Predilection for Perplexion: Preoperative microstructural damage is linked to postoperative delirium

Abstract

Postoperative delirium is the most common postsurgical complication in older adults and is associated with an increased risk of long-term cognitive decline and Alzheimer's disease (AD) and related dementias (ADRD). However, the neurological basis of this increased risk-whether postoperative delirium unmasks latent preoperative pathology or leads to AD-relevant pathology after perioperative brain injury-remains unclear. Recent advancements in neuroimaging techniques now enable the detection of subtle brain features or damage that may underlie clinical symptoms. Among these, Neurite Orientation Dispersion and Density Imaging (NODDI) can help identify microstructural brain damage, even in the absence of visible macro-anatomical abnormalities. To investigate potential brain microstructural abnormalities associated with postoperative delirium and cognitive function, we analyzed pre- and post-operative diffusion MRI data from 111 patients aged ≥60 years who underwent non-cardiac/non-intracranial surgery. Specifically, we investigated preoperative variation in diffusion metrics within the posterior cingulate cortex (PCC), a region in which prior work has identified glucose metabolism alterations in the delirious brain, and a key region in the early accumulation of amyloid beta (Aβ) in preclinical AD. We also examined the relationship of preoperative PCC NODDI abnormalities with preoperative cognitive function. Compared to patients who did not develop postoperative delirium (n=99), we found increased free water (FISO) and neurite density index (NDI) and decreased orientation dispersion index (ODI) in the dorsal PCC before surgery among those who later developed postoperative delirium (n=12). These FISO differences before surgery remained present at six weeks postoperatively, while these NDI and ODI differences did not. Preoperative dorsal PCC NDI and ODI values were also positively associated with preoperative attention/concentration performance, independent of age, education level, and global brain atrophy. Yet, these diffusion metrics were not correlated with cerebrospinal fluid Aβ positivity or levels. These results suggest that preoperative latent brain abnormalities within the dorsal PCC may underlie susceptibility to postoperative delirium, independent of AD-related (i.e., Aβ) neuropathology. Furthermore, these preoperative microstructural differences in the dorsal PCC were linked to preoperative deficits in attention/concentration, a core feature of postoperative delirium. Our findings highlight microstructural vulnerability within the PCC, a key region of the default mode network, as a neuroanatomic locus that can help explain the link between preoperative attention/concentration deficits and increased postoperative delirium risk among vulnerable older surgical patients.

Keywords: Postoperative delirium; atrophy; attention; diffusion-weighted imaging; neurite density.

Figure 1:

STROBE Diagram

Figure 2:

PCC region of interest constrained diffusion metric analysis, from preoperative MRI scans. (A) Overlay of significant (p < 0.05) voxels for all three diffusion metrics. (B-D) Violin plots displaying the distribution of each diffusion metric by postoperative delirium status: (B) Fraction of isotropic diffusion (FISO), (C) Neurite Density Index (NDI), and (D) Orientation Dispersion Index (ODI).

Figure 3:

Scatter plots illustrating the relationships between diffusion metrics and cognitive factors. Spearman’s r, 95% confidence intervals and p values are provided for each plot. (A-E) Fraction of isotropic diffusion (FISO), (F-J) Neurite Density Index (NDI), (K-O) Orientation Dispersion Index (ODI). Each NODDI metric is subdivided by row for each of the following cognitive factors: Factor 1 - Narrative Memory, Factor 2 - Structured Memory, Factor 3 - Executive Function, Factor 4 - Figural Memory, and Factor 5 - Attention/Concentration. Significant associations were identified via Spearman’s (r) between both NDI (J) and ODI (O) and Factor 5 - Attention/Concentration (p < 0.05), respectively. All other associations were not significant (p > 0.05).

Figure 4:

Scatter plots illustrating the relationships between diffusion metrics and global cognitive function (Continuous Cognitive Index, or CCI). Spearman’s r, 95% confidence intervals, and p values are provided for each plot. No significant relationships (p > 0.05) were identified via Spearman’s (r) between any of the three NODDI metric (A) Fraction of Isotropic Diffusion (FISO), (B) Neurite Density Index (NDI), (C) Orientation Dispersion Index (ODI) and CCI.

Figure 5:

Violin and Scatter plots illustrating the relationships between diffusion metrics and cerebrospinal fluid Aβ. A-C) Diffusion metrics by Aβ42 positivity status. No significant differences were found in diffusion metrics between those with versus those without positive CSF levels of Aβ. D-F) Scatter plots for diffusion metrics by numeric levels of Aβ. Spearman’s r, 95% confidence intervals and p values are provided for each plot. No significant relationships (p > 0.05) were identified via Spearman’s (r) between the diffusion metrics (A, D) Fraction of isotropic diffusion (FISO), (B, E) Neurite Density Index (NDI), (C, F) or Orientation Dispersion Index (ODI) with CSF Aβ levels.