Assessing brain morphology and inflammation in relation to episodic memory profiles among aging adults with HIV

Rationale: Identifying persons with HIV (PWH) at risk for Alzheimer’s disease (AD) is complicated because memory deficits are common in HIV-associated neurocognitive disorders (HAND) and amnestic mild cognitive impairment (aMCI; a precursor to AD). Due to differences in underlying brain changes, recognition deficits are common in aMCI but not in HAND; therefore, recognition may be a useful neurocognitive marker to differentiate these etiologies. There is a paucity of research aimed at identifying PWH on an AD trajectory and understanding biological mechanisms that may put PWH at higher risk of aMCI/AD. This study examined the relationship between baseline brain integrity and memory (i.e., recall and recognition) and the relationship between inflammation, brain integrity, and memory.Design: This study utilized longitudinal data from the CNS HIV Antiretroviral Therapy Effects Research program from 92 PWH between ages 45-68. Aim 1 used multivariable linear regression and logistic regression to examine neuroanatomical correlates of memory. Aim 2 utilized multilevel modeling to examine if baseline structural neuroimaging predicts memory decline (average follow-up = 5.7 visits over 6.5 years). Aim 3 examined if medial temporal lobe (MTL) structures mediate the association between peripheral inflammatory biomarkers (i.e., IL-6, TNF-a, CCL2, CRP, CXCL10) and memory. Results: Thinner pars opercularis thickness was associated with impaired recognition (p=0.012) and worse delayed recall (p=0.001), and thinner rostral middle frontal thickness was associated with worse delayed recall (p=0.006). Memory was not associated with MTL, basal ganglia, primary motor cortex, or other prefrontal structures. In Aim 2, recognition impairment was variable over time, and there was little decline in delayed recall (i.e., T-score change of -0.041 per year). Baseline MTL integrity was not associated with memory decline (ps>0.40). Lastly, MTL structures did not mediate the association between inflammatory biomarkers and memory. Conclusions: Memory in this sample of middle-aged PWH was associated with prefrontal structures and not MTL structures. MTL structures did not predict memory decline. This may suggest that memory is more related to frontally mediated etiologies, such as HIV, rather than AD pathology in middle-aged PWH. Additional research is needed to clarify if recognition can be used clinically to differentiate aMCI and HAND