Cerebral autoregulation

1. Bedside vignette: has this brain stopped defending its own flow?#

A 9-year-old fell from a balcony and arrived with a severe head injury; an ICP monitor was placed on arrival. On day 3 the MAP is 65 and the ICP 15, so CPP is 50, a number that looks defensible on a standard chart. But this child's injured brain has lost the ability to buffer blood-pressure swings. Across the morning, every dip in MAP is followed, almost beat for beat, by a fall in cerebral oxygenation, and every rise by a rise. The vessels are behaving as passive pipes: flow is tracking pressure rather than being held steady against it.

The useful bedside question is no longer "is CPP above a fixed threshold?" but "has this brain stopped defending its own flow, and if so, what perfusion pressure does it actually need today?" That is the question cerebral autoregulation, and the monitoring built around it, exists to answer. The rest of this page is the physiology behind it.

2. What is cerebral perfusion pressure?#

Before flow, pressure. The brain is perfused by the difference between the pressure pushing blood in and the pressure inside the skull pushing back:

Cerebral perfusion pressure (CPP) is the driving pressure across the cerebral vascular bed: mean arterial pressure (MAP) comes from the arterial line, intracranial pressure (ICP) from an EVD or parenchymal monitor.

CPP is not the same as cerebral blood flow. Flow also depends on how wide or narrow the vessels are, that is, on cerebrovascular resistance (CVR):

This one relationship is the whole point of autoregulation. By changing arteriolar calibre (and therefore resistance), the brain can hold CBF roughly constant even as CPP moves. When the mechanism works, CPP can swing and flow barely changes; when it fails, flow follows CPP directly.

Two clinical consequences follow. First, the same CPP can mean very different things: CPP 60 from MAP 75 and ICP 15 is a healthy brain, while CPP 60 from MAP 100 and ICP 40 is a decompensating one reaching the same number by a very different path. Second, the "right" CPP is age-dependent and, ultimately, patient-dependent. The CPP page covers the number, its age-banded targets, and those paths in full; here we use CPP simply as the horizontal axis of the autoregulation curve.

3. What autoregulation is: the Lassen plateau#

The brain runs a roughly fixed metabolic appetite, about 3.5 mL O₂ per 100 g per minute in an adult and more per gram in a school-age child, and must feed it across a constantly moving systemic blood pressure. It does this by changing the calibre of its arterioles: they constrict when perfusion pressure rises and dilate when it falls. The result is the Lassen plateau, a band of CPP (or MAP) across which CBF stays approximately constant, bounded by a lower limit of autoregulation (LLA) and an upper limit (ULA). 1

Below the LLA the arterioles are already maximally dilated and cannot compensate further, so flow falls as pressure falls. Above the ULA they are maximally constricted, so flow rises as pressure rises. The plateau is not identical in every patient: it is narrower and lower in children, and acute brain injury narrows it or abolishes it altogether.

4. How a single arteriole holds the line#

Three mechanisms act in parallel to hold flow against a swinging pressure, with different time constants and different things that modulate them.

1. Myogenic. Smooth muscle in the arteriolar wall responds to transmural pressure. As pressure inside the vessel rises, mechanosensitive channels (notably TRPM4 and Piezo1) depolarise the muscle, voltage-gated Ca²⁺ enters, and the vessel constricts. This is fast (~5–15 s), survives denervation, and is the dominant short-latency mechanism (originally the Bayliss effect, 1902).
2. Metabolic. Local accumulation of CO₂, lactate, and adenosine, together with shifts in interstitial pH, dilates vessels where metabolic demand is high. This underlies functional hyperaemia and most of CO₂ reactivity.
3. Neurogenic. Sympathetic activation shifts both the LLA and the ULA rightward, narrowing the window; parasympathetic input to the cerebral vessels is small.

5. The three operating zones at the bedside#

The plateau has three operating zones, and the consequence of sitting in each one is different.

• Below the LLA (hypotensive zone). CBF falls with pressure, so even modest hypotension causes ischaemia. This is the territory of secondary injury in severe TBI, of watershed injury in neonatal HIE, and of injury after deep anaesthetic induction.
• On the plateau. CBF is buffered. Routine blood-pressure swings are absorbed by the arterioles, and the bedside team does not have to chase them.
• Above the ULA (hypertensive zone). CBF rises with pressure, so hypertensive surges drive vasogenic oedema, blood-brain-barrier breakdown, and haemorrhagic transformation. This is the enemy in post-recanalisation stroke, the post-cardiotomy hypertensive patient, eclampsia, and malignant hypertension.
• Off the plateau entirely (pressure-passive). CBF tracks pressure linearly and there is no buffer at all; every pressure swing is a flow swing. In severe brain injury the plateau narrows or disappears. 2

A "normal" MAP can therefore be too low or too high for this patient today.

What bends the curve. Chronic hypertension and hyperviscosity shift the whole curve rightward; pregnancy and ageing shift it leftward; acute brain injury narrows it asymmetrically, with the LLA rising and the ULA falling. 3

A note on management. When the bedside picture suggests autoregulation has failed (flow tracking pressure, ischaemia or oedema appearing with pressure swings), the first moves are the cheap ones: correct hypoxaemia, hypercarbia, fever, seizures, pain, and under-sedation, each of which can narrow or shift the curve faster than any haemodynamic manoeuvre. Only then reach for blood-pressure targeting, and, where the monitoring exists, for an individualised perfusion-pressure target (CPPopt, below).

The extreme case: the Cushing reflex. When CPP collapses far enough, the brainstem itself becomes ischaemic and triggers a stereotyped systemic response: a sympathetic surge driving hypertension, vagally mediated bradycardia, and irregular respiration, the Cushing triad. 4 5 It is not a form of autoregulation but a last-ditch attempt to force perfusion pressure back up, and it is a late and ominous sign of critically raised ICP and impending herniation. Treat it as an emergency; raised-ICP management and the herniation cascade are covered on the ICP page.

6. Why pediatric autoregulation is different#

The pediatric autoregulatory plateau is narrower, lower, and more variable than the adult one. 6 Three concerns dominate at the bedside:

1. Narrow plateau. Piglet data suggest the plateau may be only 10–20 mmHg wide in neonates, with wide variation between individuals. 7
2. Low lower limit. The LLA can sit close to baseline MAP, so a normal-looking blood pressure may already be on the pressure-passive segment.
3. Variable closure. Post-arrest, post-cardiopulmonary-bypass, and HIE states all shift the LLA upward unpredictably.

This is why brief hypotension during induction or in the catheterisation lab matters disproportionately in infants, and why individualisation of the perfusion-pressure target is at once most attractive and least validated in the youngest patients. 8

7. How we measure autoregulation at the bedside#

Autoregulation is a property, not a signal: no probe reads it directly. At the bedside it is inferred from a simple idea. When the plateau is intact, slow waves in arterial pressure do not produce matching slow waves in a flow- or pressure-related brain signal, because the arterioles are actively absorbing them. When autoregulation fails, the two move together. Measuring how tightly they move together, as a moving correlation over roughly 5-minute windows, grades autoregulation continuously.

Three indices do exactly this, differing only in the brain-side signal they use:

• PRx (pressure reactivity index) uses ICP. It is the invasive gold standard when an ICP monitor is already in. See the PRx page.
• Mx (mean velocity index) uses transcranial Doppler flow velocity. It is non-invasive and useful before a monitor is placed or after it is removed. See the Mx page.
• COx (cerebral oximetry index) uses NIRS tissue oxygenation. It is fully non-invasive and the workhorse in neonates and in cardiac surgery. See the COx page.

All three read near zero or negative when autoregulation is intact and turn positive when it fails. Their exact thresholds, the underlying math, the trace patterns, and the pitfalls live on their own pages.

8. CPPopt: the individual sweet spot#

Because each brain's plateau sits at a different place, and moves over hours, there is an individual perfusion pressure at which autoregulation works best. This is CPPopt. The idea is straightforward: if you watch one of the autoregulation indices across a range of CPP values over several hours, it tends to trace a U. Autoregulation is worst when CPP is too low (passive collapse) and when it is too high (hyperaemia), and best somewhere in between. The bottom of that U is CPPopt, and the operational target is a few mmHg either side of it.

Two things matter at the bedside. CPPopt is a moving target, not a number to set once: it is re-derived as the brain's state changes. And it is not always obtainable, when the curve is flat there is no optimum to read, and the right move is to fall back to the age-banded CPP floor rather than invent one.

In children CPPopt is lower than the adult-style default and is age-dependent; the pediatric evidence is early but consistent with the adult physiology. 8 The method itself, the dose-response between time spent away from CPPopt and outcome, and the trial evidence are covered in full on the CPPopt page. 10 11

9. Combine with…#

• CPP: the perfusion pressure this curve is plotted against.
• PRx: the invasive gold-standard autoregulation index.
• Mx: the TCD-based non-invasive index.
• COx: the NIRS-based index for the patient without an ICP monitor.
• CPPopt: the individualised, data-driven moving target.
• ICP: raised-ICP management and the herniation cascade.
• Foundations: CO₂ and O₂ reactivity: how PaCO₂ shifts the curve.
• Foundations: pediatric cerebral physiology: age-band CPP and CMRO₂ values.

10. Evidence summary#

11. Self-check#