Intra-Operative Cerebral Protection
Presented at the October 1998 ANZCA Neurosurgical Anaesthesia Special Interest Group Meeting
Lindeman Island, Queensland, Australia
Chris Thompson, Royal Prince Alfred Hospital, Sydney, Australia
(Please use the 'frames' version' if your screen is big enough)
There are no authoritative double-blind randomised controlled human clinical papers which clearly indicate which techniques of cerebral protection improve outcome. Hence, for any action, eg temperature management, one cannot argue, on an outcome basis, that 'doing nothing' is better (or worse) than 'doing something else'! So, how do we make sensible decisions in the absence of unequivocal human outcome studies? An allegory illustrating some possible approaches to such a dilemma is presented.
This discussion intends to:
- Consider possible mechanisms of cerebral injury
- Ask some fundamental questions about these processes
- Examine some hypotheses about the injury process (and how it may be ameliorated)
- Critically evaluate some of the literature
- Show some new data on drug effects in a human carotid endarterectomy model
- Clarify, the the extent that this is possible, how the neuroanaesthetist can provide useful intraoperative cerebral protection in some common clinical situations.
Surgery and anaesthesia interfere with the usual homeostatic mechanisms which protect the brain of an intact individual from both traumatic and physiological insults.
Anaesthetists can confer cerebral protection by utilising therapies which minimise surgical trauma, maintain the physiological milieu of brain cells close to normal and attenuate self-destructive processes on both the macro and micro scales. Unfortunately, the potential for adverse effects, limited appreciation of the benefits and the absence of definitive clinical studies have resulted in confusion and considerable debate as to the relative merits of one drug or technique over another in many situations.
Obvious examples include 'brain shrinkage' to reduce retractor trauma and make the surgeons job easier, and appropriate management of blood pressure is also important.
More difficult issues include manipulation of CO2, temperature and pharmacological interventions.
CO2 management is a particularly complex issue. Lowering PaCO2 by hyperventilation is a common practice in neuroanaesthesia, primarily since it reduces brain volume and cerebral blood flow, improving surgical conditions, reducing bleeding and anaesthetic requirements. A further theoretical benefit from 'inverse steal' has also been proposed, but can only be helpful if regional flow inequality exists. Unfortunately metabolism is not changed; ischaemia is therefore a necessary consequence. The net benefit from hyperventilation of a patient (and the degree to which this should be applied) varies according to the clinical setting. While mild hyperventilation is, on balance, probably beneficial for most neurosurgical procedures, it may be harmful in some settings.
Hypocarbia worsens global ischaemia (where redistribution cannot take place) since it further reduces global cerebral blood flow. Prolonged hyperventilation has been shown to worsen outcome following severe head injury, implying that the brain cannot adequately compensate for global hypocarbia induced vasoconstriction. I think that hyperventilation should not be used in association with deliberate global hypotension.
In temporary aneurysm clipping and similar situations, hypocarbia may also be disadvantageous since if ischaemia occurs, it may be more dependent on dilation of nearby arteriolar collaterals than on the theoretical application of 'inverse steal' to 'watershed ischaemia'. This debate has not been answered. Experimental data from intra-cerebral probes is required to resolve the question. I am 'sitting on the fence' on this one, and aim to keep the CO2 normal if possible.
On the other hand, it seems likely that hyperventilation is capable of protecting against major regional ischaemia, such as occurs during carotid endarterectomy, by inverse steal. I have some nice data demonstrating steal (ipsilateral stump pressure fall and TCD velocity deterioration) during acute increases CO2 in patients with low stump pressures while undergoing carotid endarterectomy. It is important to recognise that carotid patients who are not subject to ischaemia, or with good collaterals, may demonstrate the opposite effect - increasing flow with hypercarbia - since CO2 induced vasodilation may, in these patients, be able to increase blood flow in both hemispheres. That opposite outcomes are possible with the same therapy, depending on the state of the collaterals, helps explain the longstanding difficulty we have had in resolving this question. On balance it seems likely that modest hypocarbia should be deliberately induced in carotid endarterectomy (and other states of gross regional flow inequality or collateral-dependent ischaemia), since it will almost certainly benefit those patients with ischaemia and have no adverse effect if ischaemia is not present.
Traditionally the degree of 'cerebral protection' conferred by a given agent has been equated with its ability to provide metabolic suppression. This simple hypothesis has proved inadequate to explain the degree of benefit which arises, for example, from the use of barbiturates or mild hypothermia.
Temperature management is still disputed. Deep hypothermia and some anaesthetic agents are well known to provide cerebral protection. Cell survival under circulatory arrest is prolonged by hypothermia, however the protected period is almost twice as long as would be expected from metabolic suppression alone. The 'extra' benefit seen with these agents is most likely due to reduced release of metabolically active mediators (such as excitatory amino acids) during or after the period of ischaemia. Anything which reduces metabolic activity will reduce production of these mediators and hence reduce the severity of an ischaemic insult. The search is on for clinically useful mediator inhibitors, however none, with the possible exception of nimodipine, are in common clinical use.
Mild (34°C) hypothermia has been advocated for routine neurosurgery. Hyperthermia is definitely bad. Increasing degrees of hypothermia (mild, moderate or profound) are increasingly protective against ischaemia. On logical grounds mild hypothermia must be be protective! Some papers have suggested that the degree of clinical benefit from mild hypothermia greatly exceeds that which would be predicted from the magnitude of the fall in metabolic rate alone. Adverse effects include wound infection and persisting hypothermia in the postoperative period which may cause increased oxygen consumption, cardiac ischaemia, patient discomfort and confusion. The rate of reduction of temperature is so slow that it must be prophylactically administered to all at-risk patients - exposing all to risk, but only a few to benefit.
Hence mild hypothermia is beneficial but not without hazard. I think that that mild hypothermia is strongly indicated in situations in which cerebral ischaemia is a substantial issue (carotids, aneurysms, deep lesions etc), however, it should not be used, ad-hoc, for all neurosurgical cases! If indicated, induction of hypothermia should commence as soon as the patient goes to sleep, by infusion of cold IV fluids, prompt removal of all clothing and the use of room temperature forced-air devices. Rewarming with forced air devices is essential and needs adequate time, particularly for obese patients. Even if the core temperature does not return to normal at the end, shivering is reduced and the patient feels better if their periphery is warm.
Barbiturates and other 'cerebral protective' drugs are also a subject of controversy, despite years of investigation. Initial hypotheses suggesting that protection was, like cold, directly due to metabolic suppression are probably not correct.
Under normothermic gobal total ischaemia or global total anoxia, barbiturates provide little or no direct benefit . Barbiturates only reduce the rate of ATP fall for the first 20-30 seconds. This is because profound ischaemia flattens the EEG in 15-20 seconds, after which time the rate of ATP fall will be the same regardless of the presence or absence of barbiturates. This contrasts with hypothermia, which prolongs cell survival and reduces the rate of ATP fall in proportion to the degree of hypothermia.
Many workers evaluated barbiturates (and other therapies) with brief (3-5 minute) periods of total normothermic ischaemia or anoxia . While these often demonstrate some benefit, many underplay the fact that the benefit is of such brief duration as to be clinically irrelevant. Only models of incomplete ischaemia show clinically useful degrees of protection.
It is highly likely that the benefits of barbiturates are essentially indirect, and arise from mechanisms such as inverse steal, excitatory neurotransmitter mediator inhibition, reductions in ICP (leading to improvements in cerebral perfusion pressure, particularly so at low perfusion pressures) and retractor ischaemia etc, rather than reduction in metabolic rate per se. Inhibition of the release of excitatory neurotransmitters (aspartate, taurine, glutamate and gaba) has been demonstrated even if barbiturates are administered after the period of ischaemia, suggesting that at least some of the benefit occurs after reperfusion.
Drug-induced inverse steal requires reductions in cerebral metabolism (usually to the point of EEG burst suppression) with an agent which maintains flow-metabolism coupling ('autoregulation'). This will result in reductions in blood flow in well perfused regions with subsequent increases in upstream perfusion pressures leading to the redistribution of this 'excess' flow down pressure gradients to more ischaemic areas.
I have looked for evidence of inverse steal in carotid endarterectomy patients by continuously acquiring EEG, radial arterial pressure, 'stump' (distal internal carotid artery) pressure and transcranial doppler data during methohexitone or sevoflurane induced EEG changes. Stump pressure is measured before, during and after clamping by a needle in the distal internal carotid, and during surgery by means of an intraluminal Fogerty balloon catheter, which is used by my surgeon in place of a distal ica clamp. EEG is recorded raw and with an analog RMS to DC converter device, which allows me to easily quantify burst suppression or interhemishperic total EEG power differences. CO2, temperature and arterial pressure are maintained as contstant as possible (nor-adrenaline is used, if needed, to increase arterial blood pressure).
Stump pressure is a poor predictor for the presence of cerebral ischaemia, since there are many reasons why the measured stump pressure threshold for ischaemia may vary considerably from one individual to another. These include variability of the lower limit of the autoregulatory threshold due to chronic hypertension, the presence of cerebral vasodilators (volatile agents, hypercarbia) and vasoconstrictors (hypocarbia) which will increase (or decrease) flow for a given pressure, variability in metabolic requirements of the brain (drugs, hypothermia), variability of ICP (a significant part of the CPP equation) and technical errors like incorrect calibration of transducers, etc.
Despite this, stump pressure is our best indicator of ipsilateral cerebral perfusion pressure during carotid endarterectomy. Once clamped, the distal internal carotid provides a zero flow liquid tube which is directly connected to the middle cerebral artery. Stump pressure will reliably measure the arterial side of the cerebral perfusion pressure equation - much more accurately than the radial arterial pressure!
A most important observation is that stump pressure is profoundly affected by arterial pressure. It is vitally important to maintain adequate arterial pressure during carotid endarterectomy, and this is reinforced every time I measure stump pressure throughout the case - particularly if the stump pressure is low! Another observation is that autoregulation of flow is an extremely rapid phenomenon, as seen in this slide showing changes in stump pressure and flows during crossclamping and after ventricular ectopic beats.
It is true that low flow, not low perfusion pressure per se, causes ischaemia, however, at low cerebral perfusion pressures arterial vasodilation becomes maximal, cerebral blood flow autoregulation fails and flow becomes perfusion dependent. Failure of the circulation to provide enough perfusion pressure to ensure adequate flow is the root cause of hypotensive ischaemia. Continuous measurement of stump pressure continuously monitors this 'adequacy of the circulation' and permits evaluation of the effect of anaesthetic interventions (drugs, systemic blood pressure, CO2, etc) on the perfusion pressure of that hemisphere. When combined with EEG and TCD measurement, we can relate stump pressure to changes in cerebral metabolism and bloodflow, an ideal situation for examining inverse steal.
I have found that barbiturate-induced burst suppression consistently induces significant inverse steal in an EEG-effect-dependent manner, to such an extent that very significant increases in stump pressure (up to 50 mmHg) may be seen. Increases of stump pressure of this magnitude explains a significant part of the 'protective' effect seen with barbiturates in carotid endarterectomy. These three diagrams (pre-clamp, post-clamp, and after barbiturates), illustrate the mechanism of the pressure rise and flow redistribution of inverse steal.
This elevation in stump pressure is directly related to the instantaneous electrical activity of the brain. The stump pressure rises during periods of electrical silence as short as 2 seconds and falls again during 'bursts' or brief periods of electrical activity. This is graphically illustrated in these slides (1, 2, 3, 4, 5 and 6). Stump pressure increase is maximal when the EEG is nearly flat. I have seen TCD confirmation of inverse steal in a patient with particularly low low stump pressures (mean of 30mmHg) in whom barbiturates increased both stump pressure and flow velocity on the ischaemic side, while reducing flow velocity on the non-operated side. In the same patient, transient elevation of arterial CO2 caused hyperaemia of the non-ischaemic hemisphere and reduced both stump pressure and perfusion of the ischaemic hemisphere, suggesting that modest hyperventilation is beneficial in ischaemic carotid endarterectomy.
Volatile agents not only suppress metabolism but are also cerebral vasodilators; in contrast with the barbiturates, cerebral blood flow does not change much when the EEG is rendered flat. In the same carotid model, stump pressures did not rise when sevoflurane was administered to the same EEG effect as barbiturates, suggesting that inverse steal does not occur with these agents. It is worth noting sevoflurane, when titrated to burst suppression, causes much more hypotension than methohexitone.
Extrapolation of this data to situations other than carotid endarterectomy may be unwise. Profound deliberate hypotension does not result in the gross blood flow inequalities seen in carotid endarterectomy, so benefits from redistribution of flow (inverse steal) are likely to be trivial. Theoretically the best recipe for cerebral protection in this setting involves both metabolic suppression (hypothermia and EEG silence) and cerebral vasodilation (nitroprusside and/or sevoflurane). Total cerebral blood flow at any given pressure (under equal conditions of electrical silence) is likely to be higher with sevoflurane than with a barbiturate. Hence it is likely that an agent like sevoflurane may be better than barbiturates in this setting, since it induces the triad of metabolic suppression, hypotension, and vasodilation.
When confronted with the need for 'cerebral protection', a clear understanding of the underlying mechanisms of both injury and treatment is required to decide on the best approach. Appropriate monitoring (EEG, EP, stump pressure, TCD) is needed to optimise therapy. Barbiturates, mild hypothermia, mild hypocarbia and hypertension are recommended for regional ischaemia, eg carotids. Profound induced hypotension is best induced with vasodilators like nitroprusside, rather than cardiac depressants or haemmorrhage.
