Post hoc delayed recall analyses were conducted that excluded participants who were excluded in post hoc analyses in aim 1 as well as those not consistently on ART and participants with a substance use disorder during follow-up. The total included after these exclusions was n=55. A single mediator model was used to investigate if medial temporal lobe structures had a mediating effect between the peripheral biomarkers of inflammation and recognition and delayed recall. The “lavaan version 0.6-12” R package was used to conduct all mediation analyses . Inflammation biomarkers were the independent variable , medial temporal lobe structures were the mediator , and recognition and delayed recall were the outcomes. A single mediator model using a robust diagonal weighted least squares approach with bias-corrected bootstrapping to obtain 95% confidence intervals was used for the dichotomous recognition outcome. However, there was poor model fit, and thus logistic regressions were conducted instead to examine if plasma biomarkers of inflammation had a cross-sectional association with recognition. A single mediator model with bias-corrected bootstrapping to obtain 95% confidence intervals were used for the continuous delayed recall outcome. Age, imaging covariate, and other covariates identified in aim 1 were included in analyses. There were not significant associations between plasma biomarkers and delayed recall; however,cannabis grower supplies mediation models were still conducted as there does not need to be a direct effect in order for a mediation effect to be tested or established .
Delayed recall mediation models were re-examined excluding participants that were excluded from post hoc analyses in aim 1 . Demographic and clinical characteristics are displayed in Table 3. Data presented in Table 3 represents participants’ first visit for this analysis, which is when the MRI and blood draw for biomarker analysis were completed. On average, participants were in their early-50s [range = 45 – 68], approximately half were African American/Black, were predominantly male, and had some college education. With regard to neurocognitive functioning, approximately one-third were classified as neurocognitively impaired via Frascati criteria . The average T-score for global cognition, recall, and processing speed was around 50, and the average Z-score for recognition was Z=-0.1; however, the average psychomotor skills T-score was lower than average at a T-score of 44. Approximately 20% of the sample was classified as impaired on delayed recall, and 13.2% were impaired on recognition. Recognition and delayed recall were correlated at r=0.358 . Of the 18 participants that were impaired on delayed recall, 7 were impaired on recognition whereas 11 were not impaired on recognition. Meaning that of the 12 that were impaired on recognition, 7 participants were impaired on delayed recall whereas 5 participants were not impaired on delayed recall. The majority of participants had HCV, a history of major depressive disorder, and a lifetime history of a substance use disorder , cocaine , cannabis , opioid , and methamphetamine. Ten participants met criteria for current major depressive disorder and four met criteria for a current substance use disorder . Approximately one quarter of participants had at least one APOE ε4 allele.
Post hoc sensitivity analyses excluding participants who were excluded in post hoc analyses in aim 1 as well as those not consistently on ART and participants with a substance use disorder other than Alcohol Use Disorder and Cannabis Use Disorder were conducted . The ICC for the delayed recall derived from the intercept-only model equaled 0.67, indicating that 67% of the variability in the delayed recall scores could be attributed to the between-person differences in average delayed recall performance. In the random slopes and random intercept model with no additional covariates, the average slope was -0.154, indicating that, on average, the delayed recall T-score decreased by 0.154 every year, and the standard deviation of the slope was 0.802. None of the cross-level interactions were significant, indicating that in this smaller group of participants that baseline hippocampal volume , entorhinal cortex thickness , and parahippocampal cortex thickness did not moderate delayed recall performance over time. A single-mediator model that accounted for relevant covariates was used. No significant direct effects from any plasma biomarkers of inflammation to delayed recall were found . None of the plasma biomarkers of inflammation were significantly associated with any of the medial temporal lobe structures , and none of the medial temporal lobe structures were significantly associated with delayed recall . When examining the indirect effect, all bias-corrected 95% CI contained 0; therefore, no meditated effect was established. See Table 13 for direct and indirect effects. Post hoc analyses excluding participants who were not on ART, with detectable viral load, and those with methamphetamine were examined.
Accounting for covariates, no significant direct effects from any plasma biomarkers of inflammation to delayed recall were found . Greater plasma CRP was associated with a thinner parahippocampal gyrus , but no other plasma biomarkers of inflammation were significantly associated with any of the medial temporal lobe structures . As demonstrated in aim 1, a thinner entorhinal cortex was associated with better delayed recall performance , but no other medial temporal lobe structures were significantly associated with delayed recall . When examining the indirect effects, all bias-corrected 95% CI contained 0; therefore, no meditated effect was established.This study closely examined episodic memory and the biological correlates of episodic memory. This is one of the first studies to examine the relationship between episodic memory, particularly recognition, and brain integrity both crosssectionally and longitudinally in middle-aged PWH without significant confounding comorbid conditions. This study had the potential to help validate recognition as a clinical marker that could aid in distinguishing aMCI and an AD trajectory from HAND, which could have led to diagnostic improvements in disentangling HAND and aMCI. While this study did not support that recognition is associated with more AD-related markers , it did demonstrate that episodic memory in middle-aged PWH is possibly related to frontally mediated etiologies, such as HIV, and that there was little decline in episodic memory in this group. Moreover, this is one of the first to examine the role of peripheral inflammation and its association with brain integrity and episodic memory in middle-aged and older PWH, although there were few associations with peripheral inflammation. While many of the findings of this study were not in line with the hypotheses, these findings are still clinically important and help to address gaps in our understanding of the biological associations of episodic memory in middle aged PWH. Broadly, the first aim of this study was to examine how recognition and delayed recall were associated with the medial temporal lobe, which is more implicated in preclinical AD, and the basal ganglia and prefrontal cortex,dry racks for weed which are more associated with HAND. Aim 1a examined the relationship between recognition and these three regions of interest. It was hypothesized that recognition memory would be more strongly related to medial temporal lobe structures given that recognition memory deficits are more associated with AD. Because recognition scores were skewed, as is common given that most people attain a near-perfect score with low variability in scores, this variable was dichotomized as examining it continuously violated multi-variable linear regression assumptions. In this sample of 92 participants, only 12 of participants were impaired on recognition. As discussed later, this sample of PWH may not be generalizable and thus the group that was impaired on recognition is both small and may also not be generalizable to other groups; additionally, recognition impairment at baseline may not be reliable given the instability of the recognition impairment across visits. Contrary to the hypothesis, medial temporal lobe structures were not significantly associated with odds of being impaired on recognition.
Given the limited number of participants that were impaired on recognition, there may not have been enough power to detect an effect; however, the odds ratios were fairly close to 1 indicating the association was neither statistically nor clinically significant. Also contrary to the aim 1a hypothesis, a thinner pars opercularis, part of the prefrontal cortex, was significantly associated with greater odds of being impaired on recognition. No other prefrontal regions or basal ganglia regions were significantly associated with odds of being impaired on recognition. Delayed recall was hypothesized to be more equally associated with all three regions, given that delayed recall deficits are observed in both aMCI/AD and HAND. Somewhat consistent with the hypothesis, thicker rostral middle frontal gyrus and pars opercularis were associated with better delayed recall. Examining laterality, these findings were somewhat more driven by the right. Additionally, thicker right pars triangularis was significantly associated with better delayed recall whereas the left pars triangularis was not. Contrary to the hypothesis, delayed recall was not significantly associated with the medial temporal lobe nor the basal ganglia. In post hoc analyses that excluded participants not on ART, or those with a detectable viral load or methamphetamine use disorder – a group of participants who are closer to those who are ideally treated in medical care – these associations held and thicker rostral middle frontal gyrus and pars opercularis were associated with better delayed recall and relationships were somewhat stronger within this subset of participants. It is important to note that given that delayed recall was examined continuously, this does not imply that these prefrontal regions are associated with delayed recall impairment, as that was not examined. Moreover, mean cortical thickness was included in the models as a covariate, so this means that this association is observed while accounting for average cortical thickness. Taken together, the finding that episodic memory was associated with some prefrontal structures may suggest that, at least in middle age, episodic memory performance is more likely related to frontally mediated etiologies, such as HIV, rather than early AD pathology. The inferior frontal gyrus, which includes the pars opercularis, pars triangularis, and pars orbitalis, as well as the middle frontal gyrus are not part of the medial limbic circuit implicated in memory formation, but they still contribute to memory deficits. The prefrontal cortex is of course associated with memory retrieval . Additionally, more recent models of memory formation stress the importance of the prefrontal cortex in memory formation given that there is some research to suggest that the prefrontal cortex aids in enabling long-term memory formation through connections with the anterior thalamic nuclei . Additionally, these more updated models of memory formation could account for why recognition was associated with prefrontal structures as well, although there could be several other explanations for this observed association. For example, recognition may also be associated with prefrontal structures due to poor initial encoding, which was not explicitly examined in these analyses. Nevertheless, functional MRI studies have shown alterations in prefrontal and hippocampal regions during memory tasks in PWH compared to controls further highlighting that prefrontal regions are implicated in memory in PWH . As highlighted in the introduction, HIV studies have found structural changes throughout the brain, including frontal regions, as compared to persons without HIV . Additionally, studies have demonstrated accelerated age-related atrophy or greater than expected “brain age” in middle-aged and older PWH compared to HIV-negative participants . For example, Milanini et al., 2019 found that, in a group of 19 participants with HAND who were on average 64 years old, HAND individuals showed faster atrophy in the cerebellum and frontal gray matter compared to HIV-negative controls. Additionally, Pfefferbaum et al., 2014 found accelerated changes in the frontal lobe, temporal pole, parietal lobe, and the thalamus in PWH compared to HIV-negative controls. Of these studies examining longitudinal brain changes, all found some involvement of the frontal lobe, but most studies did not examine the specific regions within frontal lobe that were driving these associations. Additionally, results from these studies were mixed as to if brain changes were associated with changes in cognition. Given that the current study only examines structural MRI at one time point, we cannot assume that there has been atrophy of the prefrontalcortex; however, given the literature demonstrates atrophic changes in PWH in the frontal lobe and accelerated aging in the frontal lobe, is possible that changes in the prefrontal cortex have occurred in this cohort and are contributing to the observed associations with memory.