Article Figures & Data
Figures
- Figure 1.
Schematic representation of various modality datasets in the study, their processing pipelines on a within-subject level (light blue), as well as data-reduction techniques and statistical strategy on between-subject level (dark blue) to test for associations between the datasets. FC, Functional connectivity; Cov, covariates; COG PC1, latent variable (cognitive deficit, which summarizes the largest portion of shared variance as the first principal component); GM, gray matter. NIMROD study (Neuroimaging of Inflammation in Memory and Other Disorders), AD/MCI+, Alzheimer's Disease/Mild Cognitive Impairment (positive PET amyloid, 11C PiB PET), PET PK, positron emission tomography [11C]PK11195 ligand (microglia activation), PK-IC maps, PET [11C]PK11195 independent component maps, PK IC3, 3rd component of [11C]PK11195 ligand maps, 11CPiB PET (amyloid PET), fMRI, functional magnetic resonance imaging, Cam-CAN, Cambridge Centre for Ageing and Neuroscience, G, diagnostic group (AD/MCI+ or Controls), ACE-R, Addenbrookes' Cognitive Examination - revised test, MMSE, mini-mental status examination, RAVLT, Ray Audio-Visual Learning Test, MLR, multiple linear regression.
- Figure 2.
Source Based Inflammetry (SBI) for the component differentially expressed between groups: (left) independent component spatial map reflecting increase in PK ([11C]PK11195 ligand- activated microgia) binding values in cortical and subcortical areas (red blobs) including inferior temporal cortex and hippocampus, these regionally specific increases are over and above the global PK differences between groups. Right, Bar plot of subject loading values for AD/MCI+ and control group (each circle represents an individual) indicating higher loading values for AD/MCI+ than control group as informed by two-sample unpaired permutation test (a robust linear regression was used to down-weight the effects of extreme data points). AD/MCI+, Alzheimer's Disease/Mild Cognitive Impairment (positive PET amyloid, 11C PiB PET), HC, healthy controls.
- Figure 3.
The Source Based Inflammetry (SBI) identified five independent components (ICs) that reflected PK ([11C]PK11195 ligand-activated microgia) binding values in cortical and subcortical areas. The PKIC3 component differed between AD/MCI+ patients and controls (first row, third column). This PKIC3 component negatively correlated with total gray-matter volumes in all individuals as well as in patients only (but not controls only; third column and second, third, and fourth rows). In other words, the patients expressing higher [11C]PK11195 binding PKIC3 component (reflecting higher binding in the inferior temporal cortex and hippocampus as shown in Fig. 2) displayed higher levels of brain-wide atrophy. GM, gray matter, AD/MCI+, Alzheimer's Disease/Mild Cognitive Impairment (positive PET amyloid, 11C PiB PET), HC, healthy controlls.
- Figure 4.
Mean effects (left) and group difference effects (AD/MCI+>Controls, right) between default mode network (DMN) and subcortical (SC) regions using univariate approach. vACC, ventral Anterior cingulate cortex; PCC, posterior cingulate cortex; IPL, intraparietal lobule; FPN, frontoparietal network; Put, putamen; Hipp, hippocampus, AG, angular gyrus; CN-MFG, middle frontal gyrus (cortical network), SFG, superior frontal gyrus; r, right; l, left. Note that the whole pattern of brain connectivity rather than each connection separately was used to study how subject-specific neuroinflammatory levels influence large-scale network connectivity (Fig. 5). AD/MCI+, Alzheimer's Disease/Mild Cognitive Impairment (positive PET amyloid, 11C PiB PET).
- Figure 5.
Left, First-level Multiple Linear Regression (MLR) indicating that functional connectivity differences (deviating from groups effects in Fig. 3) are associated positively with PKIC3 (PK PET ligand 3rd Independent Component). Connections surviving a threshold of p < 0.05 corrected for multiple comparisons are highlighted with a black contour, although it is important to bear in mind that the whole-pattern of brain connectivity was used in the analysis shown on the right. Right, Second-level MLR association between PKIC3 pattern of functional connectivity and cognitive performance for patients with AD pathology (including MCI+; orange) and control (green) groups. The group difference in slopes was significant (p < 0.0001). By using multiple linear regression and correlations with cognitive performance, we found that the change in patients' cognition was correlated to a pattern of brain connectivity that was itself linked to neuro-inflammation. This relationship between cognition and a PET-rsfMRI association (i.e., neuroinflammation-functional connectivity) was only seen in AD/MCI+ patients, not controls. vACC, ventral Anterior cingulate cortex; PCC, posterior cingulate cortex; IPL intraparietal lobule; FPN, frontoparietal network; SC, subcortical, DMN, default mode network, DMNd, dorsal DMN, Put, putamen; Hipp, hippocampus, AG, angular gyrus; SFG, superior frontal gyrus; R, right; L, left.
Tables
- Table 1.
Participant details (mean, with SD and range in parentheses) and group differences by χ2 test, one-way ANOVA, or independent samples t test
Demographic and clinical data AD/MCI+ (N = 28) Controls (N = 14) AD/MCI+ < Controls Sex, females/males 12/16 8/6 NS Age, years (SD, range) 72.7 (±8.5, 53–86) 68.3 (±5.4, 59–81) NS Education, years (SD, range) 12.9 (±3.0, 10–19) 14.1 (±2.7, 10–19) NS MMSE (SD, range) 25.6 (±2.2, 21–30) 28.8 (±1.0, 27–30) t = 4.9, p < 0.0001 ACE-R (SD, range) 78.9 (±7.7, 62–91) 91.6 (±5.3, 79–99) t = 5.5, p < 0.0001 RAVLT (SD, range) 1.5 (±1.6, 0–6) 9.6 (±3.2, 3–15) t = 10.8, p < 0.0001 NS, Not significant with p > 0.05 (uncorrected).
In this issue
