Cerebral &bgr;-amyloid deposition predicts HIV-associated neurocognitive disorders in APOE &egr;4 carriers

Objective:The apolipoprotein E (APOE) &egr;4 allele enhances cerebral accumulation of &bgr;-amyloid (A&bgr;) and is a major risk factor for sporadic Alzheimer's disease. We hypothesized that HIV-associated neurocognitive disorders (HAND) would be associated with the APOE &egr;4 genotype and cerebral A&bgr; deposition. Design:Clinicopathological study of HIV-infected adults from four prospective cohorts in the US National NeuroAIDS Tissue Consortium. Methods:We used multivariable logistic regressions to model outcomes [A&bgr; plaques (immunohistochemistry) and HAND (standard criteria)] on predictors [APOE &egr;4 (allelic discrimination assay), older age (≥50 years), A&bgr; plaques, and their two-way interactions] and comorbid factors. Results:Isocortical A&bgr; deposits generally occurred as diffuse plaques and mild-to-moderate amyloid angiopathy. Isocortical phospho-Tau-immunoreactive neurofibrillary lesions were sparse. The APOE &egr;4 and older age were independently associated with the presence of A&bgr; plaques [adjusted odds ratio (OR) 10.16 and 5.77, 95% confidence interval (CI) 2.89 − 35.76 and 1.91−17.48, P = 0.0003 and 0.0019, respectively, n = 96]. The probability of HAND was increased in the presence of A&bgr; plaques among APOE &egr;4 carriers (adjusted OR 30.00, 95% CI 1.41−638.63, P = 0.029, n = 15), but not in non-&egr;4 carriers (n = 57). Conclusion:The APOE &egr;4 and older age increased the likelihood of cerebral A&bgr; plaque deposition in HIV-infected adults. Generally, A&bgr; plaques in HIV brains were immunohistologically different from those in symptomatic Alzheimer's disease brains. Nonetheless, A&bgr; plaques were associated with HAND among APOE &egr;4 carriers. The detection of APOE &egr;4 genotype and cerebral A&bgr; deposition biomarkers may be useful in identifying living HAND patients who could benefit from A&bgr;-targeted therapies.


Introduction
In the current era of HAART, HIV-associated neurocognitive disorders (HAND) continue to affect the clinical outcome of HIV infection [1,2]. Specifically, the milder forms of HAND, asymptomatic neurocognitive impairment (ANI) and mild neurocognitive disorder (MND), are more common than HIV-1-associated dementia (HAD). The differential susceptibility to HAND may be explained by individual differences in HIV variants, host genetic polymorphisms, and comorbid factors (e.g. aging, substance use, and adverse effects of HAART), which may interact with each other in contributing to neural injury [3]. Some of these factors may trigger or promote a cascade of metabolic disturbances, leading to neurodegeneration and thereby neurocognitive impairment. For instance, postmortem studies showed extracellular b-amyloid (Ab) deposition (as plaques) in the isocortex [4][5][6][7][8] and hippocampus [9] in subsets of HIV-infected adults.
The disturbance in cerebral Ab metabolism may be one of the potential pathophysiologic pathways leading to HAND. Although no systematic correlative analyses between cerebral Ab deposition and neurocognitive impairment were available in previous autopsy studies [5,8,9], a clinical study by Clifford et al. [10], in agreement with a report by Brew et al. [11], showed that Ab42 levels in the cerebrospinal fluid (CSF) were decreased in HAND patients similar to the levels in patients with mild Alzheimer-type dementia, when compared with those in cognitively normal patients. The decrease in CSF Ab42 levels reflects generally the presence of cerebral Ab deposition detected by [ 11 C] Pittsburgh compound B (PiB) PET [12,13]. However, the findings in those CSF studies were not confirmed in a similar study by Gisslen et al. [14], which might be explained by between-study differences in patients' age and antiretroviral therapy.
The apolipoprotein E (APOE) e4 allele correlates with the earlier onset and greater extent of cerebral Ab accumulation [15,16] and is a major risk factor for sporadic Alzheimer's disease and cerebral amyloid angiopathy (CAA) [17]. The major codominant alleles (i.e., e2, e3, and e4) in the human APOE gene are associated with differential biological activities of their protein products [18]. The APOE is an Ab-binding molecule that may influence the clearance of soluble Ab at the blood-brain barrier and affect Ab seeding and aggregation [18][19][20]. Across several studies in HIVinfected adults, it remains controversial as to whether the APOE e4 increases the susceptibility to HAND [21][22][23][24][25][26].
The present study was aimed at exploring the influence of APOE e4 on cerebral Ab deposition in HIV-infected adults and studying their significance in contributing to HAND. We followed a clinicopathological correlative approach in studying HIV-infected adults who received detailed clinical, neuropsychological, and laboratory assessments as part of the National NeuroAIDS Tissue Consortium (NNTC). We hypothesized that HAND would be associated with the APOE e4 genotype and cerebral Ab deposition. If so, the detection of APOE e4 and brain Ab deposition may be useful in identifying HAND patients who could benefit from Ab-targeted therapies.

Study cohort
We assembled 160 HIV cases in total (age range 27-67 years) autopsied during 1999-2010. Frozen tissues were available for APOE genotyping in 151 cases and formalin-fixed middle-frontal sections for immunohistochemistry in 105 cases. These brains were obtained from HIV patients who participated in neuropsychological testing at a median of 20.7 weeks before death [interquartile range (IQR) 37.7 weeks]. Seven domains of neurocognitive functioning were assessed: information processing speed, attention/working memory, learning, recall memory, verbal fluency, abstraction/executive functioning, and motor/psychomotor speed, with statistical correction for demographic variables (i.e. age, sex, ethnicity, and education), as described previously [27]. Based on standard criteria [28], HIV-associated neurocognitive diagnoses were made, including normal cognition (n ¼ 32), ANI (n ¼ 19), MND (n ¼ 37), and HAD (n ¼ 22). There were 47 individuals affected by neuropsychological impairment due to other or undetermined causes, and three individuals whose diagnoses were inconclusive; these cases were excluded from the statistical analysis regarding HAND.
Histories of antiretroviral treatment available in 101 HIV patients were recorded within a median of 17.6 weeks (IQR 32.3 weeks) before death and grouped into 'no treatment' (n ¼ 31), 'non-HAART regimens' (n ¼ 6), and 'HAART regimens' (n ¼ 64). The antiretroviral regimens and their durations varied markedly among HIV patients. Hepatitis C virus (HCV) infection was present in 47 (37.6%) of 125 HIV patients having serological testing. We used either Psychiatric Research Interview for Substance and Mental Disorders [29] or Composite International Diagnostic Interview [30] to ascertain lifetime substance use disorders based on the Diagnostic and Statistical Manual of Mental Disorders (fourth edition). Of 122 HIV patients evaluated for methamphetamine use, 46 (37.7%) were recorded as having lifetime methamphetamine use (combining abuse and dependence categories); at the final premortem visits, only two of these 46 had current dependence and none had current abuse. Of 121 HIV patients evaluated for major depressive disorder (MDD), 56 (46.3%) were recorded as having lifetime MDD; 12 of these 56 had current MDD at the final premortem visits. Because of the high prevalence of comorbid factors described above, we included them as covariates in the statistical analysis.
Systemic autopsy findings were commonly diagnostic of AIDS; other frequent findings included hepatic cirrhosis and bronchopneumonia. Of 160 HIV brains, 44 were normal, 24 had minimal histopathologic changes, 18 with Alzheimer type II gliosis, 29 with vascular pathology (e.g. hypoxic-ischemic changes, infarcts, and hemorrhages), 16 with HIV encephalitis, eight with leukoencephalopathy, 10 with microglial nodules, 19 with one or more opportunistic infections (e.g. cytomegalovirus encephalitis, toxoplasmosis, cryptococcosis, and progressive multifocal leukoencephalopathy), and 10 focally involved by primary or secondary non-Hodgkin's lymphoma. Of 105 HIV brains available for immunohistochemistry, only one showed HIV encephalitis.
Non-HIV controls (n ¼ 22, age range 24-90 years) with no clinical history of neurological diseases were autopsied during 1992-2009. The systemic autopsy findings included organ transplantation, hepatic cirrhosis, lymphomas, and cardiovascular diseases. The neuropathologic diagnosis was either normal or minimal histopathologic changes. The formalin-fixed isocortex sections were available for immunohistochemistry.

Apolipoprotein E genotyping
Tissue samples obtained at autopsy were stored at À808C until the time of total DNA extraction using DNeasy Blood & Tissue Kit (Qiagen, Germantown, Maryland, USA). The amount of genomic DNA was quantified by using NanoDrop Spectrophotometer (NanoDrop Technologies, Wilmington, Delaware, USA). For APOE genotyping, we used the allelic discrimination assay (Taqman SNP Genotyping Assays; Applied Biosystems, Carlsbad, California, USA) according to the manufacturer's instructions. The allele calls and genotypes of samples were determined by using the Taqman Genotyper software.

Light microscopy
The presence of Ab plaques or CAA was confirmed when these lesions were found in any of Ab-4G8, Ab40, and Ab42 immunostained slides due to the fact that cerebral Ab deposits characteristically exhibit an uneven multifocal distribution [32]. The density of Ab plaques was graded as focal and widespread. CAA was qualitatively graded according to Vonsattel criteria [33] as mild, moderate, and severe. The density of p-Tau-immunoreactive neurites was graded as 1 (barely present at Â 100 magnification), 2 (easily noted at Â 100 magnification), and 3 (notable with naked eye inspection), a scoring system adapted from a BrainNet Europe Consortium study [34].

Statistical analysis
We used multivariable logistic regressions to model outcomes [i.e., cerebral Ab plaques and HAND (vs. normal cognition)] on predictors [i.e., APOE e4, older age (!50 years), Ab plaques, and their two-way interactions] and each of four covariates (i.e. antiretroviral treatment, HCV infection, methamphetamine use, and MDD). The statistical analyses were performed using R (version 2.10.0, 2009, http://www.r-project.org). All two-tailed P values were considered significant at a threshold of P < 0.05.

Cohort characteristics
Between HIVand non-HIV control groups, there was no significant difference in age (median 46 and 50 years, IQR 14 and 22.3 years, n ¼ 160 and 22, respectively; P ¼ 0.42, Mann-Whitney U test) or postmortem interval (median 12 and 15 h, IQR 13.4 and 13 h, n ¼ 158 and 21, respectively; P ¼ 0.33, U test). The proportion of women in the HIV group (n ¼ 19 of 160) was lower than in the control group (n ¼ 10 of 22; P ¼ 0.0004, Fisher's exact test).

Cerebral Ab deposition
In both HIV and control groups, parenchymal Ab deposits were found in most instances as diffuse plaques in the cortical gray matter, often exhibiting perivascular and perineuronal accumulation ( Fig. 1a and b). Cored Ab plaques were seen in only one HIV case and two controls, in concurrence with diffuse Ab plaques. The prevalence of Ab plaques was 29.5% of 105 HIV cases (focal 24, widespread 7) and 22.7% of 22 controls (focal 2, widespread 3). CAA was found in 6.7% (mild 3, moderate 3, severe 1) of 105 HIV cases and in 4.5% (moderate 1) of 22 controls, always together with Ab plaques. Intracellular Ab immunoreactivity in neuronal soma was focally observed (more often on Ab-4G8 than Ab40 or Ab42 staining) in subsets of HIV cases and controls, without regard to the presence of Ab plaques in the same section.
Apolipoprotein E e4 and older age independently predicted cerebral Ab plaque deposition In univariable logistic regression, the presence of Ab plaques was significantly associated with the APOE e4 and older age (P < 0.001 and ¼ 0.016, respectively), but not with each of the four covariates (i.e. antiretroviral treatment, HCV infection, methamphetamine use, and MDD) (P > 0.15). In multivariable logistic regression (Model: Ab ¼ APOE e4 þ older age þ covariate), the APOE e4 and older age remained significant independent predictors for Ab plaques after adjusting for each of the four covariates (P < 0.05). In contrast, none of these covariates showed significant association with Ab plaques after adjusting for the APOE e4 and older age (P > 0. 16). Accordingly, all the covariates were excluded. In Model (n ¼ 96): Ab ¼ APOE e4 þ older age, the APOE e4 predicted Ab plaques (adjusted OR 10.16, 95% CI 2.89À35.76, P ¼ 0.0003), as did older age (adjusted OR 5.77, 95% CI 1.91À17.48, P ¼ 0.0019). The interaction effect of APOE e4 and older age on the presence of Ab plaques was not statistically significant (P ¼ 0.97) (Fig. 2). Furthermore, the APOE e4 was significantly associated with the abundance of Ab plaques (none, focal, widespread) on multinomial logistic regression (overall P ¼ 0.002, n ¼ 96). That is, the odds of having focal Ab plaques (relative to none) was higher among APOE e4 carriers compared with none4 carriers (OR 3.35, 95% CI 1.22À9.19, P ¼ 0.019), as was the odds of having widespread Ab plaques (relative to none) (OR 11.73, 95% CI 2.05À67.20, P ¼ 0.006).
Interaction effect of apolipoprotein E e4 and cerebral Ab plaque deposition on HIV-associated neurocognitive disorders Univariable logistic regression showed no significant association between HAND and each of demographic and biologically relevant variables (P > 0.09) ( Table 1). We used multivariable logistic regression to explore the effects of APOE e4, Ab plaques, older age, and their twoway interactions on HAND. The model selection process was pursued according to the Akaike Information Criteria (a measure of the relative goodness of fit of a statistical model) to include only those variables and interactions that provided the most accurate prediction of HAND. The interaction effects of older age and APOE e4 or Ab plaques, as well as the main effect of older age, on HAND were not statistically significant.

Potential effects of comorbid factors on HIV-associated neurocognitive disorders
We further investigated whether the interaction effect of APOE e4 and Ab plaques on HAND remained after adjusting for older age and each of the four covariates. Age remained irrelevant in all of these models. In Model:  of MDD (P ¼ 0.28 and 0.95, respectively). The issue of multicollinearity was of trivial concern because there was no significant association between methamphetamine use (or MDD) and APOE e4 (or Ab plaques) (P > 0.6, all x 2 tests).

Discussion
Generally, Ab plaques first appear in the isocortex and then expand with increasing age hierarchically into further brain regions, representing different phases of Ab deposition [35]. The middle-frontal gyrus is one of the isocortex regions having relatively high Ab plaque density [36]. Accordingly, we chose to examine this brain region for Ab plaques. We found that cerebral Ab deposits both in HIV cases and non-HIV controls occurred mostly as diffuse plaques and were rarely associated with p-Tau-immunoreactive neurofibrillary lesions. These findings agree with those in previous studies of HIV brains [4][5][6]. Diffuse Ab plaques likely represent the earliest stage of temporal progression of Ab plaques [37], in contrast to neuritic cored Ab plaques characteristically present in symptomatic Alzheimer's disease brains.
Regarding the APOE genotypic distribution, our HIV case series appeared to represent the general population. We found the APOE e4 and older age were independently associated with the presence of cerebral Ab plaques after adjusting for each comorbid factor. Furthermore, the APOE e4 was associated with the abundance of cerebral Ab plaques. These findings in HIV patients concur with those in the general population [15,16,38].
Notably, we found an interaction effect of the APOE e4 and cerebral Ab plaques on HAND. That is, the presence of Ab plaques was associated with HAND among APOE e4 carriers, but not in non-e4 carriers. Our finding suggests APOE isoforms differentially modulate the association between cerebral Ab plaques and HAND. Indeed this concurs with a clinical study in the older population by Kantarci et al. [39] showing that higher brain Ab loads detected by PiB PET correlated with poorer cognitive performance among APOE e4 carriers.
In a small study by Ances et al. [40] with the assessment of cortical PiB retention, none of five HAND and 11 cognitively normal HIV patients had increased PiB retention in contrast to symptomatic Alzheimer's disease patients. On the contrary, CSF Ab42 levels were decreased (<500 pg/ml cutoff value) in two of four HAND and three of eight cognitively normal HIV patients, but in only one of eight non-HIV controls apparently matched for the APOE e4 status and age. Previous clinical studies by Clifford et al. [10] and Brew et al. [11] also showed that CSF Ab42 levels were reduced 2332 AIDS 2012, Vol 26 No 18 in HAND patients compared with those in cognitively normal participants. Decreases in CSF Ab42 levels correlate generally with increases in cortical PiB retention indicating the presence of cerebral Ab deposition [12,13]; however, the CSF changes begin at earlier ages than changes in cortical PiB retention [16,41]. As PiB (a derivative of thioflavin-T [42]) binds to b-pleated sheet aggregates of peptides (i.e., amyloid, including fibrillar Ab), PiB readily marks cored Ab plaques (whether or not they are neuritic) [41,43]. Due to its higher affinity to fibrillar Ab compared with the affinity of thioflavin-S [44], PiB also marks diffuse Ab plaques [41,43], characteristically composed of small amounts of fibrillar Ab [45]. Taken together, it is likely that HIV patients with reduced CSF Ab42 levels have cerebral Ab deposition, which (depending on the fibrillar Ab load) may or may not be detected by PiB PET [16,41]. Accordingly, measurement of CSF Ab42 levels may be more sensitive than PiB PET for the detection of cerebral Ab deposition in HIV-infected adults. PiB PET may be useful in the event that the cerebral Ab load is high as is seen with the presence of widespread Ab plaques (found in 6.7% of 105 HIV cases in our study).
Although APOE e4 correlates with the earlier onset and greater extent of cerebral Ab accumulation [15,16], it is not an indispensable factor for cerebral Ab deposition. Progressive Ab accumulation may be caused by increased Ab production by neurons, increased influx of Ab from the circulation, decreased enzymatic degradation of Ab, and defective efflux of soluble Ab from the interstitial fluid (ISF) [46]. In addition to receptor-mediated transcytosis of Ab across the blood-brain barrier, Ab elimination may be mediated by perivascular macrophages [47], via bulk flow of ISF into the ventricles [48], and through perivascular ISF drainage along the basement membranes of capillaries and arteries [49,50]. APOE isoforms may differentially regulate the clearance of soluble Ab at the blood-brain barrier and the propensity for Ab to aggregate [18][19][20]. In addition to its enhancing effect on cerebral Ab accumulation, the APOE-4 isoform may potentiate the effect of Ab plaques on the neurodegenerative process leading to HAND through other mechanisms yet to be determined.
Unexpectedly, we found that methamphetamine use and MDD were individually associated with the lower probability of HAND, after adjusting for the APOE e4 and cerebral Ab plaques. In our study, the majority (71.8%) of HAND cases were in milder forms (ANI and MND). Accordingly, neural injury in most HIV cases was probably not at irreversible stages, that is, the brain retained a degree of plasticity while being exposed to methamphetamine. Previous studies showed that methamphetamine (low dose) enhanced cognitive performance [51], especially in tasks that required long periods of sustained attention in individuals with relatively low (prefrontal cortex-dependent) working memory capacity at baseline [52]. The combined effects of HIV and methamphetamine in this context are of particular interest and can be investigated in future studies. Regarding the association between MDD and HAND, the HIV-infected patients affected by MDD might be treated with selective serotonin reuptake inhibitors, which were shown to correlate with reductions in cerebral Ab accumulation due to increased serotonin signaling [53]. However, we did not find any significant association between the presence of Ab plaques and MDD.
In conclusion, we investigated the influence of APOE e4 on cerebral Ab deposition in HIV-infected adults and their significance in contributing to HAND, by using clinical, laboratory, and postmortem tissue resources available from the NNTC. We found that APOE e4 and older age independently increased the likelihood of cerebral Ab plaque deposition. Although Ab plaques in HIV brains were immunohistologically similar to those in aging brains and different from those in symptomatic Alzheimer's disease brains, cerebral Ab deposition was associated with HAND among APOE e4 carriers after adjusting for each comorbid factor. Accordingly, the detection of APOE e4 and biomarkers of cerebral Ab deposition (e.g. decreases in CSF Ab42 levels) may be useful in identifying HAND patients who could benefit from Ab-targeted therapies. Still, future studies in the HIV-infected population are warranted to confirm the inverse relationship between CSF Ab42 levels and the abundance of cerebral Ab plaques. Based on our finding that isocortical p-Tau-immunoreactive neurofibrillary pathology was sparse in HIV patients, CSF p-Tau measurement may be useful in differentiating HAND from Alzheimer's disease and other tauopathies in older patients [10,54,55].