Newly identified precipitating factors in mechanical ventilation-induced brain damage: implications for treating ICU delirium

Abstract

Delirium is 1.5 to 4.1 times as likely in intensive care unit patients when they are mechanically ventilated. While progress in treatment has occurred, delirium is still a major problem in mechanically ventilated patients. Based on studies of a murine mechanical ventilation model, we summarize evidence here for a novel mechanism by which such ventilation can quickly initiate brain damage likely to cause cognitive deficits expressed as delirium. That mechanism consists of aberrant vagal sensory input driving sustained dopamine D2 receptor (D2R) signaling in the hippocampal formation, which induces apoptosis in that brain area within 90 min without causing hypoxia, oxidative stress, or inflammatory responses. This argues for minimizing the duration and tidal volumes of mechanical ventilation and for more effectively reducing sustained D2R signaling than achieved with haloperidol alone. The latter might be accomplished by reducing D2R cell surface expression and D2R-mediated Akt inhibition by elevating protein expression of dysbindin-1C.

Keywords::

• apoptosis
• cognition
• critical care
• dopamine
• dopamine D2 receptor
• dysbindin
• haloperidol
• hippocampus
• mechanical ventilation
• vagus nerve

Twenty-five percent of intensive care unit (ICU) patients receive mechanical ventilation for more than 12 h [1]. Of these, 74–83% develop delirium [2–4] compared with 20–48% of nonventilated ICU patients [5,6]. In other words, the prevalence of delirium is 1.5–4.15 times higher in ventilated patients. This poses serious risks because the occurrence and duration of delirium in such patients are associated not only with longer hospitalization and increased mortality [7] but also with forebrain damage [8–10] and the related constellation of long-term cognitive deficits [4,11,12], anxiety and depression [12], and daily living disabilities [12,13]. Along with pain (P) and agitation (A), then, delirium (D) belongs to the PAD triad of immediate medical concerns in treating ICU patients [14,15].

Over the last decade, progress has been made in treating ICU delirium experienced by mechanically ventilated patients. This is has been achieved by identifying modifiable risk factors and minimizing them safely. As summarized in recent reviews and reports [16–19], the incidence and/or duration of delirium in mechanically ventilated ICU patients can be reduced in multiple ways, specifically (a) managing agitation with light sedation using dexmedetomidine (a centrally acting α2 adrenergic agonist), not benzodiazepines (GABA_A receptor modulators) apart from some exceptional circumstances [19]; (b) daily interruption of sedative administration coupled with daily spontaneous breathing trials; (c) managing pain with analgesic, not sedative, doses of opioids; (d) early ICU physical rehabilitation (early mobilization); and (e) aiding restorative, rapid eye movement sleep nonpharmacologically (e.g., with ear plugs). Several of these recommendations have been effectively combined in what is called the ABCDE strategy (Awakening & Breathing Coordination, Delirium monitoring/management and Early exercise/mobility bundle) [16–18].

Nevertheless, the occurrence and duration of delirium in mechanically ventilated ICU patients remains high [18]. While pharmacological treatments have shown promise in ICU cases following elective surgery [17,19], none have proven effective in preventing delirium in critically ill patients [14,16,19,20]. With respect to delirium reduction, only dexmedetomidine has proven even moderately effective in these patients [14,16,19]. As a result, the American College of Critical Care Medicine does not currently recommend any pharmacological treatments for delirium other than dexmedetomidine [14].

We believe discovery of more potent treatments for delirium in mechanically ventilated ICU patients requires further study of its precipitating biological factors. Consistent with studies on delirium in general, three such factors are commonly noted, though they are based only on correlational studies: (a) hypoxic ischemia in multiple brain areas, most consistently the hippocampus [8]; (b) systemic inflammation elevating brain levels of proinflammatory cytokines [21,22] and (c) altered neurotransmission in the brain, most consistently reduced cholinergic transmission and increased dopaminergic transmission [7,23]. The latter two factors may be related given that acetylcholine reduces [24] while dopamine increases [25] inflammation-induced cytokine secretion by macrophages. It has been argued, in fact, that conditions promoting neural degeneration such as occur in ICU delirium patients [8–10] may prime microglia (brain macrophages) to amplify CNS responses to systemic inflammation [26].

To test proposed precipitating factors in mechanical ventilation-induced delirium and to identify new ones, we developed an experimental model of ventilation-induced brain injury in mice [27]. The relevance of this model to delirium is based on evidence that the duration of delirium in mechanically ventilated ICU patients is positively associated with the degree of brain volume loss [9] and brain white matter abnormalities [10] in those patients. We found that mechanical ventilation of mice for just 90 min raised levels of three apoptosis biomarkers (cleaved caspase 7, cleaved poly (ADP-ribose) polymerase 1 [PARP-1] and intact caspase 9) in the hippocampal formation (HF = hippocampus [CA1–3] + dentate gyrus + subiculum), a brain area playing an important role in diverse cognitive functions [28]. Significant elevations in cleaved caspase-7 and PARP-1 occurred with both low- and high-pressure mechanical ventilation delivering tidal volumes of air at 8 and 15 ml/kg body weight, respectively. Significantly elevated casapase-9 occurred only with high-pressure ventilation.

Such rapid apoptosis may seem surprising since many apoptotic stimuli are reported to cause cell changes only after a delay of at least 6 h. Earlier apoptotic effects are known, however. For example, elevated apoptosis biomarkers are found in neurons 30 min to 2 h after onset of toxic, ischemic or traumatic conditions [29–33].

The early apoptotic response to mechanical ventilation in the HF could not be attributed to classic apoptotic triggers because it was not accompanied by elevated biomarkers for hypoxia, oxidative stress or proinflammatory cytokines (i.e., interleukins 1β and 6). While these factors may contribute later to brain damage, they clearly did not initiate mechanical ventilation-induced apoptosis. The early apoptotic trigger proved instead to be novel, specifically dopamine-induced inhibition of the antiapoptotic Akt → GSK-3β pathway [34] activated via dopamine D2 receptors (D2R) [27] at both low- and high-pressure ventilation. Prolonged activation of D2R causes association of Akt with β-arrestin 2 and protein phosphatase 2A [35]. Formation of this protein complex results in dephosphorylation of Akt at its Thr308 activation site [35], which prevents Akt inhibition of several proteins enabling apoptosis, including GSK-3 [34]. Since gene expression levels of tyrosine hydroxylase, a catecholamine synthesizing enzyme, were elevated in the HF of mice ventilated at high (but not low) pressure, hyperdopaminergia in the HF may contribute to apoptosis seen in these mice [27]. This may help explain why D2R antagonists (the antipsychotics haloperidol, olanzapine and risperidone) show promise in reducing the duration and/or severity of delirium in mechanically ventilated ICU patients following elective surgery [19,36], although not in those critically ill [16,20].

Our study [27] also addressed how mechanical ventilation can cause dopamine-related HF apoptosis. Passive lung inflation activates slowly adapting receptors of pulmonary sensory neurons with cell bodies in the nodose ganglion and axons entering the brain via the cervical vagus nerve to innervate so-called pump cells (P cells) in the nucleus of the solitary tract (NTS) [37–39]. P-cell axons terminate in the caudal NTS and in several output targets of the NTS, mainly the caudal ventrolateral medulla (also known as the paragigantocellular nucleus [40–42]), the retrotrapezoid nucleus in the rostral medulla [43–45] and the parabrachial-Köllliker-Fuse nuclear complex (PB-KFC) in the pons [46–48]. Via the PB-KFC [45,48,49] and/or the locus ceruleus [50–52], these three cell groups (caudal ventrolateral medulla, retrotrapezoid nucleus and PB-KFC) innervate the ventral tegmental area (VTA) of the midbrain.

The VTA, along with the adjacent substantia nigra, pars compacta, is the source of HF dopaminergic input [53], which normally facilitates cognitive processes in the HF [54]. Aberrant VTA input due to chronic mechanical ventilation, however, may account for the elevated tyrosine hydroxylase gene expression we found in the HF of ventilated mice [27] because chronic vagal stimulation elevates dopamine in VTA targets studied to date (i.e., prefrontal cortex and nucleus accumbens) [55]. Via several pathways that converge on the VTA, then, aberrant vagal stimulation may cause a hyperdopaminergic state in the HF. This may be deleterious not only because it can promote apoptosis in that brain area as noted above, but also because dopamine via D2R suppresses CA1 responses to input from CA3 [56].

Consistent with this hypothesis, we found that HF apoptosis induced by mechanical ventilation within 90 min is mediated in large part by vagal stimulation and by D2R-mediated dopamine signaling in the HF [27]. Such apoptosis was absent in mice given either a bilateral cervical vagotomy or the D2R antagonist haloperidol (0.5 mg/kg) prior to ventilation. Activation levels of signaling molecules in the antiapototic Akt → GSK-3β pathway similarly remained normal in mice that were either vagotomized or haloperidol before ventilation. As the latter finding predicted, intracerebroventricular injection of the D2R antagonist raclopride (but not the D1R antagonist SCH-23390) before mechanical ventilation reduced apoptosis and actually increased the activation state of the Akt → GSK-3β pathway above normal levels.

Abnormal vagal activity and related elevation of D2R-mediated dopamine signaling in the HF thus emerge as likely precipitating factors in mechanical ventilation-induced brain damage. This has two clinical implications for treatment of mechanically ventilated ICU patients. First, to minimize apoptosis in the HF, disruption of which is associated with cognitive dysfunction [28] and delirium [8], both the tidal volume and duration of mechanical ventilation should be minimized. This adds urgency to earlier arguments for minimizing time spent continuously on such ventilation [16,18]. Second, to minimize dopamine-induced HF apoptosis, more potent reduction of D2R signaling is needed given the ineffectiveness of D2R antagonists on ICU delirium in the critically ill [16,20].

Expert commentary

Our work suggests that minimizing D2R-mediated apoptosis may be facilitated by augmenting D2R antagonist protocols with procedures to reduce cell surface levels of D2R. Compared with nonventilated ICU cases, we found that mechanically ventilated ICU cases had elevated levels of dysbindin-1 in HF neurons [27], especially in CA2 and CA3 containing the highest density of tyrosine hydroxylase-positive axon terminals in the HF [57]. Although the specific isoforms have not been identified, dysbindin-1 is known to reduce cell surface levels of neuronal D2R (but not D1R) [58,59] and to protect neurons against cell death [59,60]. The same apparent compensatory response was found in the HF of our mechanically ventilated mice, which showed elevated gene expression of dysbindin-1 isoform C (a pre- and postsynaptic protein [59,61]) within 90 min and elevated protein expression of that isoform within 330 min. Dopamine by itself elevates gene expression of dysbindin-1C in the HF, an effect blocked by haloperidol and thus presumably mediated via D2R [27]. Likewise, mechanical ventilation elevated dysbindin-1C protein in the HF, an effect blocked by haloperidol or by cervical vagotomy [27]. A postventilation rise in dysbindin-1C was clearly insufficient to prevent ventilation-induced HF apoptosis within 90 min, but may do so if dysbindin-1C levels were raised prior to mechanical ventilation. At present, however, dysbindin-1C remains a drug target still unexplored for clinical purposes.

What may prove most effective in minimizing delirium in mechanically ventilated ICU patients is a combination of the best practices now used, notably the ABCDE strategy [16], with more effective prophylactic reductions in D2R dopamine signaling than safely achieved with D2R antagonists alone and proinflammatory cytokines that may contribute to delirium after the initial apoptotic effects of mechanical ventilation. This combination of treatments may finally lead to reliable prevention or reduction in delirium experienced in this highly vulnerable population of patients.

Five-year view

Basic questions remain regarding mechanical ventilation-induced HF apoptosis and its relationship to cognitive deficits and delirium. In what way is vagal sensory input to the brain abnormal under mechanical ventilation? How does that cause elevated dopamine expression in VTA axons innervating the HF? Does the early HF apoptosis induced by mechanical ventilation actually result in cognitive deficits as expected from such brain damage? How long after the 90 min interval tested might brain inflammatory process contribute to brain damage and cognitive deficits? These questions could be answered in the near future.

Over the next five years, we anticipate progress in preventing delirium or limiting its duration in mechanically ventilated patients as refinements in the ABCDE strategy are coupled with more effective means of reducing D2R-mediated dopamine signaling. The latter may require, however, development of brain accessible drugs that safely elevate dysbindin-1C expression and/or diminish D2R-induced inhibition of Akt via β-arrestin 2 [30]. Parallel development of anti-inflammatory prophylaxis in the ICU will probably also occur in the same time frame given evidence that proinflammatory processes may well contribute to delirium in the critically ill [21,22,26]. As clinical strategies become more multifactorial, the pace of progress in treating delirium in mechanically ventilated ICU patients should continue to accelerate.

Key issues

• The percentage of intensive care unit (ICU) patients experiencing delirium is as much as four times greater in those who are mechanically ventilated.

• Delirium in ICU patients poses serious risks apart from increased mortality, notably brain damage and what are likely to be related, long-term impairments in cognition, mood and daily living activities.

• While progress has been made in its treatment, delirium remains common in mechanically ventilated patients, reflecting in part our still limited knowledge of its precipitating factors.

• We recently addressed this knowledge gap by studying mechanical ventilation experimentally in mice to determine how it can cause brain damage and thereby the cognitive deficits expressed as delirium.

• Just 90 min of mechanical ventilation in mice, especially at high tidal volumes, was sufficient to induce apoptosis in the hippocampal formation (HF), a brain area whose disruption is associated with cognitive impairment and delirium.

• The HF apoptosis quickly induced by mechanical ventilation is not associated with hypoxia, oxidative stress or inflammatory responses in the HF, but appears to be driven instead by aberrant vagal input to dopaminergic midbrain neurons that trigger proapoptotic signaling in the HF via dopamine D2 receptors.

• This hypothesis is supported by evidence that cervical vagotomy or D2R blockade by haloperidol prevents mechanical ventilation-induced apoptosis in the mouse HF.

• Such HF apoptosis is followed within 4 h by an increase in HF levels of the C isoform of a protein (dysbindin-1) known to reduce cell surface D2R in neurons. A similar compensatory increase in dysbindin-1 occurs in the HF of mechanically ventilated ICU patients.

• These findings provide an additional reason to minimize the tidal volume and duration of mechanical ventilation and suggest that the effectiveness of D2R antagonism in reducing delirium may be augmented by other means of reducing D2R signaling (e.g., raising levels of dysbindin-1C).

• The most effective means of reducing incidence and duration of delirium, however, will probably require reducing D2R-mediated dopamine signaling along with other factors (e.g., proinflammatory cytokines) promoting delirium most comprehensively addressed in the ABCDE treatment strategy.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.