DELUSIONAL SEVERITY IN A COGNITIVELY MIXED SAMPLE OF OLDER ADULTS IS ASSOCIATED WITH ABNORMALITIES IN WHITE MATTER TEXTURE

Introduction Background: Delusions occur in older patients and may be a harbinger of later cognitive impairment. Studies to date have suggested a correlation with white matter lesion burden [2] [3] but the correlation between delusions and white matter texture (WMT) is not well understood. White matter texture refers to tissue organization in normal-appearing brain matter (NABM) related to the overall diffusion capabilities of brain tissue. There is growing interest in analyzing these regions of the brain since they may contain early signs of disease [4]. Whether delusion severity is correlated with WMT is not so clear. Knowing the relationship between delusional severity and WMT may provide insight into potential mechanisms. Objective: Using neuroimaging biomarkers measured from 212 FLAIR MRI volumes, we investigated the relationship between delusional level and texture of the NABM in a cognitively mixed cohort extracted from the Alzheimeru0027s Disease Neuroimaging (ADNI) data set. Methods Method: We extracted a cognitively diverse sample of patients with delusions based on NPI-Q scores from the ADNI data set. The delusion group was organized according to severity (1, 2, 3) based on Neuropsychiatric Inventory-Quick (NPI-Q) scores, with 1 being the lowest severity, and FLAIR MRI volumes for each subject was downloaded. There were 32 subjects (130 volumes) in group 1, 11 subjects (49 volumes) in group 2 and 9 subjects (33 volumes) in group 3 for analysis. Using validated, in-house brain extraction and intensity standardization algorithms [5][6][7], the normal-appearing brain matter (NABM) of FLAIR images were extracted. Texture analysis was performed on the NABM with a 2D texture feature that measures spatial correlation among pixel intensities [1]. This 2D spatial correlation metric was extracted on a per pixel basis, using 3mm x 3mm windows to find local, microstructural texture differences. The spatial feature maps were averaged over all slices to find a volume representation and a slight erosion was applied to remove outliers caused by the large brain-background edge. From the volumeu0027s correlation histogram, the bottom 2% and top 98% of the features are removed to further improve robustness to outliers. The mean and median of these feature maps were statistically compared between delusion groups (1, 2, 3). High spatial correlation indicates neighboring pixels have similar intensities (and is more homogeneous). Results Results: Boxplots for the NABM texture measures are shown for groups 1, 2 and 3 in Figure 1. Subjects with high delusional severity (group 3) had the most noticeable change in WMT when compared to lowest delusional severity (group 1). Group 3 had a higher spatial correlation value of 6.0947 e+03 +/- 592.6178 than group 1 with values of 5.7630 e+03 +/- 832.5040. A higher value indicates more spatial correlation among pixels (i.e. local homogeneity since closer pixels are more likely to have similar intensity values). This suggests that there are larger regions that are smooth in Group 3 compared to Group 1, which could indicate less WM “structure” in the more severe delusions group. This effect is shown in Figure 2. T-test results in Table 1 shows the highest difference is between groups 1 and 3 when examining the mean and median values with p-values both Conclusions Discussion: This study provides the first evidence that severity of delusions in a cognitively diverse sample of older adults is correlated with abnormalities in white matter texture. This finding suggests that alterations in white matter texture may contribute to the development of delusions in older patients. There is a notable difference between an individual with mild symptoms and severe symptoms based on their white matter texture as shown. Due to the subjective nature of an individual being scored either NPI-Q 1 or 2, or, 2 or 3, differentiating between mild and moderate, or moderate and severe disease is often ill-defined. This may contribute to challenges in differentiating between these groups. Further study comparing our results to patients without delusions and controlling for severity of cognitive impairment is required to further validate our findings. This research was funded by: St. Michaelu0027s Hospital Foundation, Heather and Eric Donnelly endowment