Cerebral Perfusion: Err on the Side of Caution

William L. Lanier, MD


As controversy continues regarding the hemodynamic management of patients in the head-up or beach chair position, the APSF Newsletter turns to Dr. William Lanier for editorial perspective. Dr. Lanier is Editor-in-Chief of Mayo Clinic Proceedings as well as a highly regarded neuroanesthesiologist and neurophysiology investigator.

In the Summer 2007 issue of the APSF Newsletter, Cullen and Kirby reported on 2 patients in whom a catastrophic, new-onset brain injury was discovered after surgery in the beach chair (barbershop) position.1 The authors presented views on the effect that blood pressure monitoring and management may have had on neurologic injury and provided a formula for correcting hydrostatic blood pressure gradients from the site of measurement to the site of vulnerable brain tissues. This publication generated a series of letters to the Newsletter, either supporting or challenging the need for the blood pressure corrections suggested by Cullen and Kirby. Notable among those letters was that of Munis who argued that a correction for hydrostatic gradients was not needed because, in the head-up position, the circulation above the heart functions as a siphon.2 Cucchiara took another approach and chided practitioners to place an arterial catheter in head-up patients and measure blood pressure at the level of the head to avoid the need for arithmetically corrected measurements altogether.3 This debate continues in the current issue of the Newsletter with letters from Drummond et al. who argue that clinical management of head-up patients must account for hydrostatic gradients,4 and Kirby and Cullen5 who expand on concepts raised in their earlier publication.1

Figure 1

Figure 1

This debate about blood pressure monitoring and management in head-up patients is unavoidable because of inadequate empirical data involving anesthetized, head-up patients who are at risk for rare, but debilitating, postoperative neurologic deficits.1,6 Various forms of head-up positioning are used not only for neurosurgical procedures (e.g., posterior fossa craniectomy and cervical laminectomies) where the effects on hemodynamics have been more intensely pondered, but also for surgery to the thyroid gland, shoulder, and other non-neurosurgical sites where debate about blood pressure management has been less common. Placing the patient supine or prone to avoid physiologic challenges imposed by a head-up position is not always an option, as the sitting position for posterior fossa craniotomy is reported to diminish operative blood loss and significantly improve postoperative cranial nerve function.7 With cervical spine surgery or posterior fossa intracranial surgery, converting from the sitting to prone position may potentially worsen pulmonary gas exchange in patients having medically complicated obesity, or may contribute to the risk of postoperative visual impairment in rare instances. Other surgeries (e.g., thyroid and shoulder surgery) are simply made more technically difficult by varying from an ideal head-up position. As such, it appears that the head-up position during anesthesia and surgery is here to stay, even though ideal blood pressure monitoring and management in these patients is controversial.

One of the core features of the current debate about blood pressure management in the head-up position revolves around whether the circulation above the heart functions as a siphon system2 or as a waterfall system.1,4,5 Based on the available evidence, either scenario is probably an oversimplification in anesthetized, surgically positioned patients. The siphon concept is very appealing when speaking of the physiology of unanesthetized healthy humans or giraffes; however, anesthetized surgical patients placed head up—often with the head position deviating considerably from neutral—may introduce more complex physiology. As we will see later, these head-position variations, independent of a gravity effect, have a bearing on cerebral circulation. Further, the siphon analogy assumes that vessels will function in series, when in fact the vessels connecting the heart to the most remote areas of the brain tissues and spinal cord have some elements in series and some in parallel. These parallel aspects of the circulation may place tissues within remote watershed regions at risk for ischemic injury coincident with global cerebral and spinal cord blood flow remaining adequate. It is not so simple to model the cerebral circulation as a waterfall either, because a waterfall analogy dictates that the hydrostatic gradient of the column of blood in vessels meaningfully influences the relationship between the pressure at the aortic root and the remote regions of the brain. This analysis, too, overlooks the input of vessels in parallel, some of which may be occluded at baseline (e.g., from atherosclerosis) or as a result of surgical positioning. Some examples are in order:

Toole and Tucker8,9 reviewed the literature concerning awake patients who acquire new-onset neurologic symptoms related to changes in head position, and they identified multiple contributing factors such as: 1) intraluminal atherosclerosis, 2) deviations from classic vessel configurations within the neck (most commonly involving a diminutive or non-functioning vertebral artery unilaterally), 3) changing relationships between the geography of the brainstem and vertebral vessels during head flexion, and 4) external compression of the carotid and vertebral arteries by osteophytes or normal vertebral anatomy. In a prospective study,9 they examined the effect of head flexion/extension, rotation, and tilt on blood flow through the carotid and vertebral arteries in 20 fresh cadavers. They determined that, if a change in flow was to occur at all, it occurred at flexion/extension of <45°, rotation of <45°, or tilt of <30°. A positive response was manifested as simultaneous cessation of blood flow in both vertebral arteries in 30% of cadavers, and in both internal carotid arteries (but not simultaneously) in 45% of cadavers. This research also determined that the diminution or ablation of blood flow in these vessels was not linear with head movement, but instead developed precipitously over an incremental 5-10° change. Additionally, they determined that it was not possible to predict in which vessel, or even on which side of the body, vessel occlusion would occur during head rotation. Elsewhere Perkins et al.10 reported on 2 patients who underwent right carotid endarterectomy while the patients were supine with the head rotated to the left. Inadvertent lidocaine injection into the right carotid arteries (during attempted local anesthesia of the carotid sinus baroreceptors) produced electroencephalographic (EEG) changes in both patients, but the EEG patterns varied greatly for reasons made clear by the preoperative angiogram. In 1 patient, atherosclerotic changes limited the contributions of the right carotid artery to the right side of the brain. Not surprisingly, EEG changes in this patient were unilateral and ipsilateral to the site of lidocaine injection. In contrast, the other patient had simultaneous EEG changes in both cerebral hemispheres, though more prominent in the right. Angiography revealed that, because of widespread atherosclerosis, the left carotid artery contributed nothing to the circulation of either cerebral hemisphere; however, the right carotid artery supplied blood for both hemispheres. Clearly these collective observations of Toole and Tucker8,9 and Perkins et al.10 speak to the fact that the plumbing of the human brain can be variable, dependent on changes in head positioning, and conceptually quite different from household plumbing.

Parallel Plumbing Important

If this is the case, one should examine the extremes of blood pressure required to prevent permanent neurologic injury. At the lower end of this range, we could assume a young, healthy, normotensive patient, with classic vessel anatomy, and an intracranial pressure never deviating from 0 mmHg or regional cerebral blood flow distribution never deviating from parity. Assuming a siphon based physiology, then it should be possible to measure blood pressure at the level of the heart, and maintain blood pressure at the lower limit of autoregulation without causing ischemic neurologic injury. Any small errors created by deviations from a pure siphon system, and some uncertainty as to whether there is a precise lower limit of autoregulation and where it might occur in this patient,11 would be somewhat offset by the fact that, even as perfusion pressure declines below the lower limits of autoregulation, blood flow does not fall into the abyss but instead declines gradually, perhaps still leaving enough circulation to prevent permanent neurologic injury. At the other extreme, if we assume a waterfall-based physiology, we must not only account for a hydrostatic gradient imposed by the sitting position, but we must also take into account the parallel plumbing feeding the waterfall, and the effects that regional variations in intracranial pressure, surgical retractor pressure, head positioning, atherosclerosis, geographic variants of blood vessel distribution, and other factors may have on the flow through contributing vessels, some of which may be critical to patient well-being. Clearly there is a considerable difference between the physiologies described by these 2 extremes.

Simple Study May Not Yield Simple Answer

It is tempting to rush to the animal laboratory to try to mimic and study the exact patterns of physiology during anesthesia and patient positioning. However, such studies will likely reflect the physiology of healthy animals in which the various combinations of heart and head positioning, species-related anatomic variations, and other factors, will not accurately reproduce the conditions of the rare, highest-risk humans. If such studies are eventually performed in animals to better explore the issue of monitoring site versus cerebral well-being as related to siphon versus waterfall hemodynamic models, it must be remembered that measurements of well-being must take into account the watershed regions of brain, eyes, and spinal cord, using techniques such as microspheres, laser Doppler flowmetry, or multidimensional radiologic imaging to quantify regional blood flows, and multiple-lead electrical recordings to assess electrical well-being. Crude assessments of well-being, using transcranial Doppler sonography of conducting vessels, and processed or geographically non-discriminating eletrophysiologic measurements, will simply not address the root of the problem. Unfortunately, attempting to monitor and assess individual patients will be problematic, if for no other reason than that the patients at greatest risk of injury during the head-up position are probably those with some atypical anatomy or baseline physiology. Such patients will be hard to identify, the influence of variations in patient positioning may be impossible to explore in the clinical environment, and data from these patients will be hard to generalize to other high-risk patients.

Absent such evidence, it is tempting to instead analyze and rationalize blood pressure monitoring and management in individual patients, based on core principles. However, we anesthesiologists should be reluctant to choose this approach, recognizing how such a process has ill served us in the past. We need not be reminded that for a period of 3 or more decades, this type analysis of a possible intracranial pressure increase in response to intravenous succinylcholine,12,13 or to “bucking” and coughing in tracheally intubated subjects,14,15 erroneously ascribed increases in intrathoracic pressure and central venous pressure as the operant mechanisms. However, when such concepts were first tested experimentally in the 1980s and ‘90s, neither clinical condition was even remotely related to the long-touted operant mechanism.12-15 Instead, other altogether different mechanisms appeared to be responsible, and the onset, magnitude, and duration of the intracranial pressure increases were not at all what anesthesiologists had long envisioned. There are a sufficient number of similar, faulty analyses in the history of anesthesiology to make us fearful of introducing new errors in management, based on core-principle analysis absent empirical support. However, unlike previous examples involving transient increases in intracranial pressure, the end result of the current discussion of blood pressure management in head-up patients is not to declare a winner of some innocuous academic pillow fight, but instead to optimize patient management for the purpose of avoiding irreversible neurologic injury.

Without the data we need to definitively identify ideal blood pressure monitoring and management in head-up, anesthetized patients, what should we do for contemporary blood pressure measurement and management? It would seem appropriate that our practices should err on the side of providing excessive blood pressure to non-critical tissues, and adequate blood pressure to critical tissues. Such an approach has merit not because we have proven that a modified watershed model of cerebral circulation is operant in head-up patients or that core principles have led us to an unimpeachable conclusion, but instead because such an approach moves us in a management direction away from hypoperfusion (whatever the cause). This approach also has merit because experience tells us that small reductions from normal blood pressure are statistically more likely to produce long-term injury (e.g., from ischemia) than are small elevations in blood pressure (e.g., from hemorrhage or edema formation). Risk of cerebral aneurysm rupture is a notable exception.

In the face of inadequate information, pursuing good outcomes primarily by avoiding bad outcomes is not new to anesthesiologists and nurse anesthetists. Indeed, with an ongoing, decades-long debate about alpha-stat versus pH-stat management of blood gases and pH during clinically induced hypothermia,16 the most commonly accepted management philosophy is directed toward avoiding harm, not pursuing perfection.

It should be remembered that invoking a siphon-related analysis of cerebral perfusion is basically an exploration of the minimal blood pressure required to provide adequate blood flow from the heart, through the brain, and back to the heart, and does not adequately account for the distribution of that blood flow within the brain. It is an analysis of extremes, to determine how far we can push our management approach yet not do harm. Indeed, we are sometimes called upon to transiently push the extremes of systemic blood pressure, to permit the clipping of a cerebral aneurysm, allow the placement of a suture in a critical cardiovascular structure, or ensure adequate perfusion and oxygenation of a fetus. However, these infrequent instances are different from the discussion of blood pressure management in head-up patients. Here, we are not exploring the transient, extreme manipulation of physiology to permit benefit (as in the aforementioned examples), but the prolonged management of blood pressure to avoid harm (e.g., watershed cerebral ischemia).

As such, until we have definitive data proving otherwise, it seems prudent to direct our blood pressure management in head-up patients in a manner that will accommodate for hydrostatic gradients, patient’s baseline blood pressure (with its implications for cerebral autoregulation), and the impact of atherosclerotic and other vascular anomalies, regional intracranial pressure, and head positioning. Such an analysis dictates measuring blood pressure at the level of the most vulnerable tissue (i.e., the brain), and maintaining blood pressure well within the patient’s normal range of blood pressures observed while unanesthetized. This management philosophy is consistent with our historic role as the vulnerable patient’s last homeostatic defense for avoiding injury during anesthesia and surgery.

William L. Lanier, MD
Professor of Anesthesiology
Mayo Clinic
Mayo Clinic Proceedings
Rochester, MN


  1. Cullen DJ, Kirby RR. Beach chair position may decrease cerebral perfusion: Catastrophic outcomes have occurred. APSF Newsletter 2007;22(2):25,27.
  2. Munis J. The problems of posture, pressure, and perfusion. APSF Newsletter 2008; 22(4):82-83.
  3. Cucchiara RF. Hazards of beach chair position explored. APSF Newsletter 2008;22(4):81.
  4. Drummond JC, Hargens AR, Patel PM. Hydrostatic gradient is important—blood pressure should be corrected. APSF Newsletter 2009;24(1):6.
  5. Kirby RR, Cullen DJ. Lower Limit of Cerebral Autoregulation Questioned. APSF Newsletter 2009;24(1):5.
  6. Wilder BL. Hypothesis: The etiology of midcervical quadriplegia after operation with the patient in the sitting position. Neurosurgery 1982;11:530-1.
  7. Black S, Ockert DB, Oliver WC Jr, Cucchiara RF. Outcome following posterior fossa craniectomy in patients in the sitting or horizontal positions. Anesthesiology 1988;69:49-56.
  8. Toole JF. Effects of change of head, limb, and body position on cephalic circulation. New Engl J Med 1968;279:307-11.
  9. Toole JF, Tucker SH. Influence of head position upon cerebral circulation. Studies on blood flow in cadavers. Arch Neurol 1960;2:42-9.
  10. Perkins WJ Jr, Lanier WL, Sharbrough FM. Cerebral and hemodynamic effects of lidocaine accidentally injected into the carotid arteries of patients having carotid endarterectomy. Anesthesiology 1988;69:787-90.
  11. Drummond JC. The lower limit of autoregulation: Time to revise our thinking? Anesthesiology 1997;86:1431-3.
  12. Lanier WL, Milde JH, Michenfelder JD. Cerebral stimulation following succinylcholine in dogs. Anesthesiology 1986;64:551-9.
  13. Lanier WL, Iaizzo PA, Milde JH. Cerebral function and muscle afferent activity following intravenous succinylcholine in dogs anesthetized with halothane: the effects of pretreatment with a defasciculating dose of pancuronium. Anesthesiology 1989;71:87-95. Erratum in: Anesthesiology 1989;71:482.
  14. Lanier WL, Iaizzo PA, Milde JH, Sharbrough FW. The cerebral and systemic effects of movement in response to a noxious stimulus in lightly anesthetized dogs. Possible modulation of cerebral function by muscle afferents. Anesthesiology 1994;80:392-401.
  15. Lanier WL, Albrecht RF 2nd, Iaizzo PA. Divergence of intracranial and central venous pressures in lightly anesthetized, tracheally intubated dogs that move in response to a noxious stimulus. Anesthesiology 1996;84:605-13.
  16. Kern FH, Greeley WJ. Pro: pH-stat management of blood gases is not preferable to alpha-stat in patients undergoing brain cooling for cardiac surgery. J Cardiothorac Vasc Anesth 1995;9:215-8.