Perioperative neurocognitive disorder (PND), which encompasses delirium and postoperative cognitive dysfunction, commonly occurs in the elderly after anesthesia and surgery. Previously, all forms of postoperative cognitive impairments were called “postoperative cognitive dysfunction,” but more recently, the use of “perioperative neurocognitive disorder” is recommended as an overarching term for cognitive impairment identified in the preoperative or postoperative period (Evered et al., 2018a). This change aligns PND with the phenotypically similar neurocognitive diagnoses listed in the Diagnostic and Statistical Manual of Mental Disorders, version 5, such as Alzheimer's disease (AD) (Evered et al., 2018a). Furthermore, PND has emerged as a significant global public health issue that leads to more extended in-hospital stays, higher hospitalization costs, and higher mortality rates (Yang and Terrando, 2019). The risk factors for PND are multifaceted, which might be related to anesthetic management, tissue damage, surgery duration, psychological stress, and genetic susceptibilities (Subramaniyan and Terrando, 2019; Eckenhoff et al., 2020). As the elderly population continues to increase, the number of cases of PND will continue to rise. The incrementally growing prevalence forces researchers to explore the mechanisms underlying the pathogenesis of PND and to seek optimal prevention and treatment measures (Evered and Silbert, 2018; Berger et al., 2019).

Preclinical and clinical studies support the notion that neuroinflammation plays a significant role in the pathogenesis of PND (Nathan, 2019; Yang et al., 2019, 2020). Trauma experienced during surgery triggers the release of endogenous factors known as damage-associated molecular patterns that can activate immune cells such as neutrophils and monocytes, promote pro-inflammatory cytokines, and subsequently cause systemic inflammation (Huber-Lang et al., 2018). Uncontrolled systemic inflammation is associated with a compromised blood-brain barrier (BBB). BBB impairment promotes immune cell movement and pro-inflammatory cytokines from the blood into the brain parenchyma, triggering neuroinflammation and ultimately leading to postoperative cognitive impairment (Abrahamov et al., 2017; Yang et al., 2017, 2020). Microglia, the resident immune cells of the central nervous system (CNS), perform “immune surveillance,” and survey their assigned brain regions (Kettenmann et al., 2011). In addition, with prolonged systemic inflammation, microglia start to develop an activated phenotype characterized by an increase in pro-inflammatory mediators, such as interferon-γ, interleukin (IL)-1β, tumor necrosis factor (TNF)-α, and reactive oxygen species (Liu L. R. t al., 2020). Pro-inflammatory factors released by microglia such as IL-1α and TNF contribute to the subsequent activation of astrocytes and further promote neuroinflammation (Liddelow et al., 2017; Liu L. R. t al., 2020). Complement system activation is another essential inflammatory response that is activated by surgically triggered damaged-associated molecular patterns. For example, the deposition of the C-reactive protein, a biomarker for delirium, can activate and regulate the classical complement pathway, thereby contributing to dysregulated inflammation. Blocking the complement cascade (e.g., complement C3 gene deficiency and selective complement inhibitor for complement C3) improves neuroinflammation and functional outcomes of PND (Alawieh et al., 2018) also suggests the contribution of neuroinflammation in PND.

Several studies have shown that the pathological mechanism of PND is similar to that of AD (Evered et al., 2016; Gerlach and Chaney, 2018). It was also demonstrated that a significant accumulation of amyloid-β (Aβ) and tau in the brain parenchyma after anesthesia and surgery (Terrando et al., 2011). Moreover, increased Aβ and tau in the cerebrospinal fluid (CSF) have been reported in patients after surgery (Evered et al., 2018b), which are risk factors for PND (Xie Z. et al., 2013; Evered et al., 2016; Cunningham et al., 2019). The accumulation of Aβ and tau induces neuroinflammation, leading to glial activation, pro-inflammatory factor release, and neuronal damage (Calsolaro and Edison, 2016).

The glymphatic system is a recently discovered waste removal system that utilizes a unique perivascular channel system formed by astrocytes (Iliff et al., 2012; Benveniste et al., 2017; Zhang C. et al., 2018). Although the glymphatic system contributes to the delivery of nutrients, specifically glucose, its most influential and recognized function is to clear extracellular metabolites and waste products from the parenchyma into the CSF (Abbott et al., 2018; Nedergaard, 2013; Simon and Iliff, 2016). Since its discovery, researchers have proposed that waste or protein aggregates induced by glymphatic dysfunction are associated with neurodegenerative diseases, including AD (Tarasoff-Conway et al., 2015; Harrison et al., 2020; Nedergaard and Goldman, 2020). Given the characteristics of the glymphatic system and the complexity of possible changes during the perioperative stage, we herein speculate a causal relationship between the glymphatic system and symptomatic progression in PND. Glymphatic dysfunction in the perioperative period, which leads to waste accumulation, could trigger or exacerbate neuroinflammation and eventually lead to cognitive dysfunction. Through this review, we discuss the function and driving mechanism of the glymphatic system, outline the current evidence to illustrate the impact of anesthesia and surgery on the glymphatic system, and emphasize the viewpoint that glymphatic dysfunction is involved in the pathogenesis of PND.

The glymphatic system is a perivascular network that subserves a pseudo-lymphatic function throughout the brain (Figure 1). Iliff et al. named this brain-wide fluid transport pathway the glial-associated lymphatic system or glymphatic system due to its dependence on glial water flux and the lymphatic function of the brain (Iliff et al., 2012). As the macroscopic waste clearance system of the CNS, the glymphatic system's fundamental role is to eliminate soluble proteins and metabolites, including Aβ and tau, based on the evidence that 40%−80% of neurotoxic compounds can be cleared from the CNS via this system (Iliff et al., 2012). The neurotoxic compound clearance process within the glymphatic system is described as a 3-step serial process (Benveniste et al., 2019b). First, CSF is continuously transported from the subarachnoid space and Virchow-Robin space to the peri-arterial spaces in a bulk-flow driven manner; subsequently, CSF is propelled from the peri-arterial compartment into the interstitial fluid (ISF) space. The convection and mixing of CSF and ISF are facilitated by aquaporin 4 (AQP4) in the dense and complex brain parenchyma. Ultimately, CSF-ISF fluid mixes with interstitial waste solutes and is subsequently transported to the perivenous space, from the meningeal lymphatic vessels to the lymphatic vessels and circulatory system (Plog and Nedergaard, 2018).

AQP4 is required to maintain the glymphatic function and is abundantly expressed in the end-feet of astrocytes that surround arteries and veins. It is also anchored to the astrocyte membrane by the carboxyl terminus of α-syntrophin. Syntrophin-dependent AQP4 mediates the bidirectional transport of water across the brain-blood interface (Amiry-Moghaddam et al., 2003). As a bidirectional channel, the effects of AQP4 on glymphatic function may be diverse and require further investigation. Although the role of AQP4 in promoting ISF clearance is debated, a large body of evidence suggests that it facilitates CSF movement from perivascular spaces into the interstitial space and activates flushing of ISF (Smith et al., 2017). Studies on different strains of AQP4-knockout mice have suggested that AQP4 is necessary for the rapid movement of CSF from the perivascular space into the ISF and through the brain (Mestre et al., 2018a). Moreover, AQP4 reduces the resistance to CSF movement from the periarterial space into the interstitium and subsequently from the interstitium into the perivenous space (Groothuis et al., 2007). The flow of CSF via the glymphatic pathway is an essential part of the CNS that eliminates metabolic waste products, such as Aβ and lactic acid, from deep inside the brain and delivers nutrients, such as glucose, or therapeutic drugs to the brain parenchyma. This CSF circulation differs from typical lymphatic vessels and acts as a pseudo-lymphatic drainage within the CNS. Therefore, more recently, alternative CSF circulation in the brain has been proposed as a glymphatic circulation (Benveniste et al., 2019b).

Currently, most animal experiments on glymphatic function are performed by applying anesthetic drugs to simulate the sleep state. The effect of anesthetics cannot be ignored when studying the effect of sleep on glymphatic function. Currently, limited research has been conducted to investigate the effects of anesthetics or sedatives on the rodent glymphatic system. A study demonstrated that inhaled isoflurane (2–2.5%) inhibited CSF circulation and waste clearance in the brain (Gakuba et al., 2018). This result indicates that inhaled isoflurane inhibited glymphatic function, especially at a high dose (3%) (Gakuba et al., 2018). This research also showed that ketamine, a common intravenous anesthetic, also inhibited glymphatic activity at an intraperitoneal dose of 150 mg·kg⁻¹ in mouse models (Gakuba et al., 2018). The inhibitory effect of ketamine on glymphatic function could be reversed using a combination of ketamine and 10 mg·kg⁻¹ of xylazine (Gakuba et al., 2018). However, Xie et al. previously found that ketamine (100 mg·kg⁻¹) combined with xylazine (20 mg·kg⁻¹) significantly increased the influx of the CSF tracer in all mice analyzed (Xie L. et al., 2013). The opposite effects of the two different anesthetic regimens require further study. One possible explanation is whether ketamine inhibits or increases lymphatic inflow, depending on the dose of ketamine and its combination with xylazine, an α₂-adrenergic agonist and an analog of clonidine used as a sedative and analgesic in animals. Xylazine increases glymphatic CSF influx and may remedy the impairment of ketamine use in the glymphatic system (Gakuba et al., 2018).

Dexmedetomidine (DEXM) is another α₂-adrenergic agonist widely used for anesthesia or sedation in clinics and has also been shown to enhance glymphatic function in animal research (Lilius et al., 2019a). Recently, Ozturk et al. reported that DEXM supplemented with low-dose isoflurane increased glymphatic transport compared to that of isoflurane only (Ozturk et al., 2021). DEXM has also been shown to increase the delivery of intrathecally administered drugs, such as oxycodone and naloxone, by increasing glymphatic flow (Lilius et al., 2019b).

Propofol is a widely used intravenous anesthetic in humans with anti-inflammatory (Ren et al., 2014) and neuroprotective properties (Miller et al., 2018). Recent studies have found that propofol can increase glymphatic activity (Gakuba et al., 2018). Pentobarbital has also been shown to significantly increase glymphatic function in animal models (Hablitz et al., 2019), although benzodiazepines have mainly been replaced by human anesthesia. Other anesthetics used in animal anesthesia such as α-chloralose and avertin inhibit CSF tracer influx in the glymphatic pathway (Hablitz et al., 2019). Therefore, glymphatic function is affected by the dose, type, and combination of anesthetics. Although the exact mechanisms underlying these preclinical experiments remain unclear, they may provide valuable information for studying the functional effects of the glymphatic system during clinical anesthesia (Table 1).

Interestingly, these animal studies on glymphatic dysfunction agree with those of clinical studies on PND. Anesthetics are a risk factor for postoperative cognitive decline or neuropathological changes (Schenning et al., 2016). It has been confirmed that inhaled anesthetics, such as isoflurane or sevoflurane, have more chance of causing postoperative cognitive dysfunction (Ologunde and Ma, 2011; Hu et al., 2014), whereas some intravenous anesthetics, such as DEXM and propofol, cause slight harm or even improve perioperative cognitive function (Miller et al., 2018). Anesthetics that have a significant inhibitory effect on the glymphatic system are more likely to cause perioperative cognitive impairment, whereas some anesthetics that enhance glymphatic function are less likely to cause cognitive impairment. The effects of ketamine on the glymphatic system and postoperative cognition have been controversial (Morrison et al., 2018). DEXM enhances glymphatic function and has also been shown to prevent postoperative delirium and cognitive dysfunction caused by inhaled anesthetics in elderly patients (Zhang et al., 2018). Therefore, anesthetics may affect postoperative brain function by affecting the glymphatic function. Currently, the mechanisms underlying the effects of anesthetics on glymphatic function remain unclear. Further clinical research is necessary to determine the optimal variety, dose, and combination of anesthetics and drugs to reduce glymphatic function impairment during anesthesia.

The mechanisms that regulate glymphatic function under normal physiological conditions include arterial pulsatility, respiration, and neural activity. However, these mechanisms may be disrupted during anesthesia, including general anesthesia and spinal anesthesia (Scott, 2018). During clinical anesthesia, the heart rate, blood pressure, and arterial pulsation are significantly changed, and spontaneous negative pressure respiration is replaced with positive pressure ventilation. The arousal mechanism is also significantly suppressed during general anesthesia (Schiff, 2020). Anesthesia-induced pathological processes, such as the breakdown of glymphatic CSF-ISF exchange, have been linked to disease initiation and progression. As a result, the impact of anesthesia on glymphatic function needs to be considered.

Persistent systemic inflammation caused by surgery can disrupt the integrity of the BBB (Vacas et al., 2014; Hughes et al., 2016; Bi et al., 2017; Wang et al., 2017; Ni et al., 2018; Verheggen et al., 2018). The BBB is a complex functional and anatomical system that prevents the entry of neurotoxic plasma elements, blood cells, and pathogens into the brain. It is comprised of endothelial cells, perivascular microglia, pericytes, neurons, and astrocytic end-feet (Sweeney et al., 2019). Tight junction proteins located between adjacent endothelial cells include claudins, tricellulin, occludin, and zonula occludens (Sweeney et al., 2019). Acute endothelial dysfunction caused by surgically induced systemic inflammation may lead to BBB disorders (Plog et al., 2015; Ekeloef et al., 2017). Inflammation after laparotomy can decrease the levels of tight junction proteins such as claudin, β-catenin, occludins, and ZO-1, all of which can contribute to tight junctions and BBB permeability (Yang et al., 2017). Moreover, astrocytes are easily activated by postoperative inflammatory cytokines such as TNF-α, IL-1β, IL-6, complement C3, and high mobility group box-1, which can subsequently damage the integrity of the BBB, especially in the hippocampus (He et al., 2012; Jin et al., 2014; Xiong et al., 2018). Activated astrocytes can also encourage tissue proliferation, resulting in scar-like perivascular barriers (Voskuhl et al., 2009). The endothelial cell membrane and the BBB close junctions comprise the vascular components of the glymphatic system, while the avascular components consist of astrocyte end-feet integrated with glia-line boundaries between the neuropil and axon tract regions (Plog and Nedergaard, 2018). Therefore, systemic inflammation influences the function of the glymphatic system based on the evidence that the structure and function of the glymphatic system are closely related to that of the BBB (Verheggen et al., 2018). Notably, a recent study found that lipopolysaccharide-induced systemic inflammation reduced perivascular CSF tracer flow and penetration into the parenchyma (Manouchehrian et al., 2021). These observations may be beneficial for our understanding of the role of systemic inflammation in glymphatic clearance.

Notably, the adverse cerebral effects of anesthesia were primarily reported in the elderly (Bedford, 1955; Berger et al., 2015). It is not yet fully understood why elderly patients are highly susceptible to PND, while young people who undergo surgery rarely develop it (Hu et al., 2020). Researchers have discovered some PND-susceptible patients with preoperative disease, such as preclinical AD (Xu et al., 2014), hyperglycemia (Zhang et al., 2014), cerebral microemboli (Wang et al., 2020), and other underlying neurovascular diseases or neurodegeneration (Wang et al., 2020). These preoperative diseases may share common underlying mechanisms and contribute to neurocognitive dysfunction or dementia. With advances in imaging technology, it has been found that these diseases have glymphatic impairment (Gaberel et al., 2014; Kress et al., 2014; Kyrtsos and Baras, 2015; Jiang et al., 2017; Hadjihambi et al., 2019). A theory states that susceptible patients with PND and aging have preexisting glymphatic dysfunction but progresses slowly without significant cognitive impairment (Bolte et al., 2020). Perioperative risk factors exacerbate the preexisting glymphatic damage beyond the compensatory limit, resulting in significant cognitive changes in these individuals. AD is a progressive neurodegenerative disorder characterized by gradual cognitive decline, which is characteristic of extracellular Aβ deposition and intracellular accumulation of hyperphosphorylated tau (Long and Holtzman, 2019). Patients in the preclinical stage of AD, particularly those with APOE4 genetic risk factors (Mentis et al., 2020), already have Aβ accumulation and impaired glymphatic clearance, but no clinically detectable symptoms of cognitive impairment (Masters et al., 2015). Moreover, aging is associated with degenerative changes in the brain's structure and function, such as reduced cerebrovascular pulsatility, depolarized or downregulated AQP4, and sleep deprivation, which impair glymphatic function (Zeppenfeld et al., 2017). Glymphatic dysfunction may result in the deposition of toxic solutes (including amyloid) in the aging brain, which exacerbates glymphatic dysfunction and creates a vicious cycle (Iliff et al., 2013). Vascular dementia is caused by many neurovascular diseases, including hypertension, ischemic stroke, traumatic brain injury, and microinfarcts, which significantly impair arterial CSF pulsation and glymphatic clearance (Riba-Llena et al., 2018). Diabetes can also impair cardiac pulsation due to pathological changes in the vascular endothelium (Jiang et al., 2017).

Due to technical limitations, few previous studies have used non-invasive methods to investigate glymphatic function in humans, particularly hospitalized surgical patients susceptible to PND. Recent advances in imaging, such as real-time MRI (Ding et al., 2018), ultra-fast MRI (Kiviniemi et al., 2016), and gadolinium-based contrast agents enhanced MRI (Ringstad et al., 2018; Deike-Hofmann et al., 2019), will provide a sensitive and non-invasive tool for assessing glymphatic function in clinical trials. Notably, it is essential to develop intraoperative monitoring tools for glymphatic function. A new method of measuring human glymphatic function that uses near-infrared spectroscopy has been developed for long-term monitoring of brain function, which is highly compatible with perioperative monitoring (Myllylä et al., 2018). These non-invasive methods may help translate glymphatic measurements from the laboratory to the clinic (Jiang, 2019). Advances in technology will yield new and valuable information about the glymphatic system, providing a quantitative map for the diagnosis, monitoring, and prognosis of neurological disorders, including PND.

Recent discoveries concerning the glymphatic system anatomy and mechanisms have contributed to a better understanding of CSF circulation and waste removal. Furthermore, the glymphatic system has an immune function and may affect the onset and progression of PND, either directly or indirectly. Because clinical research on perioperative glymphatic changes is still lacking, scientists and clinicians, particularly gerontologists, neurologists, psychiatrists, anesthesiologists, and surgeons, can collaborate to uncover the etiological mechanisms of perioperative glymphatic changes. The glymphatic concept may provide a strategic breakthrough in rethinking, treating, and, most importantly, preventing PND.

Human studies exploring the physiological and pathological mechanisms between the glymphatic system and brain cognition over time have greatly improved our knowledge; however, some key issues still need to be addressed by future research. First, the definition of normal or healthy glymphatic function may be one of the biggest challenges. A simpler and quicker intraoperative monitoring method remains to be developed. A better description of glymphatic dynamics will help demarcate “normal” and “abnormal” glymphatic function in humans and identify preventive strategies for PND. Second, perioperative factors such as anesthetic type, intraoperative hemodynamics, surgery position, surgical inflammation, and previous health status may confound research on the glymphatic system and PND. Cross-sectional clinical studies have demonstrated specific changes in the glymphatic transport of patients susceptible to neurological diseases, which is helpful in predicting the incidence of PND. Third, more clinical studies should be conducted instead of purely animal studies. Although association analysis may provide critical information for cause-effect deductions, correlation does not necessarily mean causation. Fourth, basic research targeting the perioperative glymphatic system in PND merits optimization; specifically, appropriate model systems should be carefully selected. Finally, the translation of basic research results into clinically relevant effects in humans should be expedited. Most data on the role of the glymphatic system are based on animal studies. Preclinical animal studies frequently result in unexpected failures during clinical transition due to unidentified reasons. Alternatively, future therapeutic interventions will likely be based on individual factors due to significant differences in anatomical variation and disease complexity among human populations.

XR, GZ, LL, and CL contributed significantly to conception of the review and edited the review. XR undertook the literature, drafted, and corrected most of the manuscript and tables. SL, KL, and HL assisted with part of the pictures and manuscript. CL contributed to the sections of glymphatic system and anaesthetics. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.