Cerebral perfusion in sepsis

This article is one of ten reviews selected from the Yearbook of Intensive Care and Emergency Medicine 2010 (Springer Verlag) and co-published as a series in Critical Care. Other articles in the series can be found online at http://ccforum.com/series/yearbook. Further information about the Yearbook of Intensive Care and Emergency Medicine is available from http://www.springer.com/series/2855


Introduction
Sepsis, the host's reaction to infection, characteristically includes multi-organ dysfunction. Brain dysfunction is often one of the fi rst clinical symptoms in sepsis and may manifest as sepsis-associated delirium in up to 70% of patients [1,2], less often as focal defi cits or seizures [3]. As severely reduced global perfusion leading to hypotension, maldistribution of regional blood fl ow, and tissue hypoperfusion is a key feature of severe sepsis and septic shock, the question whether there is a link between cerebral perfusion and brain dysfunction in sepsis is obvious. However, clinical and experimental data on cerebral perfusion in sepsis are often inconsistent and most reports only include small numbers of animals or patients. We summarize the current literature on the eff ects of the infl ammatory response on cerebral per fusion and review the eff ects of altered cerebral perfusion on brain function in sepsis.

Sepsis and the brain
In sepsis, the brain may be aff ected by many systemic disturbances, such as hypotension, hypoxemia, hyperglycemia, hypoglycemia, and organ dysfunction (e.g., increased levels of ammonia in liver dysfunction or urea in acute kidney injury). Direct brain pathologies, such as ischemic brain lesions, cerebral micro-and macrohemorrhage, microthrombi, microabscesses, and multifocal necrotizing leukencephalopathy, have also been described in histopathologic examinations [4,5]. However, in addition to these metabolic and `mechanical' eff ects on the brain, infl ammation by itself causes profound alterations in cerebral homeostasis in sepsis.

Infl ammation and the brain
Sepsis at the outset causes a hyperinfl ammatory reaction, followed by a counteractive anti-infl ammatory reaction. Pro-and anti-infl ammatory cytokines are initially upregu lated. Despite its anatomical sequestration from the immune system by the blood-brain barrier, the lack of a lymphatic system, and a low expression of histocompatibility complex antigens, the brain is not isolated from the infl ammatory processes occurring elsewhere in the body. Th e circumventricular organs lack a bloodbrain barrier, and through these specifi c brain regions blood-borne cytokines enter the brain [5,6]. Th e circumventricular organs are composed of specialized tissue and are located in the midline ventricular system. Th ey consist of the organum vas culosum, the pineal body, the subcommissural organ, and the subfornical organ. Th ey also express components of the immune system (Toll-like receptors [TLR]), and receptors for cytokines such as interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α).
A further mechanism by which the brain can detect systemic infl ammation is through aff erent vagal fi bers ending in the nucleus tractus solitarius, which senses visceral infl ammation through its axonal cytokine receptors. In response to the detection of systemic infl ammation, behavioral, neuroendocrine, and autonomic responses are generated including expression of immune receptors and cytokines, inducible nitric oxide synthase (iNOS), and prostaglandins leading to oxidative stress, mitochondrial dysfunction, and apoptosis [5,7,8].

Eff ects of sepsis on the blood-brain barrier and the vascular endothelium
Th e blood-brain barrier, established by the tight junctions of the endothelial cells in interaction with astrocytic foot processes and pericytes, is responsible for a tightly regulated microenvironment in the brain. It prevents circulating noxious substances from entering into the brain and regulates brain capillary blood fl ow [1]. In sepsis, cerebral endothelial cells are activated by lipopolysaccharide (LPS) and pro-infl ammatory cytokines, including bradykinin, IL-1β, and TNF-α; TNF-α also activates iNOS [9]. Th ese changes in the cerebral microcirculation are associated with the upregulation of mRNA for local production of IL-1β, TNF-α, IL-6, and NO by induction of iNOS. Furthermore, leukocytes stick to the wall of blood vessels and enter the brain, mediated by adhesion molecules. Th e expression of one such adhesion molecule, the intercellular adhesion molecule (ICAM), is increased in septic rats [10]. Th ese local factors can promote endothelial dysfunction and result in blood-brain barrier breakdown leading to an increased permeability of the blood-brain barrier and to perivascular edema, as has been demonstrated in several animal models of sepsis [11][12][13]. Th e former facilitates the passage of neurotoxic factors, while the latter impairs the passage of oxygen, nutrients, and metabolites. Th e increased diapedesis of leukocytes and the perivascular edema decrease microcirculatory blood fl ow in the brain capillaries. Further evidence for an alteration in the blood-brain barrier comes from work by Alexander and colleagues [14]. In an animal model, these authors demonstrated that endotoxemia-triggered infl ammation in the brain led to an alteration in the blood-brain barrier, including an upregulation of aquaporin 4 and associated brain edema. Th is sequence of events appeared to be mediated by TNF-α signaling through the TNF receptor 1 [14].
In a recent magnetic resonance imaging (MRI) study in nine humans with septic shock and brain dysfunction, sepsis-induced lesions could be documented in the white matter suggesting blood-brain barrier breakdown [15]. However, in a pathologic study no evidence of cerebral edema was reported in 23 patients who died of septic shock [4].
NO is produced by the endothelium and plays an important role in the regulation of vascular tone; its increased release may be responsible for the vasodilatation and hypotension in sepsis [16]. iNOS is activated by endotoxins and cytokines leading to local and general vasodilatation [8,17,18]. NO is also considered a potent cerebral vasodilator [19]. Th us, NO may play an important role, not only in mediating systemic vascular resistance, hypotension, and cardiac depression, but also in cerebral vasodilatation during sepsis. However, in an ovine model of hypotensive-hyperdynamic sepsis, Booke and colleagues [20] demonstrated that inhibition of NOS did not alter cerebral blood fl ow (CBF) and postulated that CBF is regulated by mechanisms other than NO during sepsis. Nonetheless, in situations of ischemia and reperfusion the presence of great amounts of NO can cause an increased production of reactive oxygen species (ROS), like peroxynitrite, responsible for the destruction of membranes in cells and mitochondria.
Finally, another mechanism by which the brain is aff ected in sepsis is the generation of ROS by activated leukocytes. Exposed to these radicals, erythrocyte cell membranes become less deformable and may be unable to enter the brain microcirculation, thus aggravating the cerebral hypoperfusion seen in sepsis [21,22]. Th e brain itself with its high oxygen consumption and low antioxidant defense is susceptible to damage by ROS. Generation of ROS may alter oxidative phosphorylation and cytochrome activity in the mitochondria and impair cerebral energy production.

Cerebral perfusion pressure
Mean arterial pressure (MAP) is notoriously low in severe sepsis and septic shock. Accordingly, cerebral perfusion pressure (CPP) is low. Moreover, in view of the possible presence of brain edema, the infl uence of intracranial pressure (ICP) on CPP must be considered. Pfi ster et al. [23] measured ICP non-invasively in 16 patients with sepsis and reported moderate elevations of ICP > 15 mmHg in 47% of patients; an increase > 20 mmHg was not observed. CPP < 50 mmHg was found in 20% of their patients. Assuming that cerebrovascular pressure autoregulation is intact and the plateau of the autoregulatory curve is not shifted, their results suggest that CPP in the majority of the patients they investigated was likely to remain in the lower range of the autoregulatory plateau. However, this interpretation is partially in contrast to measurements of CBF in patients with sepsis. Bowton et al. [21] demonstrated that CBF was reduced in patients with sepsis independent from changes in blood pressure or cardiac output. Th ese authors used the 133 Xe clearance technique to measure CBF in nine septic patients. Similarly, Maekawa et al. [22] found signifi cantly lower CBF in six patients with sepsisassociated delirium than in awake controls. In an experimental model of human endotoxemia, Moller and colleagues [24] reported a reduction in CBF after an intra venous bolus of endotoxin in healthy volunteers. However, the authors assumed that CO 2 reactivity was intact in their subjects and explained this CBF reduction to hypocapnia occurring because of general symptoms of malaise, although they did not measure CO 2 reactivity in their subjects.

Regulation of cerebral perfusion CO 2 -reactivity
Using transcranial Doppler (TCD) and arterial partial pressure of CO 2 (PaCO 2 ) levels between 3.0 and 7.0 kPa, Matta and Stow [25] found relative CO 2 -reactivity to be within normal limits in ten patients with sepsis. Th eir patients were in the early stages of sepsis (< 24 h after admission to ICU), were all mechanically ventilated, and received infusions of midazolam and fentanyl. Absolute CO 2 -reactivity was lower than had been reported in subjects who were awake but consistent with values obtained during sedation and anesthesia. Similarly, Th ees and colleagues [26] reported a normal response to a decrease in PaCO 2 in ten patients with sepsis using TCD and cardiac output measurement by thermal dilution. Th eir patients were all mechanically ventilated, and sepsis had been established for > 48 h. Bowton and colleagues [21] also reported normal specifi c reactivity of the cerebral vasculature to changes in CO 2 in nine septic patients. However, Terborg and colleagues [27] reported impaired CO 2 -reactivity in septic patients, independent of changes in MAP. Th ey used TCD and near-infrared spectroscopy (NIRS) to assess CO 2 -induced vasomotor reactivity by inducing hypercapnia through reductions in the ventilatory minute volume in eight mechanically ventilated septic patients. It is important to note that all their patients suff ered from a neurological or neurosurgical illness, which may have aff ected the results. Similarly, Bowie and colleagues [28] observed significantly impaired cerebral CO 2 -reactivity in septic patients in a study of 12 sedated and ventilated patients who had sepsis for > 24 h using TCD at normocapnia, hypocapnia, and hypercapnia. Th e small sample sizes, diff erences in timing of the measurements of CO 2 -reactivity and in the severity of illness between groups, which is refl ected by the signifi cant diff erences in mortality as well as in some of the drugs used in the management of these patients, may be responsible for the confl icting fi ndings.

Cerebrovascular pressure autoregulation
Only a few studies have addressed the eff ects of sepsis on cerebral autoregulation. Matta and Stow [25] reported intact pressure autoregulation in ten mechanically ventilated patients with sepsis (not in septic shock) using a phenylephrine infusion to increase MAP by 20 mmHg and calculated an index of autoregulation by dividing the percentage change in estimated cerebral vascular resistance by the percentage change in MAP. Conversely, Smith and colleagues [29] reported loss of cerebrovascular autoregulation in 15 patients with septic shock as they were able to demonstrate a correlation between cardiac index and CBF using TCD and cardiac output measured by thermodilution. In a recent study, Pfi ster and colleagues [30,31] found disturbed cerebral autoregulation in patients with sepsis-associated deliriumbut not in patients with `plain' sepsis -using TCD and NIRS. Th is suggests that cerebral autoregulation is possibly intact in patients with sepsis but disturbed with more severe disease or complications manifesting as septic shock or sepsis-associated delirium.

Perfusion and brain dysfunction Cerebral ischemia
Cerebral ischemia is a reality in sepsis: In a post-mortem analysis of the brain of patients who died from sepsis, multiple small ischemic lesions could be identifi ed in diff erent areas of the brain [4]. Possible explanations are the hypotension seen in sepsis, especially when concurrent with preexisting cerebrovascular disease or autoregulatory failure. Th rombotic mechanisms due to a high hematocrit and increased viscosity of blood in sepsis may lead to watershed infarction as has been described in a septic patient with prolonged hypotension [3].

Cerebral perfusion and sepsis-associated delirium
Sepsis-associated delirium is a common organ dysfunction in sepsis and may actually occur before failure of other organs. It can be found in up to 70% of patients with the sepsis syndrome and is correlated with the severity of sepsis [32][33][34]. Depending on the criteria used for diagnosis, it may be detected in almost all patients with sepsis [32,35]. Sepsis-associated delirium has been reported as an independent predictor of death [36]; however it may only refl ect the severity of illness and may not be the cause of death itself. Sepsis-associated delirium presents as an alteration of the mental state and may range from lethargy or mild disorientation to obtundation and coma. Th e pathophysiology of sepsisassociated delirium is incompletely understood and is probably multifactorial. Mechanisms postulated to cause sepsis-associated delirium include brain activation by infl ammatory mediators via the vagus nerve and the circumventricular organs, which interfere with the liberation of neurotransmitters and neurohormones. Oxidative stress and formation of ROS compromising cell function and endothelial activation resulting in disruption of the blood-brain-barrier are other mechanisms proposed to play a role in development of sepsisassociated delirium [5]. However, cerebrovascular autoregulation may also play a role in sepsis-associated delirium [25,27,29,30,36]. Pfi ster and colleagues [30] reported less effi cient autoregulation in patients with sepsis-associated delirium compared to patients without sepsis-associated delirium. However, in the same patients, cerebral oxygenation measured by NIRS did not diff er between patients with and without sepsisassociated delirium. Th us, reduced cerebral blood fl ow and disturbed cerebrovascular autoregulation mayamong others -be important precipitating factors for sepsis-associated delirium [2,30]. Alternatively, it could also be argued that disturbed autoregulation is merely a refl ection of a more severe infl ammatory stimulus that is associated with a more profound dysfunction of the blood-brain barrier and hence endothelial/autoregulatory dysfunction.

Eff ects of catecholamines on cerebral perfusion in patients with sepsis
Data on the cerebrovascular eff ects of catecholamines in sepsis are scarce. Th e blood-brain barrier prevents catechol amines from entering the brain as long as it is intact. Cerebral hemodynamics are not directly aff ected by norepinephrine and phenylephrine in anesthetized patients without cerebral pathology [37]. After head injury however, dopamine, norepinephrine and phenylephrine all seem to increase CBF with the eff ect of norepinephrine being more predictable than that of dopamine [38]. Th is is possibly due to the fact that in head injury there is also a disruption of the blood-brain barrier that allows, e.g., norepinephrine to access intracerebral β receptors leading to an increase in cerebral metabolism and, hence, CBF [39]. Accordingly, it could be speculated that in sepsis also the cerebral eff ects of vasopressors may be unpredictable depending on the degree of blood-brain barrier dysfunction.
A representation of documented and hypothetical factors infl uencing cerebral perfusion in sepsis is shown in Figure 1.

Conclusion
Th e infl ammatory response observed in sepsis triggers profound changes in the brain. Blood-brain barrier permeability is increased, and substantial changes in regulation of CBF and cerebral perfusion may occur.
Hypoperfusion due to severe hemodynamic instability will obviously lead to ischemic brain injury. Furthermore, the changes in pressure autoregulation may result in an increased vulnerability of the brain to hypoperfusion. However, this does not explain the full range of brain dysfunction found in septic patients. So far it has not been possible to establish a clear link between cerebral perfusion and sepsis-associated delirium. It is conceivable that the eff ects of the infl ammatory response on the brain per se are the key events leading to sepsis-associated delirium, and that the observed changes in CBF regulation are rather a consequence of infl ammation than a cause of sepsis-associated delirium.

Figure 1. Synopsis of documented and hypothetical factors infl uencing cerebral perfusion in sepsis.
Some of the factors (e.g., nitric oxide [NO]) infl uence cerebral perfusion at diff erent levels of the brain circulation. It could be speculated that the eff ect of vasopressors may be unpredictable depending on the degree of blood-brain barrier dysfunction. MAP: mean arterial pressure; CPP: cerebral perfusion pressure; ICP: intracranial pressure.