Delirium is a neuropsychiatric syndrome characterized by acute fluctuating disturbances in attention, consciousness, and cognition (Inouye et al., 2014; Marcantonio, 2017). It is common in older hospitalized adults, affecting one-third of those aged 70 years and over (Laurila et al., 2004; Marcantonio, 2017), and is associated with numerous poor outcomes including prolonged hospital stay, cognitive impairment, functional disability, and mortality (Fong et al., 2009; Inouye et al., 2014; Marcantonio, 2017; Goldberg et al., 2020). Though some progress has been made in diagnosis, elucidation of risk factors and prognosis, the neurobiological basis of delirium has not yet been completely clarified. Consequently, few effective and mechanism-based treatments for delirium are available. It is therefore imperative to explore neurobiological biomarkers associated with delirium, which may not only advance our understanding of the pathogenesis of delirium but also aid in delirium risk stratification to implement proactive interventions.

The advancement of neuroimaging particularly magnetic resonance imaging (MRI) provides a unique chance to study the neural substrates of delirium, and certain abnormalities on brain MRI have been reported to be associated with delirium (Nitchingham et al., 2018; Kalvas and Monroe, 2019). Several studies have indicated that white matter hyperintensities, markers of cerebral small vessel disease (CSVD), can predict the occurrence of delirium (Omiya et al., 2015; Nitchingham et al., 2018; Clancy et al., 2021; Pendlebury et al., 2022). Nevertheless, for other CSVD markers, especially MRI-visible perivascular spaces (PVS), this is still unknown. PVS are fluid-filled spaces around the brain perforating small vessels and are hypothesized to act as part of the drainage systems which facilitate fluid exchange and clearance of metabolic wastes from the brain (Wardlaw et al., 2020; Gouveia-Freitas and Bastos-Leite, 2021). Although it is normal to have a few visible PVSs on MRI in healthy young brains (Piantino et al., 2020; Barisano et al., 2021), they may become increasingly common and even enlarge in the context of aging, vascular risk factors, and other features of CSVD (Wardlaw et al., 2020), which may have important clinical implications to brain health (Wardlaw et al., 2020; Gouveia-Freitas and Bastos-Leite, 2021). And growing studies have shown that PVS burden is related to cognitive decline and dementia in older adults (Ding et al., 2017; Debette et al., 2019; Paradise et al., 2021). However, no studies have previously explored PVS in relation to delirium. Whether PVS plays a role in the development of delirium remains unclear.

Therefore, in the present study, we aimed to determine the association between PVS and risk of future delirium and to examine the predictive ability of PVS for delirium in older hospitalized patients. We hypothesized that older patients with a higher PVS burden at baseline would be at greater risk of subsequent delirium.

In the present study, we found that CS-PVS on MRI acquired within 6 months prior to admission was independently associated with subsequent delirium in older hospitalized patients. Our study suggests that existing previously acquired brain MRI may represent an underlying resource, and CS-PVS visible on these MRIs may serve as a risk factor of delirium and may provide important predictive information for delirium occurrence. Incorporating CS-PVS into delirium prediction algorithms might have potential clinical utility in aiding delirium risk stratification, and this needs to be confirmed by future studies.

Although a few studies have investigated brain CSVD markers (e.g., WMH and cerebral microbleeds) in relation to delirium (Omiya et al., 2015; Nitchingham et al., 2018; Lachmann et al., 2019; Clancy et al., 2021; Pendlebury et al., 2022), no previous study has yet explored the link between MRI-visible PVS and delirium. PVS is frequently seen in the aging brain (Bown et al., 2022) and has been proposed as a key contributor to cognitive decline and dementia in older adults (Debette et al., 2019; Paradise et al., 2021). It has been reported that PVS could affect specific cognitive domains, and a negative association of PVS burden with nonverbal reasoning and visuospatial ability has been observed (Maclullich et al., 2004). Recent studies also indicated that PVS was correlated with worse executive function, information processing speed, and mild cognitive impairment (Ding et al., 2017; Passiak et al., 2019; Sepehrband et al., 2021). However, whether PVS plays a role in the development of delirium remains unknown. Our present study adds to evidence on the association of PVS with delirium, suggesting that PVS (particularly that in CS) may be an important risk marker of delirium. The previously reported significant relationship between deep WMH and delirium (Hatano et al., 2013; Omiya et al., 2015) may support our findings regarding PVS in the CS region. However, as the PVS-WMH association has not been fully elucidated (Francis et al., 2019) and the limited studies were available about the correlation of CSVD markers and their distributions with delirium, more investigations are warranted to validate our results and to explore the relationship between PVS locations and delirium. Besides, in this study, we found that the associations between CS-PVS and delirium persisted and remained significant even after adjusting for cognitive impairment, one of the strongest predictors of delirium, which implies that CS-PVS is an independent risk factor for delirium rather than a surrogate measure of cognitive performance.

Several explanations might account for the observed association between CS-PVS and delirium. First, CS-PVS is considered a marker of cerebral amyloid angiopathy (van Veluw et al., 2016; Charidimou et al., 2017). Abnormal protein aggregation (e.g., β-amyloid) can block the upstream within the cortical arteries and impair the external drainage of interstitial fluid in the deep white matter, thus contributing to the retrograde enlargement of PVS in CS (Gouveia-Freitas and Bastos-Leite, 2021), and the relationship between β-amyloid accumulation and CS-PVS has been demonstrated in previous studies (Roher et al., 2003; Charidimou et al., 2015). Furthermore, the impaired perivascular pathway may cause a vicious feed-forward cycle, further prompting PVS dysfunction and β-amyloid deposition in the vascular wall and brain (Gouveia-Freitas and Bastos-Leite, 2021; Bown et al., 2022), which may exacerbate neurovascular dysfunction, neuroinflammation, and neurodegeneration (Carrano et al., 2012; Parodi-Rullan et al., 2021), making the brain vulnerable to delirium (Idland et al., 2017; Maldonado, 2018; Chan et al., 2021).

In addition, PVS contains a range of immune cells contributing to immune surveillance and potentially neuroinflammation (Gouveia-Freitas and Bastos-Leite, 2021). Enlarged PVS is likely to involve inflammation (Wuerfel et al., 2008; Wardlaw et al., 2020) and is proposed as a marker of blood-brain barrier dysfunction (Potter et al., 2015b), which may cause disturbances in the neuronal network and predispose patients to subsequent delirium (Maldonado, 2018). Moreover, it has been reported that metabolite clearance via PVS occurs particularly during sleep (Hablitz et al., 2020; Wardlaw et al., 2020), and sleep deficiency may lead to structural changes in PVS (Del Brutto et al., 2019; Lysen et al., 2021). Growing studies have shown that sleep disruption and circadian dysfunctions, also known contributing factors of delirium (Maldonado, 2018), are involved in the occurrence of enlarged PVS (Berezuk et al., 2015; Del Brutto et al., 2019; Opel et al., 2019; Lysen et al., 2021). These data suggest that the observed relationship between PVS and delirium might be mediated by sleep disturbances. As assessment of sleep-related parameters was not available in the present study, we were unable to explore the potential effect of sleep disturbances. Further studies are warranted to elucidate the complex interrelationships between sleep, PVS, and delirium.

The strengths of our study include that delirium was ascertained at the bedside by trained assessors using the well-validated standardized assessment tool, rather than diagnosed directly from electronic medical records or hospital administrative diagnostic codes (e.g., ICD-10). The rigorous measures of delirium increase the reliability and reduce the misdiagnosis. In addition, this study was the first investigation to explore the association between PVS and delirium. Our study adds new, clinically relevant information to the literature on PVS and delirium, suggesting that CS-PVS on brain MRI could predict delirium occurring up to 6 months later. In other words, our present study suggests that existing previously acquired brain MRIs may represent an underutilized resource, and incorporating neuroimaging markers (e.g., CS-PVS) on these brain MRIs into delirium prediction algorithms may have the potential to identify delirium high-risk patients who may benefit from proactive interventions.

However, some study limitations warrant attention. First, this study was a retrospective analysis based on an internal medicine population in a single geographic area, and our findings may not be generalizable to other populations. Second, the sample size was relatively small and was unequal between delirium and non-delirium groups (the delirium group was 3 times smaller than the non-delirium group), which limited the statistical power and may raise the chance of false positives or false negatives. Besides, given the limited number of outcome events, to avoid overcontrolling, we did not adjust for other potential confounders (such as sensory impairment, treatments after admission, reasons for MRI scans, type and severity of patients’ acute diseases, etc.), and residual confounding cannot be excluded. Our study therefore should be considered preliminary, and future studies with larger sample sizes are needed to validate our findings. Third, correction was not made for multiple comparisons, which may increase the risk of type I error. Our results should be considered as hypothesis generating, and the P values should be interpreted with caution. Fourth, potential selection bias may be introduced since many patients without MRI scans were excluded, although there was no significant difference in most baseline characteristics between patients with and those without MRI scans (Supplementary Table 1). Compared to relatively healthier patients who didn’t require MRI, those with MRI might have a lower brain reserve and may have a higher PVS burden, which may potentially affect the relationship between PVS and delirium and limit the generalizability of our results to all older patients. More prospective, multicenter, large-sample studies are warranted to verify our findings. Fifth, as our original cohort was designed to investigate the incidence of delirium, patients with delirium on admission were excluded. Thus, the association between PVS and subsequent delirium observed in this study may be underestimated. In addition, we did not collect data on delirium severity, and whether PVS is related to delirium severity requires further investigation. Finally, PVS was assessed using a semiquantitative visual rating scale. Although practical to use in clinical research with good reliability and repeatability (Potter et al., 2015a), the scale is relatively insensitive and constrained by floor and ceiling effects (Wardlaw et al., 2020; Gouveia-Freitas and Bastos-Leite, 2021). The automated measures of PVS number and volume seem more precise, and further studies are needed to explore their value in predicting delirium.

CS-PVS on MRI acquired 6 months earlier was independently associated with the occurrence of subsequent delirium in older hospitalized patients. Integration of CS-PVS to the established risk factors enhanced the risk refinement and reclassification for delirium. CS-PVS may serve as a promising marker to identify delirium high-risk patients for proactive implementation of preventive interventions.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Biomedical Research Ethics Committee of West China Hospital, and conformed to the ethical guidelines of the Declaration of Helsinki. The patients/participants provided their written informed consent to participate in this study.

JY and QS conceptualized and designed the study. QS, YZ, and TL collected and interpreted the data. QS analyzed the data, prepared and reviewed the figures, and wrote the original draft. JY provided critical revisions to the manuscript. 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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.