This was a white participant in their 80s with HIV with a high school diploma. At the time of clinical evaluation, estimated duration of HIV disease was 18 years, nadir CD4 was 37, current CD4 653, and plasma HIV RNA was undetectable on dolutegravir, the nucleoside reverse transcriptase inhibitor (NRTI) abacavir and lamivudine. The participant took ARVs that included at least one NRTI for 16 years. With the exception of hyperlipidemia, osteopenia and a malignancy (not specified) in remission, they reported no major vascular, pulmonary, hepatic or other comorbidities. Concomitant medications were alendronate, aspirin and atorvastatin. They had no history of substance use disorders and reported no family history of AD.

The participant underwent three neuropsychological evaluations within a two-month period before their death eight months later. Across all evaluations, they reported dependence on others to accomplish important activities of daily living. They denied significant depression but reported fatigue and generalized weakness (Profile of Mood States vigor/activation subscale). Consistently across all evaluations, their performance indicated mild global neurocognitive impairment. Domain-specific neurocognitive performance demonstrated mild-to-moderate verbal impairment, mild executive function impairment and normal motor function. Speed of information processing performance vacillated between borderline impaired (1st evaluation), normal/average (2nd evaluation), and below average (3rd evaluation). Working memory performance was in the borderline impaired range in the first two evaluations, but normal/average at the third evaluation. Even within this short time span, deterioration was observed in memory function. Their first and second evaluations indicated below average learning and recall performance, whereas their third and final evaluations indicated borderline impaired learning performance and mildly impaired recall performance. In parallel, using an amnestic mild cognitive impairment (aMCI) diagnostic method that emphasizes impairments in recognition, the participant’s memory performance was classified as low aMCI risk at their first and second evaluations, but high aMCI risk at the third evaluation (Sundermann et al. 2020). Small declines (~ 2 T-score units) in both learning and recall performance from the first two evaluations to the third contributed to the overall deterioration in memory performance. In contrast, recognition performance was consistent across evaluations with verbal recognition in the impaired (~ 2 SD below the normative mean) range and visual recognition in the normal/below average range. The Global Deficit Score was 0.76, indicating mild-moderate overall impairment, with domain deficit scores indicating moderate impairment in learning, recall and verbal fluency (Table 1).

Neuropathology for Case #1. Gross examination of the brain at autopsy showed atherosclerosis of the right middle cerebral artery with 80% stenosis. Microscopic examination of H&E slides showed neocortical neuritic senile plaques, amyloid angiopathy, arteriosclerosis and arteriolosclerosis in subcortical white matter, and vascular mineralization in left globus pallidus. Only a few CA1 neurons displayed granulovacuolar degeneration. Immunohistochemistry showed amyloid deposition predominantly as diffuse plaques with occasional consolidated plaques, and in a few instances intraneuronal staining (Fig. 1a, b).

Immunostaining for amyloid and pTau: Frequent diffuse (a) and occasional senile plaques (b) were found in the frontal cortex and amygdala for Case #1. Diffuse amyloid plaques were found in in the frontal cortex (c), intraneuronal amyloid was found in the superior temporal cortex (d), and pTau neurofibrillary tangles and threads were found in the hippocampus for Case #2. Original magnification 10X; counterstaining with hematoxylin

This was a white participant in their 60s with HIV who had obtained a bachelor’s degree and worked in a high skill level position according to the International Standard Classification of Occupations (ISCO). The estimated duration of HIV disease was 20 years. The nadir CD4 was 143, current CD4 411, plasma viral load 981 copies/mL on atazanavir, tenofovir and emtricitabine. They had taken NRTIs for 17 years. Comorbidities were osteoarthritis, osteoporosis, hypothyroidism and hyperlipidemia. The participant had received appropriate treatment for primary syphilis in 2008 and reported a history of optic neuritis of undetermined etiology for many years. Concomitant medications were atorvastatin, escitalopram, prednisone, gabapentin, tamsulosin and levothyroxine. They had required use of a walker and had clumsy hands for many years. They endorsed mild-to-moderate depressed mood, severe fatigue and general confusion. They reported that their mother and maternal grandmother had AD.

The participant completed four neurocognitive evaluations in a two-year period (2003–2005) before death in 6/2006. Across all evaluations, they reported dependence on others to accomplish important daily activities. Global neurocognitive performance was deemed moderately impaired in the first three evaluations and deteriorated to severe impairment in the final evaluation. Verbal and working memory performance also showed decline over time from average or below average at the first and second evaluations to mild impairment at the 3rd and moderate impairment at the final evaluation. Speed of information processing was borderline impaired at the initial evaluation and then also showed deterioration to mild-to-moderate or moderate impairment at the 2nd and 3rd evaluations, culminating in severe impairment at the final evaluation. Executive function was the most consistently impaired domain throughout follow-up with moderate or moderate-to-severe impairment at the 1st through 3rd evaluations and deteriorating even further to severe impairment at the final evaluation. Motor performance vacillated within the range of mild-to-moderate to moderate-to-severe in the first three evaluations and deteriorated to severe impairment in the final evaluation. Again, using a method emphasizing recognition memory impairment, the participant was classified as high aMCI risk at all evaluations. Whereas learning performance showed a gradual decline from mild impairment (1st evaluation) to mild/moderate (2nd evaluation) to moderate impairment (3rd and 4th evaluations), recall performance showed a steeper decline from borderline impaired at the 1st evaluation to moderate-to-severe impairment in the 2nd evaluation and 4th evaluations with a brief improvement to mild impairment at the 3rd evaluation. In contrast, recognition performance showed stable impaired performance (~ 2 to 3 SD below the normative mean) across all evaluations (Table 1). They died in June 2006.

Neuropathology for Case #2. Microscopic examination of the H&E slides showed scattered diffuse plaques. Neurofibrillary degeneration of mild degree was noted within the hippocampus and amygdala. Immunohistochemistry examination showed abundant amyloid deposition as diffuse plaques in the frontal cortex (Fig. 1c), with occasional intraneuronal localization in the superior temporal cortex (Fig. 1d), and prominent neurofibrillary tangles in the hippocampus (Fig. 1e), immunohistochemistry for HIV gp41 envelope protein was negative.

This was a non-white participant in their 70s with HIV who had completed 18 years of formal education. At the time of clinical presentation, they had lived with HIV for 34 years. Their nadir CD4 + lymphocyte count was 259 cells/uL, and the current CD4 was 409 with an undetectable plasma viral load on dolutegravir, emtricitabine and tenofovir alafenamide. They had taken nucleoside NRTIs continuously for 12 years. The participant had a history of hypertension, hyperlipidemia and Type II diabetes mellitus, did not smoke tobacco and denied use of alcohol or illicit drugs with no history of substance use disorders. Concomitant medications included amlodipine, aspirin, benazepril, ezetimibe, metformin, and sitagliptin. They denied significant mood symptoms and reported no family history of AD.

In May 2016, the participant presented to the Neurology clinic complaining of transient lapses of attention without functional impairment. Mini-Mental State Examination (MMSE) score was 29/30. A brain MRI showed extensive periventricular T2 white matter hyperintensities and small bilateral basal ganglia lacunes. A carotid ultrasound demonstrated atherosclerosis, but no hemodynamically significant stenosis. A working diagnosis of mild cognitive impairment due to cerebral small vessel disease was made. In September 2016, they reported worsening memory difficulties without functional impairment, and was started on rivastigmine. By October 2016, the MMSE had dropped to 27. They were then referred to a research study for neurocognitive testing which showed high premorbid intellectual ability. The participant was classified as high aMCI risk based on impaired memory recall and recognition performance and reported dependence on others to accomplish important daily activities. At subsequent evaluations they denied dependence in activities of daily living, but we believe that this is likely due to lack of insight into their own neurocognitive deficits, since informant history indicated increasing dependence in all instrumental activities of daily living. By July 2017, the score on the Montreal Cognitive Assessment (MOCA) was 24/30. In May 2018 the patient’s family members expressed serious concerns about the participant’s memory problems and safety issues.

Plot of phospho-Tau versus Aβ42/total tau index for Case #3 showing values falling within the expected quadrant for Alzheimer’s disease (AD)

The MOCA score had dropped to 17, and a diagnosis of probable Alzheimer’s Disease with comorbid cerebral small vessel disease was made. Repeat comprehensive neuropsychological testing showed overall moderate global impairment; the greatest deficits were in speed of information processing, verbal fluency, verbal learning and delayed recall, and executive functioning. Despite the concerns of the participant’s family members, they were still driving and handling finances. Repeat neurocognitive testing showed global moderate-to-severe impairment. CSF showed Aβ42 406.65 pg/ml (reference range > 500 ng/L), T-Tau 531.1 pg/mL (reference range < 350 pg/ml) and phosphorylated Tau181 84.2pg/mL (reference range < 61) with an ATI > 0.8, consistent with AD (Fig. 2).

The patient completed four annual neurocognitive evaluations in the early 1990s when they were in their late forties, and then again between 2016 and 2019 when they were in their 70s. Their global neurocognitive performance during the four evaluations in the early 1990s fluctuated between average (1991), borderline impaired (1990 and 1996) and mild-to-moderate impairment (1995). When some degree of global impairment was evident, it was driven mostly by mild-to-moderate impairment in working memory, speed of information processing and executive function. In contrast, verbal skills, learning, recall and motor performance were classified as normal throughout the early evaluations except for borderline impaired performance at their last evaluation (Table 1).

Semiquantitative amyloid stain rating according to ApoE genotype. Increasing E4 alleles were associated with higher amyloid ratings

The larger autopsy cohort comprised 250 PWH brains from PWH that underwent neuropathological characterization of frontal lobe Aβ and ApoE genotyping as part of a prior study (Umlauf et al. 2019). The mean (SD) age at death was 47.4 (9.22) years; 17.6% were female; 21.6% were Black, 22.8% Hispanic, 52.0% Non-Hispanic White, and 3.60% other race/ethnicities; mean (SD) education 12.5 (2.87) years. 169 (67.6%) had semiquantitative amyloid ratings of 0, sixty (24%) of 1 (focal/occasional), and 21 (8.4%) of 2 (widespread/abundant). The distribution of ApoE genotypes was as follows: E2/E2 1 (0.40%), E2/E3 25 (10.0%), E2/E4 9 (3.60%), E3/E3 157 (62.8%), E3/E4 54 (21.6%), E4/E4 4 (1.60%). As seen in Fig. 3, higher amyloid ratings were associated with the presence of the ApoE E4 allele, the strongest genetic risk factor for sporadic AD (Pearson χ2 = 39.7, p = 1.88e-5). Poorer ante-mortem neurocognitive performance was not associated with higher amyloid burden (p = 0.580); however, the probability of HAND was increased in the presence of Aβ plaques when limiting to APOE E4 carriers (OR = 30.00, 95% CI = 1.41-638.63, p = 0.03) (Umlauf et al. 2019). This suggests that, in the context of higher AD genetic risk, amyloid burden has greater clinical significance among PWH. The amyloid ratings in those who took NRTIs as part of their antiretroviral regimens (n = 160, 58.6%; amyloid rating 0, 69.4%; amyloid rating 1, 20%; amyloid rating 2, 10.6%) did not differ from those who did not take NRTIs (amyloid rating 0, 72.6%; amyloid rating 1, 23.9%; amyloid rating 2, 3.54%; p = 0.178).