) from an individual over consecutive days. This abolished the problem of cross-sectional data but both CBF(V) and blood pressure increase physiologically over the first few days of life, so an observation of a rising blood pressure and CBF(V) is not evidence of a failure of autoregulation.

Doppler ultrasound is the one technique that can capture the changes in hemodynamics over the necessary time-frame of autoregulatory changes within an individual. The difficulties of Doppler are that essentially it is a measure of velocity and not flow, and that a change in the angle of insonnation will alter the velocity reading without a true change within the baby. The latter problem has been overcome by the use of fixed probes at a constant angle (29), which then allows observation of changes in hemodynamics without necessarily implying measurement of flow. It is the information gathered from Doppler studies that has produced evidence that there are several different intrinsic patterns of changing CBFV that are quite different in time-frame from those witnessed in changes provoked by external stimuli.

The well-described CBFV changes seen in babies at rest with no external stimuli applied are 1) the beat-to beat variability seen by Perlman et al. and studied in relationship to synchronous breathing during artificial ventilation (30, 31), and 2) Slow wave cyclical changes in CBFV, occurring at 3–6 cycles per minute (32, 33). Kitney and co-workers have described similar cyclical oscillations in other physiological parameters, such as heart rate and temperature (34–38). These spontaneous changes in CBFV suggest that there are constant fine tuning processes, which, because of their very differing time-frames, probably represent differing underlying controls to each other and differing controls to the changes witnessed after differing external stimuli such as posture change (39) or the stimuli employed by Fog.

Furthermore, there is another group of external stimuli, which are known to alter CBF, namely carbon dioxide, acidosis, and metabolic demand, which indirectly alter perfusion pressure and may have other modes of action upon CBF and should be considered separately in any study.

The place of the CBFV changes recorded by Boylan et al. belongs within the intrinsic, spontaneous group of CBFV changes. They provide further important evidence of the differences between preterm and term infants. However, these changes are in no way similar to the responses provoked by studies of external stimuli with sustained changes in perfusion pressure, such as seen in changes of posture (39) or the studies of Aaslid et al. (5). These latter studies witnessed changes akin to those provoked by Fog, and are of the ‘classical' autoregulatory type. The different groups of changes are likely to have different underlying physiological and pathological mechanisms.

The study of Boylan et al. also provides some additional interesting evidence for the maturation of integrity for the underlying controls of CBF(V) (33,39–41), just as we see in the child learning to walk and not to fall over.