Hyperventilation in Head Injury^*: A Review FREE TO VIEW

Modulation of Paco₂ has been used for > 40 years,¹ first in neuroanesthesia and subsequently also in neuro-intensive care. Preliminary work has shown that the volume of the swollen brain could be decreased by lowering Paco₂. With the realization that raised intracranial pressure (ICP) is a significant, treatable problem in patients with traumatic brain injury (TBI), hyperventilation became a cornerstone in the management of TBI and has remained so for decades. Hyperventilation lowers ICP by the induction of cerebral vasoconstriction with a subsequent decrease in cerebral blood volume.² The downside of hyperventilation, however, is that cerebral vasoconstriction may decrease cerebral blood flow (CBF) to ischemic levels. Already in 1942, a slowing of the EEG was observed during active hyperventilation and was interpreted as a sign of cerebral ischemia, thus illustrating the potentially harmful effects of hypocapnia.³ Over the past decade, relatively more attention has been paid to the adverse effects of hyperventilation and concern seems to exceed enthusiasm. This change in attitude would appear more emotional than data-driven and reflects the lack of conclusive data.

The aim of this review was to consider the effects of induced hypocapnia both on systemic physiology and on the physiology of the intracranial system, with specific reference to TBI. We chose to focus this review on TBI, as much of the research on hyperventilation has been conducted in this field and less information exists on other acute cerebral disorders, such as aneurysmal subarachnoid hemorrhage or stroke. This review is based on an extensive evaluation of the literature, and to this purpose we selected relevant experimental and clinical articles on hyperventilation from among > 5,000 citations found on MEDLINE since 1966. We have indicated explicitly in the text when expert opinion is being expressed rather than available evidence being quoted.

A remarkable confusion exists on terminology. What is usually referred to as hyperventilation is, in fact, hypocapnia. Since a reduction of Paco₂ below the normal level (40 mm Hg) is obtained by increasing the alveolar ventilation, hyperventilation became synonymous with hypocapnia. In this review, we will use the less precise (but much more common) term hyperventilation. Hyperventilation may be defined as “the induction and/or maintenance of levels of CO₂ tension in the arterial blood below the normal range.” In this sense, normal levels of Paco₂ should be corrected for barometric pressure at different altitudes.

The CNS, accounting for 2% of body weight (average weight of the brain, 1,300 to 1,500 g), has a high energy requirement. The cerebral oxygen consumption is 3.5 mL per 100 g/min, which corresponds to 20% of total body oxygen consumption. Under normal conditions, CBF is maintained at a constant flow rate of 50 to 60 mL per 100 g/min, with 50 mL of oxygen being extracted every minute from 700 to 800 mL of blood (Table 1 ). The extraction rate for oxygen is high, and the mean arteriovenous difference of O₂ for the CNS is 6.3 mL per 100 mL of blood. CBF depends on the differential pressure between the arterial and the venous side of the cerebral circulation, and is inversely proportional to cerebral vascular resistance. Pressure on the venous side of the capillary bed cannot be measured, and ICP, which is extremely close to venous pressure, is used for estimating the cerebral perfusion pressure (CPP). CPP is calculated as the difference between mean arterial pressure and ICP.

Normal ICP values in adults are < 10 mm Hg, and a threshold of 20 mm Hg is usually accepted for starting active treatment. A CPP of 60 mm Hg is commonly accepted as the minimum value necessary for adequate cerebral perfusion.⁴ Two important concepts are:

The Monro-Kellie doctrine states that the total volume of the intracranial contents (ie, brain tissue, blood, and cerebrospinal fluid [CSF]) remains constant as these are contained within a rigid compartment (the skull), as follows:

An increase in the volume of one of these compartments can initially be compensated for by the displacement of parts of the other components. Cerebral veins can be compressed, resulting in decreased cerebral blood volume, and the volume of the CSF compartment can decrease due to a combination of increased resorption and the displacement of CSF toward the spinal compartment. As volume increases, compensatory mechanisms are exhausted, and a further increase in volume results in a sharp rise of ICP, leading to the volume-pressure curve depicted in Figure 1 .

The high metabolic demands of the brain in combination with the limited storage of substrates necessitate maintaining CBF levels within normal ranges. In physiologic circumstances, this is effected through a number of mechanisms, which are commonly referred to as autoregulation. CBF increases with vasodilatation and decreases with the constriction of cerebral arteriolae, termed cerebral resistance vessels. These vessels respond to changes in systemic BP (pressure autoregulation), blood viscosity (viscosity autoregulation), and metabolic demand, maintaining CBF levels within limits that are appropriate to meet metabolic demands. Pressure autoregulation is shown in Figure 2 .

CBF is functionally coupled to the regional cerebral metabolism as expressed in the Fick equation CMRO₂ = CBF × AVDO₂, in which CMRO₂ is the cerebral metabolic rate of oxygen and AVDO₂ is the cerebral arteriovenous difference of oxygen content. CO₂ reactivity refers to the response of cerebral vessels and, consequently, of CBF to changesin Paco₂. An increasing CO₂ tension relaxes cerebral arteries in vitro.⁵In vivo, very localized perivascular changes of Paco₂ or pH can change the vascular diameter, indicating that elements of the vascular wall are responsible for effecting changes in the diameter of vessels. Both vascular cells (ie, the endothelium and smooth muscle) and extravascular cells (ie, the perivascular nerve cells, neurons, and glia) may be involved. In the clinical situation, CBF changes approximately by 3% for each millimeter of mercury change in Paco₂ over the clinically important range of 20 to 60 mm Hg in patients with TBI.⁶⁷ Hypoventilation resulting in hypercarbia causes vasodilatation and increased CBF, while hyperventilation results in vasoconstriction and decreased CBF.

The mechanisms underlying the three forms of autoregulation (ie, pressure, viscosity, and metabolic) have not been precisely unraveled to date, but, compared to mechanisms underlying CO₂ reactivity, the differences are recognized. Whereas pressure autoregulation seems to be located in the pial arteries, with a diameter > 50 μm, CO₂ reactivity involves smaller pial arteriolae.⁸ Different data have been obtained from in vitro and in vivo experiments. In vitro, the middle cerebral artery constricts when it is exposed to raised extracellular pH. In contrast, in vivo, large intracranial vessels are not significantly affected by changes in Paco₂.⁹ Further, in autoregulation the vascular endothelium-derived nitric oxide (NO) or endothelium-derived relaxing factors are important, but some debate exists about whether these factors are involved in the maintenance of basal CBF or actually couple regional CBF to metabolism. An astrocyte-mediated coupling of synaptic activity to local vasodilation has been proposed and will need further evaluation.¹⁰

Vessel caliber follows changes in arterial CO₂ by responding to the pH in the perivascular space without molecular CO₂ and bicarbonate ions acting independently on cerebral vessels.¹¹¹² Nevertheless, changes in pH may exert their effect on smooth muscle tone through second messenger systems or by altering the calcium concentration in vascular smooth muscles directly. Various agents have been identified as potential second messengers, including prostanoids, NO, cyclic nucleotides, potassium, and calcium.

Prostanoids are potent vasodilators, activating adenylate cyclase and increasing cyclic adenosine monophosphate, and are considered to be important regulators of the cerebral circulation in neonates but seem less important in adults. NO, which is produced by a family of NO synthase enzymes in brain vascular endothelial cells, in perivascular nerves, as well as in neurons and glia, increases the intracellular concentration of cyclic guanosine monophosphate, causing vasodilation. Although NO has been shown to act as a vasodilator in response to hypercapnia and acidosis, it cannot account for total vasodilation, as a significant portion (10 to 70%) still occurs when NO synthase is inhibited. The role of NO appears to be complex. It has been hypothesized that a vasodilatory signal is constantly produced by the brain through the synthesis of NO and that an additional signal such as hypercapnia may act on the baseline regulation of the vascular tone. Cyclic nucleotides, such as cyclic adenosine monophosphate and cyclic guanosine monophosphate, reduce the entry of calcium into vascular smooth muscle, and exert vasodilatory effects either directly or in a permissive role, allowing hypercapnia to exert its vasodilatory effects.

The opening of potassium channels indirectly reduces the influx of extracellular calcium into the cell, reducing vascular smooth muscle tone. In contrast, the disturbance of the calcium homeostasis will lead to increased intracellular calcium concentrations, resulting in vasoconstriction. This constitutes one of the reasons underpinning investigations on the efficacy of calcium channel blockers in patients with TBI.

The importance of the systemic effects of hyperventilation is often underrecognized. In some reviews,¹³ guidelines,¹⁴ editorial comments,¹⁵¹⁶ research syntheses,¹⁷ or systematic reviews¹⁸ little or no attention has been directed to the systemic effects of hyperventilation. Systemic effects are multifactorial and interrelated, affecting multiple sites of the body. Substantial differences exist between active hyperventilation (when the subject voluntarily increases his ventilation) and passive hyperventilation (by means of artificial ventilation). In the former, autonomic flow is markedly affected, while in the latter the effects of CO₂ are combined with those of the complex interaction between artificial ventilation and hemodynamics. Additionally, when hyperventilation is applied for reducing ICP, it is usually combined with a number of concurrent interventions such as sedation, paralysis, and increased fluid input.

Positive-pressure ventilation increases lung volume and intrathoracic pressure (ITP), even when a normal level of arterial Pco₂ is maintained, affecting systemic hemodynamics and lung physiology. It is likely that the induction of hyperventilation enhances this effect, as an increase in alveolar ventilation is necessary for inducing hypocapnia.

This may be achieved by increasing the tidal volume and/or the respiratory rate, or by decreasing the dead space. The most appropriate way for inducing hypocapnia has not been determined, but it is usually effected by increasing tidal volume. In stable patients, this has little effect on ITP, but sudden profound hyperventilation may cause marked hemodynamic perturbations,¹⁹ particularly in patients with relative hypovolaemia. An increase in ITP has the following four effects:

The overall effect is that venous return is impaired. This adverse effect can in part be compensated for by an increase in abdominal pressure and intraabdominal vascular pressure due to a more pronounced descent of the diaphragm following an increase in lung volume.²⁰ Increased abdominal pressure, however, may have an adverse effect on ICP.²¹ In the clinical setting, a reduced lung compliance (resulting frequently from the concomitant occurrence of pneumonia, pulmonary contusions, or ARDS) will limit the increase in ITP even if alveolar pressure increases.

The effect of positive-pressure ventilation on RV performance depends on the degree to which venous return (preload) is compromised and pulmonary vascular resistance (afterload) is affected. If compensatory mechanisms are inadequate or if a dysfunction of the RV was already present,²² the most common and important hemodynamic effect of an increase in ventilation is a decrease in cardiac output due to a decrease in the pressure gradient for systemic venous return.

The effects of positive-pressure ventilation on LV function are less important but may also lead to a reduction of cardiac output. This is primarily caused by a reduced venous return as a consequence of decreased LV end-diastolic volume, but also by a reduction of LV diastolic compliance. This may result from a septal shift (ie, ventricular interdependence) due to RV dilatation,²³ by pericardial volume limitations²⁴ or by an increase in lung volume resulting in a direct mechanical compressive effect of the expanding lung on the cardiac fossa.²⁵²⁶ These considerations emphasize the importance of maintaining normovolemia in patients with TBI in general and particularly when artificial ventilation is employed.

A beneficial effect of artificial ventilation is the reduction of LV afterload. In fact, LV systolic pressure load is represented more accurately by LV pressure relative to ITP, and an increase in ITP, reducing the transmural LV pressure, decreases the LV afterload. This mechanism may compensate for the reduction in preload and the worsening of LV compliance, thus maintaining a stable cardiac output²⁷ and may even improve cardiac output when LV end-diastolic volume is preserved.²⁸

A fall of Paco₂ is associated with a primary decrease in extracellular H⁺ concentration.²⁹ The cellular membranes, particularly the blood-brain barrier, are relatively impermeable to hydrogen ions, but permit a rapid diffusion of CO₂. Therefore, the intracellular hydrogen ion concentration is scarcely influenced by changes in extracellular pH but can be altered by changes in Paco₂. The CO₂ passes through the membrane, and, once inside the cell, is able to hydrate and ionize, thus producing hydrogen ions.³⁰ Following the onset of hyperventilation, a rapid efflux of H⁺ occurs within 10 min. In the extracellular fluid, H⁺ combines to HCO₃⁻ to produce CO₂. The extent of this compensatory reaction is, however, not very efficient and, if hypocapnia is prolonged, alkalosis develops.²⁹

A more efficient compensatory mechanism is effected by the kidney. A reduction in cellular Paco₂ in tubular cells induces an increase in intracellular pH, a reduction of H⁺ secretion, and a loss of bicarbonate, together with a decreased excretion of ammonium (NH⁴⁺).²⁹ This response begins within 2 h but is fully effective for 2 to 3 days.³¹ In the mammalian nervous system, intracellular pH is one of the most tightly regulated parameters. The maximum change in H⁺ concentration that can be tolerated is approximately 0.0005 mmol.³²

One of the effects of intracellular alkalosis is the activation of glycolysis, which occurs as a consequence of the modulation of the rate-limiting enzyme phosphofructokinase.³³³⁴ The effect on pH of lactic acid production provides a homeostatic mechanism for generating H⁺ ions to rapidly counteract a state of intracellular alkalosis.³²³⁵³⁶ The increase of lactate levels during hypocapnia, which develops in the absence of any failure of oxidative metabolism, furnishes H⁺ to compensate for the extracellular reduction of H⁺. As a consequence of the shift of H⁺ from the cellular to the extracellular compartment, an opposite movement of K⁺ (and Na⁺) from extracellular fluid to the cell occurs. The resulting hypokalemia is, however, typically mild.²⁹ The increase in cellular phosphorylation causes a rapid shift of phosphate from the extracellular fluid into the cell, with an associated reduction in plasma phosphate concentration.³⁷ In patients with severe alkalosis, albumin releases H⁺, and the binding of Ca²⁺ is increased, resulting in a reduction of the ionized fraction of calcium,³⁸³⁷ which may result in clinical symptoms including bradycardia and heart block, or even heart failure and cardiac arrest. Such serious complications are rare unless the concentration of ionized calcium falls to < 0.8 mmol (3.2 mg/dL).³⁹ It may be supposed that similar effects also pertaining to Mg homeostasis are likely, and in this regard we note that low Mg levels occur frequently following TBI.⁴⁰ In our opinion, particular attention should, therefore, be focused on electrolyte concentrations in general following TBI, and in particular if hyperventilation is employed.

Alkalosis increases the affinity of hemoglobin for O₂ and displaces the dissociation curve to the left. The following two compensatory mechanisms counteract this leftward shift: a rapid increase in lactate production⁴¹; and the induction of enzymatic activity. The increased intracellular pH activates glycolysis, increases the activity of 2,3-diphosphoglycerate (DPG) mutase and reduces the activity of DPG phosphatase.³⁰ These enzymatic adjustments lead to an increased concentration of 2,3-DPG, which over a period of several hours may normalize the dissociation curve. The influence of the Bohr effect on O₂ affinity varies inversely with the degree of hemoglobin saturation. Shifts of the dissociation curve are therefore more relevant to the venous side of the circulation. The importance of the Bohr effect on changes in venous cerebral Po₂ has been studied by various authors. Gotoh et al⁴² found a decrease of jugular Po₂ due to the Bohr effect of approximately 2.4 mm Hg. Cruz et al⁴³ reported a moderately disproportional increase in jugular bulb oxygen saturation (SjO₂) relative to jugular bulb Po₂ when pH increased to > 7.6.

Hypocapnia and respiratory alkalosis may also affect the pharmacokinetic metabolism of drugs in many of the following ways: changes in distribution due to variations of the perfusion of organs; changes of ionization of the drug due to a change in blood pH; changes in solubility and transmembrane diffusion; and changes in protein binding. Finally, the urinary excretion of some drugs also may be altered by changes in urinary pH.

Hypocapnia decreases perfusion in most of the body organ systems, including the heart,⁴⁴ the liver, the gut,⁴⁵⁴⁶ skeletal muscle,⁴⁷ and skin.⁴⁸ A reduction in coronary perfusion due to hypocapnia may cause increased risk for cardiac ischemia in patients with preexisting coronary artery disease. Kazmaier et al⁴⁴ found a mild increase in systemic vascular resistance and a mild reduction in cardiac index when passive mild hyperventilation was employed in patients with coronary artery disease. Despite the absence of significant changes in coronary perfusion pressure and myocardial blood flow, a reduction in coronary sinus Po₂ and oxygen saturation have been reported. The risk of coronary spasm is increased during hypocapnia, and in fact active hyperventilation has been used for the noninvasive diagnosis of coronary spasm.⁴⁹ The kidney is the main organ involved in the compensatory control of pH during chronic hyperventilation.⁵⁰⁵¹ Urinary electrolyte changes (ie, an increase in Na⁺ or a decrease in K⁺) together with a pH increase are part of the compensatory mechanisms in cases of alkalosis.

In the lung, respiratory alkalosis induces vasodilatation of pulmonary vessels⁵² and bronchoconstriction.⁵³⁵⁴ The cumulative resulting effect is a reduction in Pao₂ due to a ventilation/perfusion mismatch.⁵⁵ In patients with severe head injury, Turner et al⁵⁶ found a decrease in Pao₂ from 115 to 99.5 mm Hg after 1 h of hyperventilation. Other clinical research has suggested further adverse pulmonary effects, including increased airway permeability,⁵⁷ dysfunctional surfactant levels,⁵⁸ and reduced lung compliance.⁵⁹ Further, Laffey et al⁶⁰ showed in an experimental study that hypocapnic alkalosis may potentiate ischemia-reperfusion-induced lung injury. Ventilation strategies that are commonly employed in patients with TBI include high tidal volume and low positive end-expiratory pressure, and these may increase the risk of worsening acute lung injury,⁶¹⁶²⁶³ both as a consequence of an increase of lung stretch and the reversal of a “protective effect” of hypercapnia. These potentially adverse effects of hyperventilation on pulmonary function are particularly relevant to the treatment of TBI as approximately 20% of patients experience concomitant acute lung injury⁶⁴ and the incidence of pneumonia has been reported to be as high 40 to 50%.⁶⁵

Hyperventilation has been used in the management of severe TBI for > 40 years since Lundberg et al⁶⁶ reported its use to lower elevated ICP in 1959. Hyperventilation reduces ICP by causing cerebral vasoconstriction and a subsequent reduction in cerebral blood volume.² Fortune et al⁶⁷ showed that decreasing arterial Pco₂ to 26 mm Hg in eight healthy individuals decreased cerebral blood volume by 7.2% and further decreased CBF by 30.7%. Obrist et al⁶⁸ showed a beneficial effect of hyperventilation on ICP in 15 of 31 patients with severe TBI but at the same time demonstrated a reduction in CBF in 29 of 31 patients. Several investigators have reported⁶⁹⁷⁰ that the relationship between Paco₂ and ICP is not linear, and that the greatest effect is between Paco₂ values of 30 and 50 mm Hg in humans. In experimental studies over wide ranges of Paco₂, a sigmoid relation between ICP and Paco₂ has been found.⁷¹

In a clinical study of 94 patients with severe head injury, Yoshihara et al⁷² found that a blood volume change of only 0.5 mL was necessary to produce an ICP change of 1 mm Hg. Consistent with the concept of the pressure-volume curve (Fig 1), a lower blood volume was necessary to produce a significant ICP change in patients with reduced compliance. Further, it was shown that the effects on ICP were greater during hypercapnia than during hypocapnia. Similar results have also been reported by Stocchetti et al,⁷³ who calculated a mean (± SD) blood volume change of 0.72 ± 0.42 mL for each millimeter of mercury of change in Paco₂. Surprisingly, only a few studies have addressed the important question of whether beneficial effects on ICP remain present during prolonged hyperventilation. Muizelaar et al¹² have stressed that the vasoconstrictive effect will be diminished in prolonged hyperventilation as the pH of the perivascular spaces normalizes after 24 h. They further demonstrated in experimental studies that a rebound vasodilation may occur along with a risk of increasing ICP following the discontinuation of hyperventilation.

A major concern in treating raised ICP by hyperventilation is the risk of inducing cerebral ischemia, either globally or regionally. As in stroke, the risk of ischemic damage is dependent on the extent and duration of low-flow states (Fig 3 ). In the early posttraumatic phase, both global and regional CBF are markedly decreased,⁷⁴⁷⁵ and the presence of low CBF early following TBI is significantly associated with early mortality and poorer outcome.⁷⁶⁷⁷ CBF can be measured, directly or indirectly, by a number of methods, none of which, however, are easily available at the bedside in the ICU environment. CBF measurements with radioactive ¹³¹Xe were introduced for clinical use in the 1970s, but this technique was later banned from clinical use because of radiation dangers. Following the introduction of faster multislice CT scanners, Xe-CT scans became a standard technique for measuring CBF with the use of stable, nonradioactive Xe during CT scans of the brain. Inhaled Xe, which is freely diffusable from the lungs to the blood, and from the cerebral vasculature to the brain tissue, can be detected in the brain via CT scan because it increases the attenuation of x-rays.⁷⁸ Direct measurements of CBF can further be performed with positron emission tomography (PET) scanners, which offer the additional benefit of assessing metabolic parameters. PET scanning is, however, available in only a few research centers; provides, as is also the case in stable Xe-CT scanning, only momentary information; and involves transport from the ICU environment for longer periods of time. Indirect measurements of CBF can be performed with transcranial Doppler ultrasonography techniques, which permit measurements of blood flow velocity through the basal intracranial arteries. Blood flow velocity, however, does not directly correspond to CBF, as no information is available on the diameter of the cerebral arteries.

Using Xe-CT scanning to quantify regional CBF, Bouma et al⁷⁷ found CBF values below ischemic thresholds of 18 mL per 100 g of tissue per minute in 31% of TBI patients. In a retrospective analysis of TBI patients with subdural hematomas, Marion et al¹³ observed the lowest CBF within the first 24 h following injury ipsilateral to the hematoma. Studies with transcranial Doppler ultrasonography have also shown a low-flow velocity state in the early phase after injury, occurring in 63% of patients.⁷⁹

As low CBF is common in the first 24 h after a TBI, there is particular concern for aggravating the risk of ischemia by the institution of hyperventilation.¹³ In healthy volunteers, Raichle et al⁸⁰ described a 40% decrease in CBF 30 min after decreasing Paco₂ by 15 to 20 mm Hg. The response, however, was transitory, and after 4 h, CBF was restored to 90% of baseline values. Clinical studies in patients with TBI have shown a 3% change in CBF per millimeter of mercury change in Paco₂, but the response was lower in patients with lower CBF levels.⁸¹

Various clinical studies^6882838485 have confirmed an adverse effect of hyperventilation on CBF levels in patients with TBI. McLaughlin and Marion⁸⁶ further showed increased CO₂ vasoresponsivity in contusions and the surrounding penumbra, and they hypothesized that this possible hypersensitivity in combination with relative hypoperfusion may render such lesions particularly vulnerable to secondary ischemic injury, which may be aggravated by hyperventilation. The ultimate question, however, is whether the observed reduction in CBF following hyperventilation indeed leads to clinically significant ischemia, as evidenced by metabolic studies. Diringer et al,⁸⁷ for instance, showed that brief moderate hyperventilation did not impair global cerebral metabolism and oxygen extraction in patients with severe TBIs, despite a clear decrease in global CBF level. Cruz⁸⁸ has argued that the decrease in CBF level following hyperventilation is acceptable as long as the metabolic parameters are not deranged.

The monitoring of cerebral oxygenation has received considerable attention in view of the significant risk of hyperventilation to decrease CBF levels and possibly to induce/aggravate ischemia. Clinical studies have focused on jugular bulb oximetry and the monitoring of brain tissue oxygen tension (Pbro₂).

In jugular bulb oximetry, SjO₂ is monitored either continuously with fiber optic techniques or intermittently from blood sampling. It is therefore a global technique providing information on the oxygen extraction from the cerebral venous blood draining via that particular vein. However, this does not necessarily reflect hemispheric values as cross-flow may exist, with one jugular vein being more dominant.⁸⁹ It is generally preferred to measure/sample in the dominant vein.

Under normal circumstances (eg, awake or under normal hemoglobin concentration) the SjO₂ ranges from 55 to 70%. Values below 50 to 55% are generally regarded to represent global cerebral hypoperfusion with an increase in cerebral oxygen extraction.⁹⁰ Additional information can be obtained by calculating AVDO₂ or by determining the oxygen extraction fraction. Some studies⁹¹⁹² have shown that forced hyperventilation, although normalizing ICP, can lead to significantly reduced cerebral oxygenation. Other studies,⁹³⁹⁴ however, have described SjO₂ values of > 55% with a concomitant reduction in ICP. In the experimental situation, Sutton et al⁹⁵ found a significant drop in venous oxygen content following hyperventilation in two of six animals studied, accompanied by a decrease in phosphocreatine level, which was rapidly reversible after reestablishing normocapnia.

Cruz and colleagues^88969798 investigated the so-called flow-metabolism coupling and showed that in approximately 20% of patients with elevated ICP blood flow outstrips cerebral metabolic demands. Hyperventilation in this subgroup may lower CBF and improve ICP without reducing cerebral oxygenation. Cruz and colleagues^88969798 have proposed the concept of optimizing hyperventilation on the basis of SjO₂-derived parameters, aiming to both normalize ICP and decrease the cerebral extraction of oxygen by manipulating hyperventilation, to Paco₂ values ranging from18 to 30 mm Hg. Cruz et al⁸⁸ have claimed that this approach yielded better patient outcomes compared to CPP-directed therapy. However, others⁹⁴ have argued that, even with careful SjO₂ monitoring, the risk of inducing iatrogenic ischemia with hyperventilation to Paco₂ levels < 30 mm Hg is too large, and therefore the physician should adhere to CPP-based therapy by maintaining Paco₂ levels at > 30 mm Hg.

In contrast to SjO₂ monitoring, Pbro₂ monitoring is a regional technique. Most studies on monitoring Pbro₂ also have shown a deleterious effect of hyperventilation on cerebral oxygenation. Continuous Pbro₂ monitoring became possible when miniaturized probes, which can be inserted into the cerebral cortex, were manufactured. The first probe was a polarographic, Clark-type sensor, in which a cathode and an anode were contained in a membrane that was only permeable to oxygen. When oxygen diffuses from the tissue into the probe, it generates an electric current between the cathode and anode that is proportional to the oxygen tension. Subsequently, additional technologies for Pbro₂ monitoring (ie, colorimetric systems) became available.⁹⁹ All systems display numeric values, expressing the oxygen tension in millimeters of mercury. Normal values in the brains of various species, including humans, are > 20 mm Hg. Prolonged and profound reductions below this value have proven to be an independent predictor of unfavorable outcome and death.¹⁰⁰

Hemphill et al¹⁰¹ showed a linear relation between Pbro₂ and CBF with changes in end-tidal CO₂ and further confirmed a linear relation between Pbro₂ and end-tidal CO₂ levels over ranges between 20 and 60 mm Hg. Various other experimental studies^102103104 have shown a decrease in Pbro₂ following hyperventilation. In a study on 16 swine, Manley et al¹⁰³ showed a 40% decrease in mean (± SD) Pbro₂ level from 36 ± 11 to 20 ± 9 mm Hg after hyperventilation. The deleterious effect of hyperventilation on Pbro₂ has been confirmed in many clinical studies.^{91105106107108109110111} In two studies,¹¹²¹¹³ however, the decrease in Pbro₂ was not significant, and some studies^108110111 have even reported an increase in Pbro₂ in some cases. These seemingly conflicting results may be explained by differences in pathophysiology between individual patients and would seem to favor the optimized hyperventilation approach advocated by Cruz et al.⁹⁸ In patients with raised ICP that is due mainly to cerebral vasodilation (hyperemia), hyperventilation may restore blood flow within damaged regions. This is also illustrated by different responses on SjO₂ monitoring compared to brain tissue oxygenation when Pbro₂ catheters are placed near the penumbra of focal lesions.¹¹³

The effect of hyperventilation on Pbro₂ seems to be time-dependent. Initially, van Santbrink et al¹⁰⁵ showed in 1996 that the tissue oxygen response to changes in Paco₂ was most marked on day 5 after trauma. In a follow-up study in 2000, Carmona Suazo et al¹⁰⁹ showed increasing tissue oxygen response to hyperventilation over time, and found a significant relation between increased tissue oxygen response on day 5 and poorer outcome. Similar observations have been reported by others.¹⁰⁷¹¹⁴ The observation that the deleterious effect of hyperventilation on Pbro₂ increases over time is intriguing. Until now, the greatest clinical concern for the risk of ischemia following hyperventilation has been within the first 24 h after injury as low CBF frequently occurs in this time period. It may, however, be argued that within this time frame of 24 h a general state of vascular narrowing exists and that further effects of hyperventilation may not have serious adverse consequences. The increasing tissue oxygen response over time may indicate an increased risk of ischemia, particularly at later time points. However, further research is necessary to confirm these results.

Information on the metabolic status of the brain can be obtained from chemical monitoring in the jugular venous blood, from microdialysis studies, from PET scan studies, or from MRI spectroscopy. After severe head injury, elevated levels of lactate in the CSF have been frequently shown.^{115116117118119}

Based on the results of lactate determinations in jugular venous blood, various authors^116120121 have shown the increased cerebral formation of lactate. In the study by Robertson et al,¹²⁰ lactate levels increased in proportion to the severity of cerebral trauma experienced during the first 2 days after injury. Murr et al,¹²² in a study of 21 patients with severe TBI, showed that in patients with intracranial hypertension the cerebral lactate level difference remained significantly increased from the first to the fifth day after injury, but normalized over this period in the group with normal or minimally elevated ICP values. Averaged over the short-term course, patients with increased ICP had significantly higher mean lactate level differences, and a significant correlation of increased mean cerebral lactate difference to poor outcome was noted.

Cerebral microdialysis is a relatively new technique for measuring metabolic parameters in the extracellular fluid and is being increasingly used in the monitoring of TBI patients, particularly in the research setting. Artificial CSF is injected into, and recovered from, a probe inserted in the cerebral cortex. The fluid equilibrates with the extracellular concentration of various metabolites, depending on the permeability of the microdialysis membrane, the length of the probe, and the velocity of injection. Microdialysis allows the long-term measurement of extracellular fluid energy-related metabolites (eg, glucose, lactate, and pyruvate) and amino acids in the cerebral cortex.¹²³ Goodman et al⁴⁵ measured lactate and glucose levels in 126 patients with head injuries. They found an initial increase in lactate level, perhaps indicating the presence of compensated hyperglycolysis, which gradually decreased during the first 24 to 48 h. Correlations have been demonstrated between low brain tissue oxygen levels and increased lactate levels.¹²⁴¹²⁵ Reinert et al¹²⁶ further demonstrated that increases in potassium levels were correlated with lactate accumulation, and were associated with increased ICP and poorer outcome. In contrast to other studies,¹²⁶ however, some episodes of high lactate levels were not associated with low brain tissue oxygen levels. The specific effect of hyperventilation on extracellular concentrations of glutamate, lactate, pyruvate, and local CBF in patients with TBIs were reported by Marion et al.¹²⁷ Hyperventilation studies, lowering arterial Pco₂ by 8 to 12 mm Hg, were conducted 24 to 36 h after injury and again at 3 to 4 days after injury. At 24 to 36 h after TBI, hyperventilation led to a significant increase in lactate levels and in the lactate/pyruvate ratio. At 3 to 4 days after TBI, hyperventilation also led to a significant increase in lactate levels, but the differences in the lactate/pyruvate ratio were not significant. The authors concluded that hyperventilation-induced changes are more pronounced during the first 24 to 36 h after TBI than at 3 to 4 days after TBI.

The distribution and intensity of the uptake of positron-emitting radiotracers in the tissue is an indicator of metabolism. PET scanning is a technique that allows the precise measurement of biomolecules such as glucose or oxygen in a living organ, such as the brain. A short-lived radioisotope is synthetically bound to the molecule of interest to form a positron-emitting radiotracer, which can be detected and quantitatively measured by PET scanning.¹²⁸

The effect of hyperventilation on the cerebral oxygen metabolism has been studied by Diringer et al¹²⁹ with PET scan studies. Nine patients with severe TBIs were moderately hyperventilated, and four more patients were intensely hyperventilated to a mean Paco₂ of 25 ± 2 mm Hg. Although this study demonstrated a significant decrease in CBF and an increase in oxygen extraction fraction following hyperventilation, CMRO₂ remained unchanged. It was concluded that brief hyperventilation may produce large reductions in CBF but does not lead to energy failure, and the authors considered that the observed reductions in CBF are therefore unlikely to cause further brain injury. Some serious limitations of this study were that only a few patients were studied, and that the duration of hyperventilation was relatively short and was maintained only for the duration of the actual PET scan study.

Coles et al⁹⁴ have shown in studies by conducted by PET scanning (see next section) that hyperventilation increases the volume of severely hypoperfused tissue within the injured brain, despite improvements in CPP and ICP. The same group has suggested more recently¹³⁰ that the injured brain may be less capable of increasing oxygen extraction in response to hypoperfusion, so that shortly after injury the brain could be more vulnerable to the CBF reduction induced by hyperventilation.

Despite the wide use of hyperventilation in the treatment of raised ICP after TBI and the large body of evidence indicating the possible deleterious effects of hyperventilation on CBF levels, oxygenation, and metabolism, only one prospective randomized clinical trial has been reported concerning the effect of hyperventilation on clinical outcome. Muizelaar et al¹³¹ compared the outcomes of patients who were hyperventilated to a Paco₂ of 25 mm Hg for 5 days to patients in whom the Paco₂ was kept at 35 mm Hg. At both 3 and 6 months after injury, patients with an initial Glasgow coma scale motor score of 4 or 5 had a significantly better outcome when they were not hyperventilated. This study formed the basis for the recommendation at the level of a standard (class I evidence) in the guidelines for the management of TBI stating the following: “… in the absence of increased ICP, prolonged hyperventilation therapy (Paco₂ ≤ 25 mm Hg) should be avoided.” In addition, the guidelines state that “the use of prophylactic hyperventilation (Paco₂ < 35 mm Hg) should be avoided during the first 24 h after severe TBI because it can compromise cerebral perfusion during a time when CBF is reduced.” However, at the level of an option, it is recognized that hyperventilation therapy may be necessary for brief periods when there has been acute neurologic deterioration or for longer periods if intracranial hypertension is refractory to other therapy.

We consider the class I evidence underlying the standard of these guidelines debatable and open to criticism. First, the control group was in fact mildly hyperventilated with a Paco₂ of 31 to 32 mm Hg. Second, the subgroup of patients with a Glasgow coma score motor score of 4 to 5 was not prespecified, and numbers were small (control group, 21 patients; hyperventilation group, 17 patients). Third, the study was confined to patients without raised ICP. Fourth, the best outcome was achieved by a third group of TBI patients, included in the study but neglected in further discussions, who had been hyperventilated and had received tromethamine (TRAM).

The use of hyperventilation in the treatment of patients with TBI remains controversial. Studies reporting beneficial and potentially adverse effects of hyperventilation on cerebral parameters are summarized in Table 2 . The controversy has been illustrated by various editorials and comments in the literature.^{111132133134135} The proponents of hyperventilation claim that it is effective in reducing ICP and that, despite a concomitant reduction in CBF levels, there is no evidence that this results in further metabolic derangement, and from this they conclude that the risk of ischemia is a nonissue. Adversaries of hyperventilation focus on the deleterious effects on CBF level, cerebral oxygenation, and neurochemical parameters obtained in microdialysis studies. Further, the lack of evidence of a beneficial effect on clinical outcome has been emphasized.¹³⁶

How can these two widely different points of view and approaches be reconciled? The answer to this question touches on the general discussion on standardized management vs more individually targeted approach. As Chesnut¹³² summarizes in the following way: “It is unclear why the various treatment modalities are felt to be mutually exclusive and all encompassing in the area of neurotrauma management.” Even the greatest proponent of hyperventilation⁸⁸ has emphasized the need for “optimized hyperventilation” aiming at correcting the mismatch between flow and oxygen metabolism, to which purpose multimodality monitoring including jugular oximetry is required. The adversaries of hyperventilation, who usually belong to the school advocating CPP therapy, will have to admit that the inadvertent use of therapy with vasopressors and hypervolemia also carries risks concerning a prolonged course of raised ICP, fluid overload, and an increased risk of ARDS.¹³⁷ We submit that both approaches may be appropriate under specifically defined circumstances, targeting the therapy to the individual requirements of patients. The standard, as contained in the international guidelines¹⁴ on hyperventilation, stating that prolonged hyperventilation in TBI patients without raised ICP should be avoided, must be put into the appropriate perspective. It may be argued that in patients without raised ICP there is no indication for hyperventilation. To date, there is no evidence in the literature unequivocally demonstrating that hyperventilation for the treatment of raised ICP in patients with TBI is related to poorer outcome, and there is also no evidence showing beneficial effects on overall outcome. When considering a therapy without proven clinical efficacy, a careful analysis of risks and benefits is required, considering the indication for and the duration of treatment. Risks concern systemic and cerebral complications. Systemic risks would appear to be greater, particularly in patients with preexistent cardiac disease and in patients with absolute or relative hypovolemia. In this regard, it should be noted that inadvertent hyperventilation is frequent in the prehospital setting at a time when optimal volume resuscitation has not yet been accomplished. Thomas et al¹³⁸ reported an incidence of low end-tidal CO₂ in 70% of patients with TBI who were undergoing helicopter transport to an urban level 1 trauma center.

Cerebral complications particularly relate to the risk of ischemia. Considering the risk-benefit relation, it would appear to be clear that the possibility of instituting hyperventilation therapy should be considered only in patients with raised ICP. No benefit may be expected in the absence of raised ICP. Theoretically, the benefit of hyperventilation may be more particularly expected in patients in whom raised ICP is considered mainly due to increased cerebral blood volume due to vasodilation. In the opinion of the authors, this would preferentially be the pediatric and young adult population. In clinical practice, however, it may be very difficult, if not impossible, to differentiate between the contribution of edema and cerebral blood volume to traumatic brain swelling following TBI, without facilities for PET scanning or MRI diffusion-weighted imaging. Marmarou et al¹³⁹ showed in a study of 31 patients with TBI that brain swelling due to cerebral blood volume averaged 2.94% compared with an average of 9.1% for brain swelling due to edema. In this group of patients, cerebral blood volume was increased in only five patients compared to the levels obtained in seven volunteers. However, no mention was made on the time period within which these studies were performed.

It has been argued that the main risk of ischemia due to hyperventilation will be present within the first 24 h after injury, as this is the period in which low CBF levels occur. We think that this generally accepted opinion may be challenged. If indeed this acute phase is characterized by a generalized state of vascular narrowing, the additional effect of hyperventilation may be expected to be low, and this has indeed been demonstrated in various studies.¹⁰⁹ Therefore, it may be tentatively concluded that the institution of hyperventilation therapy may be more appropriate during the relative hyperemic phases of days 2 and 3 after TBI. Nevertheless, the risk of ischemic complications cannot be excluded, and the careful monitoring of cerebral oxygenation is required.

Current evidence would favor a relatively short duration of hyperventilation therapy. The general consensus is not to hyperventilate TBI patients below a Paco₂ of 30 mm Hg. In exceptional circumstances, more intense hyperventilation may, however, be considered under careful monitoring.

Jugular oximetry permits the monitoring of global cerebral oxygen extraction, and local brain tissue Po₂ monitoring may yield additional information in the penumbra around contusions. McLaughlin and Marion⁸⁶ have reported increased vasoreactivity in the penumbra zones around the contusions up to nearly three times normal, suggesting the hypersensitivity of this region to hyperventilation therapy. Metabolic studies with MRI spectroscopy or PET scanning may be required before the possibility of local adverse effects of hyperventilation can be fully evaluated.

When considering the appropriate depth of hyperventilation, two specific circumstances need to be recognized. First, at higher altitudes normal Paco₂ levels may be well below the generally accepted levels of 35 to 45 mm Hg that were determined at sea level. A correction for the influence of altitude is therefore required. Second, the influence of temperature, particularly when hypothermia therapy is used, should be considered. In the laboratory, blood gas measurements are generally performed at 37°C, and the results are not corrected for body core temperature. The validity for performing temperature corrections has been argued.¹⁴⁰ In a more recent article, Vigue et al¹⁴¹ showed that the institution of hypothermia leads to a decrease in end-tidal CO₂ and Paco₂ due to a systemic and cerebral reduction of metabolism. In fact, these authors argued that the reduction of ICP following hypothermia may be fully explained by the concomitant decrease of Paco₂.

In conclusion, controversy exists, as exemplified in Table 2, and conflicting data may support a range of therapeutic options, from the enthusiastic overuse of hyperventilation to the avoidance of hyperventilation. It is our opinion that the careful use of hypocapnia for the short-term control of raised ICP remains a useful therapeutic tool. Multimodality monitoring is required in order to safely target hyperventilation therapy to specific patients who may benefit from it.