Neuroinflammatory and morphological changes in late-life depression: the NIMROD study

Summary We studied neuroinflammation in individuals with late-life, depression, as a risk factor for dementia, using [11C]PK11195 positron emission tomography (PET). Five older participants with major depression and 13 controls underwent PET and multimodal 3T magnetic resonance imaging (MRI), with blood taken to measure C-reactive protein (CRP). We found significantly higher CRP levels in those with late-life depression and raised [11C]PK11195 binding compared with controls in brain regions associated with depression, including subgenual anterior cingulate cortex, and significant hippocampal subfield atrophy in cornu ammonis 1 and subiculum. Our findings suggest neuroinflammation requires further investigation in late-life depression, both as a possible aetiological factor and a potential therapeutic target.

To estimate TSPO binding site density, non-displaceable binding potential (BP ND ) was determined from [ 11 C]PK11195 PET data with the guidance of T 1 weighted MRI. 5 Regional white matter hyperintensity (WMH) volumes were segmented and quantified using T 2 FLAIR images. T 2 hippocampal coronal scans were manually segmented for cornu ammonis 1 (CA1), CA2, CA3/dentate gyrus, and measured for subiculum and entorhinal thickness. (See online supplement DS1 for detailed methods of PET analysis as well as WMH and hippocampal segmentation.) As a result of the relatively small sample size in the depression group, we did not assume Gaussian data distributions. Grouplevel statistical comparisons of regional [ 11 C]PK11195 PET BP ND as well as demographic, cognitive, blood, white matter lesion and hippocampal subfields data were performed using non-parametric Mann-Whitney U-test. Chi-squared test was used to test for gender differences between groups. In addition, regional [ 11 C]PK11195 BP ND data were also analysed using Monte Carlo randomisation tests to obtain P-values for each participant. (See online supplement DS1 for further details.) Results are reported without correction for multiple comparisons, noting that the primary outcome measure related to our principal hypothesis was [ 11 C]PK11195 BP ND in limbic cortical regions associated with depression and dementia.

Results
The two groups did not differ in age, gender ratio, education or global cognition (Mini-Mental State score 6 ) but the depression group had significantly higher blood CRP levels than controls (mean CRP: depression group 18.8 mg/L, control group 1. We studied neuroinflammation in individuals with late-life depression, as a risk factor for dementia, using [ 11 C]PK11195 positron emission tomography (PET). Five older participants with major depression and 13 controls underwent PET and multimodal 3T magnetic resonance imaging (MRI), with blood taken to measure C-reactive protein (CRP). We found significantly higher CRP levels in those with late-life depression and raised [ 11 C]PK11195 binding compared with controls in brain regions associated with depression, including subgenual anterior cingulate cortex, and significant hippocampal subfield atrophy in cornu ammonis 1 and subiculum. Our findings suggest neuroinflammation requires further investigation in late-life depression, both as a possible aetiological factor and a potential therapeutic target. significantly higher [ 11 C]PK11195 BP ND compared with controls in left subgenual anterior cingulate cortex (mean BP ND : depression group 0.1103, control group 0.0246; W = 54, P = 0.035), and right parahippocampus (depression group 0.1225, control group 0.0490; W = 53, P = 0.046); these are substantiated by the voxel-wise results given in Fig. 1 and online Fig. DS5. Using the individual-level Monte Carlo randomisation test, all five individuals in the depression group showed a significant increase of [ 11 C]PK11195 BP ND in the aforementioned brain regions, confirming the group-level statistical inference.
The depression group showed trends for more extensive WMH in both periventricular (depression group 7.33 ml, control group 3.74 ml; W = 13, P = 0.059) and deep (depression group 1.85 ml, control group 0.73 ml; W = 15, P = 0.095) white matter. We found a significant reduction of CA1 area in the coronal plane (depression group 22.09 mm 2 , control group 24.90 mm 2 ; W = 52, P = 0.019) and subiculum thickness (depression group 1.73 mm, control group 1.95 mm; W = 56, P = 0.004) in the depression group. (See online supplement DS2 for additional demographic, cognitive, WMH and volumetric results.)

Discussion
We found evidence of both central and peripheral inflammation in older individuals with depression, including changes in the anterior cingulate and medial temporal lobe, which play a key role in the regulation of mood and cognitive functioning. 8 Damage in these areas is linked with an elevated risk of dementia. 9 Increased [ 11 C]PK11195 binding in people with depression could be associated with cerebrovascular disease and white matter lesions, reported in the current and previous studies, 10 although some controls also had a similar burden of WMH with normal levels of [ 11 C]PK11195 binding in subgenual anterior cingulate cortex and parahippocampus (online Fig. DS5). It is notable that [ 11 C]PK11195 BP ND showed the greatest effect size compared with other modalities, with a 300% increase from controls (v. 150% for WMH and 10% for hippocampal atrophy), suggesting a strong biomarker potential for late-life depression.
There was no major cognitive impairment in our cohort, although the depression group showed significant atrophy in the hippocampus and subiculum, which have been shown to correlate with greater risk of cognitive impairment and Alzheimer's disease. 11 In addition, the hippocampus is a key component in the hypothalamic-pituitary-adrenal (HPA) axis. Increases in cytokine levels can lead to increases in oxidative stress and glucocorticoid as well as decreases in serotonin and other neurotransmitters in HPA resulting in impaired mood and cognition. 12,13 Our results were not corrected for multiple comparisons, and further replication is required in a larger cohort. However, the large effect size of [ 11 C]PK11195 was in keeping with our principal hypothesis and was supported by both a primary group-level test and secondary individual statistical tests. Cross-sectional studies provide limited information about whether neuroinflammation was the cause or consequence of neuronal damage in affected brain areas, so future longitudinal studies are needed. In conclusion, we suggest that neuroinflammation may be an important mechanism in late-life depression and merits further investigation as a potential target for novel therapeutics in a condition that responds poorly to conventional antidepressant therapy.