In the entire sample, none of the plasma biomarkers of inflammation were significantly associated with any of the medial temporal lobe structures, there were no significant direct effects between the plasma biomarkers of inflammation and delayed recall, and no mediated effect was established. As stated above, the lack of association between inflammation and delayed recall is a little surprising given that the association between inflammation and worse cognition has been demonstrated in HAND and the aging literature. Although, the effect sizes are often small, and the middle-aging literature is limited. Additionally, some of the peripheral inflammatory markers examined in this study have been associated with medial temporal lobe integrity and function in older adults , but the association between inflammation and the medial temporal lobe is much less studied in mid-life and PWH. Marsland et al., 2015 did find that IL-6 and CRP were associated with worse memory and smaller hippo campal volumes in middle-aged adults; however, it was cortical grey matter volume, not the hippocampus, that mediated the relationship between inflammation and memory. Studies in adults with HIV have found that peripheral biomarkers of immune activation but not biomarkers examined in this study were associated with frontal and temporal lobe regions . Interestingly, in post hoc analyses examining participants on ART who were virally suppressed, greater CRP was associated with a thinner parahippocampal gyrus. This finding may be in line with the Marsland et al., 2015 study. However, the current study had a small sample size, and several analyses were examined in post hoc analyses without accounting for multiple comparisons, so this finding should be interpreted cautiously. Integrating the aging and HIV literature, it is unclear if the association between peripheral inflammation,vertical grow manufacturer medial temporal lobe, and episodic memory is consistently observed in mid-life.
While the best method of determining the necessary sample size to detect a mediation effect is debated, it is still likely this study’s modest sample size of 92 is under powered to detect a mediation effect, particularly given that large effect sizes were not expected . Therefore, the role of inflammation and its association with brain integrity and episodic memory in PWH should continue to be examined, particularly in larger samples with greater power to detect these associations. It will be particularly important to examine these relationships in PWH aged 65 and over given that this is the age range in which these associations between inflammation, memory, and MCI/AD risk are more consistently found. One thing to note is that these peripheral inflammatory biomarkers were examined separately, as each biomarker may have a different relationship with memory and brain integrity. There is currently no “gold-standard” way to combine inflammation biomarkers into a single composite. However, some researchers have examined inflammation composites . Therefore, future studies may want to examine a wider array of biomarkers and employ an inflammation composite, particularly given that the impact of inflammation on brain integrity and memory may be due to the compounding effects of multiple inflammatory biomarkers. Additionally, these biomarkers were only examined at one time point, so a better understanding of how changes in these inflammatory biomarkers over time are associated with brain integrity and cognition is also needed. Lastly, this study examined peripheral inflammation. Peripheral inflammation is easier to assess more non-invasively in comparison to a lumbar puncture which is needed to collect CSF . However, peripheral inflammation may not be as reflective of neuroinflammation compared to CSF biomarkers. Although, some studies have shown that plasma inflammation may be more associated with cognition . Thus, future studies should ideally examine both plasma and CSF biomarkers to determine if examining peripheral inflammation is sufficient.
Ultimately, a better understanding of the role of inflammation and the most efficient way to measure it could help to inform interventions that could lower inflammation in PWH, if future research indicates that lowering inflammation may be cognitively beneficial. In addition to the limitations discussed above, there are additional limitations that should be considered. First, the generalizability of the sample should be considered. As noted several times in the discussion, the age range may be too young to expect a significant number of participants to have started to accumulate AD pathology. Additionally, the sample was predominantly male , which is somewhat reflective of the current demographics of PWH in the United States . Nevertheless, there are known sex differences in HIV, AD, and inflammation that this project is under powered to test but should be further examined in future studies. For example, women living with HIV are at greater risk of neurocognitive impairment, particularly in the domains of memory, speed of information processing, and motor function potentially due to a difference in psychosocial factors , comorbid conditions , and biological factors . It is also known that women are at greater risk of AD . Additionally, participants with severe confounding comorbid conditions were excluded from this study, and this sample was characterized by relatively low current drug use and relatively high ART use. These factors are also known to impact cognitive and brain functioning; for example, cannabis use has been associated with better cognitive functioning and lower inflammation in PWH . As the HIV population continues to age, it will be important to understand if there are any associations between these sociodemographic variables and AD risk that str specific to PWH. Related to generalizability, one odd finding was the higher-than-expected number of participants with the APOE e2 allele. The percentage of participants with at least one e4 allele was somewhat comparable to the general population, with estimates ranging from 10% to 25% of people having at least one e4 allele.
Additionally, it is known that Black/African American persons and persons of African ancestry have increased rates of the APOE e4 allele compared to non-Hispanic White people or those of European descent . Indeed, the CHARTER study has found an increased prevalence of the e4 allele in Black/African American participants as compared to nonHispanic White participants . The APOE e2 allele is much less studied because it is more rare, but having an APOE e2 allele is associated with a lower-than-average risk of AD. In this study, the percentage of participants with at least one APOE e2 allele was higher than the general population . Similar to the APOE e4 allele, the prevalence of APOE e2 is known to vary by ancestorial continent and latitude. The APOE e2 allele penetrance is 9.9% in Africa, which is higher than the APOE e2 allele penetrance in Europe . Even accounting for these demographic differences, the prevalence of the APOE e2 is high, and this over representation of the APOE e2 allele may mean this group is, on average, at decreased risk of AD. This increased prevalence could be due to a selection bias . Information on the APOE e2 in PWH is very limited, but more research is certainly needed to understand AD risk in diverse groups of PWH. One minor point is that four participants with the APOE e24 genotype were categorized as APOE e4-. The limited literature on this genotype does suggest a somewhat elevated risk of AD associated with this genotype,vertical grow marijuana system but much less than that of those that are APOE e34 or APOE e44 . Therefore, the APOE e24 participants were categorized as APOE e4- given the only slightly elevated risk. Other categorizations could be explored, although given the small number of participants that are APOE e24 it is unlikely to make a significant difference. In addition to the potentially limited generalizability due to the demographics and clinical characteristics of this sample, this study examined a relatively modest sample size. A sample size of 92 is not necessarily small compared to other imaging studies. However, as highlighted throughout this discussion, this modest sample size could still limit the power to detect associations. Future studies in this area would benefit from improving statistical power either by enrolling a larger overall sample and/or recruiting participants with memory impairment, particularly recognition impairment. This study is also limited in that it does not include an HIV-negative comparison group. Utilizing preexisting CHARTER data allowed for longitudinal analysis over 12 years and the ability to efficiently examine the neuroanatomical correlates of memory in middle-aged and older PWH. However, this study is therefore limited by pre-defined CHARTER protocol and design. Specifically, CHARTER did not enroll HIV-negative comparison participants, which precludes examination of how the relationship between memory profiles and brain integrity differ by HIV serostatus. While there is ample HIV-negative middle-aging literature to compare these results to, many of these HIV-negative middle-aging studies are demographically and psychosocially different than this group.
However, even with a good comparison group, it is difficult to discern the effect of HIV versus the neurotoxic effects of ART and the downstream consequences of ART . Nevertheless, future studies would benefit from a demographically and psychosocially similar HIV-negative group to better understand if the associations between memory and neuroimaging correlates are specific to PWH or if these are associations seen regardless of HIV status. In the current study, delayed recall and recognition were examined separately rather than dichotomously splitting participants into aMCI versus non-aMCI groups or comparing HAND versus aMCI groups as in Sundermann et al. . Examining recognition and delayed recall was a critical first step to inform future diagnostic improvements. Additionally, examining delayed recall continuously was advantageous because it increases variability and more subtle differences observed in mid-life may not be captured by diagnostic cut-points. However, associations between biological markers associated with AD have been found in PWH using aMCI criteria . Therefore, data could be reexamined using adapted aMCI criteria and HAND criteria to examine if a more comprehensive approach to examining episodic memory is more sensitive to the medial temporal lobe than examining delayed recall and recognition separately. As described in the Methods section, the differences in scanner by site was corrected by regressing scanner from the data. Accounting for scanner was necessary given that prior CHARTER studies have shown that pooling MRI data from multiple sites is feasible, but there are documented differences between the scanners . However, accounting for scanner is essentially accounting for study site, which is somewhat problematic given that study site has been shown to be associated with the risk of neurocognitive impairment in the CHARTER study. For example, Marquine et al. found a significant effect of study site, specifically when comparing New York and San Diego, on the risk of neurocognitive impairment that was not fully accounted for by race/ethnicity differences. It is thought that differences in the risk of neurocognitive impairment are likely due to psychosocial and environmental factors that are associated with geographic location . These psychosocial and environmental factors could also impact brain integrity, and thus accounting for scanner, while necessary, may mask real differences in brain integrity. Therefore, future studies may want to employ a different statistical method that could account for differences in scanner while not eliminating the effect of study site. Relatedly, future studies could explore alternative ways to analyze the imaging data. For example, a priori regions of interest were selected given the interest in focusing on brain structures associated with HAND and aMCI. However, the FreeSurfer processing approaches provide a broad array of additional regions that could also be explored. Furthermore, additional data-driven analytic approaches exist such as whole-brain voxel-based morphometry. This study took a hypothesis-driven approach, although examination of other regions of interest, such as subdivisions of the cingulate cortex, could be done in an exploratory fashion. Other imaging modalities such as diffusor tensor imaging to examine white matter integrity, arterial spin labeling to examine cerebral blood flow, MRS to examine neurochemical alterations, and amyloid PET imaging may also help to better understand episodic memory in PWH. Despite these limitations, this study has several clinical implications. This study showed that memory in these participants aged 45 to 68 was associated with prefrontal structures but not medial temporal lobe structures. This suggests that episodic memory in middle-aged PWH is more associated with frontally mediated etiologies such as HIV rather than etiologies associated with the medial temporal lobe such as AD. Second, recognition impairment was quite variable over time. Due to this variability over time, recognition may not serve as a good clinical marker to help distinguish aMCI from HAND. However, this group of participants is considerably younger than when late-onset AD presents; therefore, continued research is needed to examine if recognition may be a useful clinical marker to differentiate aMCI and HAND in older age.