POD is defined as a "fluctuating disturbance in attention that represents an acute change from baseline, accompanied by disturbed cognition or perception, and not due to a pre-existing neurocognitive disorder or occurring in the context of a severely reduced arousal level". Depending on the type of surgery and the studied population, it can occur in 20 to 45% of the older patients. POD is a burden to the health care providers. Indeed, it is strongly associated with increased morbidity and mortality. The pathophysiology of POD is multifactorial and not yet completely elucidated. The aging brain is more vulnerable to the development of POD. However, more than the chronological age, the patient's overall vulnerability and their preexisting cognitive status are indicators of their ability to cope with these perioperative stressors. Indeed, the patient's cognitive status is a leading cause of POD and models predicting POD show poor accuracy because they do not take into account the patient's preoperative cognitive status. Preoperative neurocognitive assessment could be performed but these tests are time-consuming and subject to various influencing factors. Hence, objective tools are required to distinguish patients with preoperative cognitive impairment.
First hypothesis: The presence of a specific intraoperative EEG signal pattern may provide a tool for such identification of patients with underlying preoperative cognitive frailty. Commonly used anesthestic agents for induction and maintaining general anesthesia (e.g. sevoflurane or propofol) provide a typical electroencephalographic pattern with slow/delta and alpha oscillations, predominantly in the frontal cerebral cortex. More specifically, alpha oscillations actually originate from parieto-occipital sites in awake patients and migrate towards frontal regions after the induction of anesthesia. This phenomenon is called the "anteriorization" of the alpha frequency band. Besides, amongst all EEG frequencies, the contribution of alpha oscillations to the global tracing evolve throughout adulthood : the alpha power tends to decrease with age, and this decrease is more pronounced in the presence of underlying cognitive disorder (e.g. mild cognitive impairment, Alzheimer's disease). More importantly, it has been demonstrated that a lower frontal alpha band anteriorization during general anesthesia is associated with lower preoperative cognitive scores. Moreover, these patients might be at higher risk of intraoperative EEG suppression in case of an overdose of anesthetics or, even often, despite a lower dose of anesthetics. In this regard, the presence and the power of frontal alpha oscillations under general anesthesia may be indicators of the patient's preoperative cognitive status and may therefore predict the risk of developing POD.
Second hypothesis: Genetic studies have demonstrated a correlation between specific genotypes and the risk of cognitive decline. APOEe4 genotype is a known risk factor for Alzheimer's disease (AD), and has been shown to be also a risk factor of POD. However, APOEe4 allele is neither necessary nor exclusive to develop AD, and this may also hold true for POD, as this late hypothesis has been rejected in other previous studies. Otherwise, few studies have looked at some EEG particularities according to APOEe4 genotyping, in AD patients and control subjects. Unfortunately, their results regarding the presence of the e4 allele and associated EEG abnormalities are conflicting. To date, no study has related APOE genotyping and intraoperative EEG patterns under general anesthesia.
Third hypothesis: In addition to perioperative episodes of cerebral hypoxia and/or hypoperfusion and neuroinflammation, pathophysiological mechanisms of POD also include a potential direct insult to the brain, induced by both anesthesia and surgery. Yet, the ideal biomarker, highly sensitive for brain injury, as well as highly specific for neuronal tissue remains to be identified. Indeed, the release of such proteins after a neuronal injury can ensue from many levels and some of them have extracranial sources. These sources may therefore influence the observed results. As a conseuquence, in clinical practice, none of the currently evaluated neurobiomarkers (e.g. interleukins, Neuron Specific Enolase, S100 calcium-binding protein B) has emerged as a reliable diagnostic and/or prognostic tool for assessing postoperative neurological complications. Recently, much focus has been given to neurofilaments, as this group of proteins is part of the scaffolding of axons and is exclusively expressed in neuronal tissue. As a consequence, abnormally high levels of neurofilaments in extracellular fluids, such as cerebrospinal fluids (CSF) or serum, correspond specifically to neuronal cell damage, which represents a significant advantage compared to other biomarkers previously tested. Among the three subunits of neurofilament, neurofilament light (NfL) subunit has been shown to be promising. High levels of NfL have been found in a large range of neurodegenerative disorders, but also in acute events such as traumatic brain injury and stroke. Since it has been possible to measure NfL in serum, obviating the need for cerebrospinal fluid samples, their analysis in the perioperative period has been facilitated. Indeed, serum NfL levels have been recently investigated in the perioperative period in various surgical patient populations. These studies provide us information about the kinetics of perioperative NfL release but they show conflicting results regarding a potential correlation between high perioperative NfL levels and the occurrence of POD.
