Postoperative delirium is an acute confusional state that occurs in up to half of surgical patients aged 65 years or older (Inouye, 2006). Several studies, including our own work, showed that although delirium is often a transient condition, it is associated with poor long-term outcomes, including the development of dementia (Inouye et al., 2016, Saczynski et al., 2012, Witlox et al., 2010). Evidence from epidemiological studies suggests that delirium and dementia are synergistically deleterious conditions. Delirium is an established independent risk factor for dementia (Witlox et al., 2010). The Delirium and Cognitive Impact in Dementia (DECIDE) study showed an increased risk of incident dementia one year after delirium in hospitalized patients aged ≥ 65 years (OR = 8.8, 95 %-CI = 1.9–41.4) (Richardson et al., 2021). A recent meta-analysis of 24 studies, including 10,549 surgical and non-surgical patients (n = 3,562 with delirium), showed that patients who experienced delirium had 2.3 times the odds of developing long-term cognitive decline (LTCD) compared to their non-delirious counterparts (OR = 2.30, 95 %-CI = 1.85–2.86) (Goldberg et al., 2020). The risk of developing delirium after surgery is 2–5 times higher in patients with dementia, and among surgical patients with dementia those who develop delirium show faster progression of cognitive decline (Fong et al., 2015). It is still unclear whether the pathophysiologic processes associated with delirium play a causative role in the subsequent development of dementia or whether delirium reflects underlying brain vulnerability, such as in the preclinical stages of diseases, that ultimately lead to dementia. The neural substrates of both postoperative delirium and LTCD are also poorly understood (Migirov et al., 2021).

Magnetic resonance imaging (MRI) has provided insights into the neural substrates of postoperative delirium and LTCD (for a review see Kant et al., 2017), and constitutes an integral part of the diagnostic workup for the assessment of patients with or at risk for dementia in clinical practice (Ikram et al., 2010, Jack et al., 2013). Neurodegenerative diseases associated with dementia produce atrophy detectable with MRI, which can be measured prior to the onset of cognitive impairment (Dickerson et al., 2011, Dickerson et al., 2009, Driscoll et al., 2009, Fox et al., 1996, Jack et al., 2013, Keret et al., 2021, Scahill et al., 2002). Brain atrophy is an established neural substrate of postoperative delirium and cognitive decline. The association between brain atrophy and delirium has been established three decades ago in computed tomography studies (Koponen et al., 1989, Koponen et al., 1987), and more recently confirmed by MRI studies (Gunther et al., 2012, Kant et al., 2022, Kant et al., 2021, White et al., 2021).

Although delirium is common in people with AD and other dementias, there is much less understanding of why delirium occurs in individuals without dementia, or why some individuals who experience delirium seem to be at increased risk for LTCD and subsequent dementia. In previous studies of the same cohort of older surgical patients without dementia used in the present work, we showed that pre-surgical evidence of thinner cortex in brain regions relevant to Alzheimer’s disease (AD) was associated with more severe delirium and faster pace of LTCD over 36 months (Racine et al., 2020, Racine et al., 2017).

Preclinical AD and vulnerable aging are two potentially important contributors to the increased risk of postoperative cognitive decline. In this study, we probed two non-overlapping sets of brain regions: the Alzheimer’s disease (AD) signature, which is a set of brain regions with reduced cortical thickness reflecting AD-related neurodegeneration, and associated with cognitive decline, progression to dementia, and positron emission tomography and cerebrospinal fluid biomarkers of AD (Bakkour et al., 2013, Bakkour et al., 2009, Dickerson et al., 2011, Dickerson et al., 2009, Dickerson and Wolk, 2013, Dickerson and Wolk, 2012); and the vulnerable aging signature, which reflects cortical thickness changes associated with the aging process measurable in older adults free of cortical amyloid positron emission tomography signal or evidence of cerebrovascular disease (Bakkour et al., 2013).

The present study investigates in older adults undergoing elective non-cardiac surgery the relationship between delirium during hospitalization, longitudinal changes in cortical thickness, as measured from MRI acquired before and one year after surgery, and LTCD over 72 months. Our overarching hypothesis is that delirium has a detrimental effect on the development of brain atrophy and long-term cognitive outcomes. Specifically, we tested the following three hypotheses.

First, that individuals with postoperative delirium would develop greater cortical atrophy in the AD or vulnerable aging signature than those without delirium over the course of 1 year after surgery.

Second, that the development of more severe cortical atrophy in the AD or vulnerable aging signature would be associated with a faster pace of postoperative LTCD. Based on the evidence that cortical atrophy can be detected years before cognitive deficits become apparent from MRI studies of aging, AD and other dementias (Dickerson et al., 2011, Dickerson et al., 2009, Driscoll et al., 2009, Fox et al., 1996, Jack et al., 2013, Keret et al., 2021, Scahill et al., 2002), we investigated whether changes in cortical thickness over the course of 1 year after surgery were associated with subsequent changes in cognitive performance occurring between 1 and 6 years after surgery. Baseline cognitive data were included as covariable in the statistical models to assess the association between MRI changes and LTCD.

Third, that the relationship between cortical atrophy and LTCD would be stronger in patients who had experienced post-operative delirium. The premise to this analysis is that the neural substrates of delirium and its consequences may include neurodegenerative changes that can be detected on MRI (e.g., the development of cortical brain atrophy) and are associated with heightened risk for developing cognitive deficits.