State of the Art and Promise of Structural Neuroimaging in Postoperative Delirium and Postoperative Cognitive Decline

Postoperative delirium (POD) is a common, potentially life-threatening complication of surgical procedures that occurs in up to 50% of patients over 65 years of age.¹ POD is intimately associated with cognitive decline. Impairment in cognitive abilities constitutes a risk factor as well as a consequence of POD.¹ Patients with delirium have about six times higher relative risk for developing dementia,² and occurrence of delirium can accelerate cognitive decline trajectories in persons without dementia³ or with Alzheimer disease (AD) and related dementias.⁴ The availability of early biomarkers may aid surgical planning and postoperative management of patients identified to be at high risk for developing POD and postoperative cognitive decline (POCD).

The review article by Kant et al.⁵ is a comprehensive summary of the existing literature on structural magnetic resonance imaging (MRI) biomarkers that could potentially be used to stratify surgical patients based on their risk of developing POD and POCD. The authors focused on cerebrovascular and neurodegenerative features of brain damage, such as white matter hyperintensities (WMHs) and cerebral infarcts, and measures of global and regional brain atrophy. In reviewing 15 neuroimaging studies, yielding a total of 1,422 surgical patients, Kant et al. found that although brain atrophy did not show a consistent association with POD and POCD, a significant association of cerebrovascular abnormalities (WMHs and infarcts) with POD and POCD was reported by most studies included. In our recent study, a cohort of 146 older individuals without dementia undergoing elective noncardiac surgery (Successful Aging After Elective Surgery, or SAGES), 32 of whom developed delirium after surgery, showed no association between WMHs and POD.⁶ Differences in the clinical settings (e.g., intensive care, elective surgery), characteristics of the study participants (e.g., inclusion/exclusion of patients with dementia at baseline), design (e.g., availability of presurgical imaging), and confounding factors (e.g., age differences between delirium and nondelirium groups) need to be critically considered, because they may explain the observed discrepancies between studies. The different approaches and settings of the already limited number of studies available made it unfeasible for Kant et al. to generate and test pooled effect sizes. However, even larger high-quality studies primarily showed either relatively mild or no association between MRI-evident brain abnormalities and POD or POCD. Given these findings, presurgical MRI does not provide a cost-effective screening procedure to assess the risk of POD and POCD based on cerebrovascular abnormalities in clinical routine.

However, neuroimaging studies may improve our understanding of the pathophysiologic underpinnings of POD and POCD. Interestingly, some of the studies reviewed by Kant et al. did show associations of POD with subcortical deep WMHs and cerebral infarcts in surgical cohorts, including 7 to 18 delirium patients (see Hatano et al., 2013; Omiya et al., 2015; and Otomo et al., 2013 in their table 4). Because cerebral WMHs may reflect different phenomena, including ischemic injury, increased blood–brain barrier permeability, and perivascular inflammation, it is possible that the association between brain damage and POD becomes more prominent in the presence of ischemic injury at the histopathologic level, which is shared by certain WMH subtypes (e.g., deep WMHs) and cerebral infarcts.

In addition to improving our understanding of the possible mechanisms underlying POD and POCD, neuroimaging studies can provide detail on the functional networks involved in the pathogenesis of POD and POCD. In our SAGES study cohort, we showed an association between POD and presurgical diffusion MRI abnormalities predominantly in attentional networks.⁷ Both WMHs and microstructural abnormalities as measured by diffusion MRI may affect the connectivity between brain areas involved in cognitive-behavioral functions. Pre-existing damage to structures of relevant brain networks (such as the frontoparietal control network and the default mode network) can lead to increased vulnerability to failure under the stress of surgery and therefore enhance the risk of developing POD.

An important open question concerns the potentially causal relationship between delirium and cognitive decline, which is particularly difficult to investigate because these conditions often coexist. To date, it is still unclear whether delirium represents the manifestation of a vulnerable brain that is already on an accelerated path to neurodegeneration (e.g., preclinical AD) or a new injury mainly dependent on factors related to the surgical procedures. It also remains unclear whether cognitive decline following delirium and AD are separate entities or if they share common pathophysiologic underpinnings. Future studies investigating the relationship between AD biomarkers and delirium may provide relevant pieces of information to address those open questions. Important considerations for future studies include the use of robust, standardized methodology to assess delirium (e.g., the Confusion Assessment Method),⁸ acquisition of presurgical MRI, and quantitative MRI analysis, combined with rigorous examination of perioperative factors.