Plasma Biomarkers of Alzheimer Disease in Women With and Without HIV

This cohort study analyzes whether biomarkers for Alzheimer disease are associated with neuropsychological performance among women with and without HIV.


Introduction
Chronic HIV is a 21st century phenomenon due to effective antiretroviral therapies (ARTs) and increasing life expectancies among people living with HIV (PLWH) and fewer HIV-related deaths. 1 Still, HIV constitutes a public health concern with more than 38 million cases worldwide. 2As a multisystem disorder, HIV directly and indirectly affects the peripheral and central nervous systems.
PLWH may experience wide-ranging neurological disorders (eg, peripheral neuropathy and cognitive impairment) and neuropsychiatric conditions including depression. 3 Although the incidence of HIV-associated neurocognitive disorder has decreased in PLWH due to effective and quickly prescribed ARTs, 4 those surviving to older ages are at risk for Alzheimer disease (AD). 5 is a global health concern with more than 50 million people affected worldwide. 5In the US, more than 6 million people aged 65 years or older live with dementia.It is projected that this number will double by 2060 and comprise more than 3% of the US population. 5AD is the seventh leading cause of death in the US 5 and globally. 6Certain populations may be more susceptible to AD, notably, those with multiple comorbidities and African American or Black people. 5Besides reducing quality of life, AD has a major economic impact due to caregiving and prohibitive costs of symptom treatment, which has reached $60 billion. 5Although there is no cure for AD, early diagnosis is critical for appropriate health and care management and caregiver and family planning.
Approximately 50% of PLWH, whether treated with ARTs or not, experience a cognitive disorder. 7Women living with HIV (WLWH) and women living without HIV (WLWOH), may be at risk for more adverse brain health outcomes with aging due to higher levels of vascular risk factors, such as hypertension, obesity, and cigarette smoking, as well as substance abuse, lower education, lower income, and inadequate health care access. 5ood-based AD biomarkers reflecting amyloid, tau, and neurodegeneration (ATN) may be keys to early diagnosis and underlie the National Institute on Aging and Alzheimer's Association 2018 research framework 8 ; the framework mandates research based on ATN, with or without clinical manifestations, to better understand the mechanisms underlying AD, the temporal sequence of AD neuropathologies, and the temporality of factors associated with increased risk of AD.Thus, novel blood-based AD biomarkers include those reflecting amyloid deposition (eg, amyloid-β [Aβ] 40, Aβ42, and the Aβ42:Aβ40 ratio), tau pathophysiology (eg, total tau [t-tau] and phosphorylated tau231 [p-tau 231 ]), and neurodegeneration (eg, neurofilament light chain [NFL] and glial fibrillary acidic protein [GFAP]). 9These biomarkers will complement and/or eventually confirm a clinical AD diagnosis based on the ATN criteria 10 and contribute to better understanding of ATX(N), 11 where the X represents other pathological factors, such as neuroinflammation, immune function, vascular events, and HIV.Thus, identification of blood-based biomarkers of AD against a background of chronic infectious disease and vascular risk is paramount to understanding HIV and ARTs in relation to cognitive impairments and, eventually, late-onset AD. 12 Given the dearth of observational, longitudinal data on blood-based AD biomarkers in WLWH, we explored the hypothesis: is 1-year change in plasma AD biomarkers associated with neuropsychological performance (NP) among WLWH and WLWOH aged 40 years or older at baseline?The Women's Interagency HIV Study (WIHS) extant data set 13 and stored biospecimens were used.

Study Population and Setting
This cohort study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline; the institutional review boards of each clinical research site approved the WIHS research protocol, and all participants provided written informed consent.
The WIHS was the largest prospective study of HIV in women (defined as sex at birth and gender) in the US. 13

Sociodemographic and Health History Measures
Sociodemographic factors were self-reported and included date of birth, race (survey options included African American or Black, White, and other race [defined as American Indian or Native Alaskan, Asian, Native Hawaiian or Pacific Islander, and multiracial]), and ethnicity (Hispanic or non-Hispanic).Race and ethnicity were self-reported and queried according to US Census definitions; race and ethnicity data were included in this study because NP T-scores were adjusted for race and ethnicity.Participants were also asked about the highest educational level attained, annual income, use of tobacco, marijuana, alcohol, and other drugs (crack, cocaine, or heroin).They were also asked about their history of hypertension, use of hypertensive medications, history of diabetes, and use of diabetes medications.

Clinical Measures
Body weight and body height measures were measured with participants wearing undergarments. 14dy mass index (BMI) was calculated as weight in kilograms divided by height in meters squared.
Categories of BMI included underweight (<18.5),healthy weight (18.5-24.9),overweight (25.0-29.9),and obese (Ն30). 15stolic blood pressure and diastolic blood pressure were recorded using a standardized protocol with the mean of multiple resting measurements used for analysis.Hypertension was defined as the use of prescribed antihypertensive medications.Diabetes was determined as fasting glucose greater than or equal to 126 mg/dL (to convert glucose to mmol/L, multiply by 0.0555) hemoglobin A 1C greater than or equal to 6.5% (to convert hemoglobin A 1C to the proportion of total hemoglobin, multiply by 0.01), or self-reported use of diabetes medications. 16Estimated glomerular filtration rate (eGFR) was calculated using the chronic kidney disease epidemiology equation. 17

Neurological and Mental Health Measures
Neurological measures included self-reported memory loss, numbness or tingling of the hands and feet, confusion and getting lost, and depression using the Center for Epidemiological Studies-Depression (CES-D) scale. 18The presence of clinically relevant depressive symptoms was defined as a CES-D score of 16 or greater of 60.

HIV-Related Measures
Laboratory-confirmed HIV status, HIV viral load, and CD4 T cell count were considered in analyses of WLWH.Self-reported ART use, and AIDS diagnosis were considered in analyses of WLWH.

JAMA Network Open | Infectious Diseases
Alzheimer Disease Biomarkers in Women With and Without HIV

NP Outcomes
The NP battery included the Letter-Number Sequencing test, 19,20 Trail Making Test Part B, 21 Stroop Test (color word and word reading), 22 Hopkins Verbal Learning Test-Revised (HVLT-R), 23 Symbol Digit Modalities Test, 24,25 Controlled Oral Word Association Test, 26 Category Fluency Test (Animals), 27 and Grooved Pegboard Test. 28Performance on these assessments was used to assess 7 (5) learning (outcome: total learning across HVLT-R trials), (6) verbal fluency (outcomes: total correct on Controlled Oral Word Association Test and Category Fluency Animals test); and (7) fine motor skills (outcomes: total time to completion for each hand on the Grooved Pegboard test). 29,30l timed outcomes were natural log-transformed and reverse scored, so that higher scores represented better performance.Cohort-derived, sociodemographically adjusted T-scores were derived for each outcome according to data for WLWOH. 30Sociodemographic factors included age, education, Wide Range Achievement Test reading subtest (WRAT-3) score, race (Black or African American compared with White individuals and individuals of other races), and ethnicity (Hispanic and non-Hispanic).T-scores were calculated to create domain-specific and global NP scores. 29,30obal NP score was the calculated mean of the 7 NP domains.

AD Biomarker Measures
Baseline and 1-year fasted plasma Aβ40, Aβ42, t-tau, p-tau 231 , GFAP, and NFL concentrations were measured in batch from stored EDTA plasma samples using single molecule array (Simoa) technology on an HD-X Analyzer (Quanterix).Plasma Aβ40, Aβ42, GFAP, and NFL concentrations were measured using the Neuro 4-Plex E kit according to instructions from the manufacturer (Quanterix).Plasma t-tau concentration was measured using the Tau Advantage assay kit.Plasma p-tau 231 concentration was measured using an in-house Simoa assay as previously described. 31Standards and controls were tested in duplicate.Paired longitudinal samples were run side by side on the same plates.The measurements were performed in 1 round of experiments using 1 batch of kits.Intraassay coefficients of variation were below 5% for all biomarkers, except for t-tau where intraassay coefficients of variation were 3.9% to 6.8%.

Statistical Analysis
Participants were characterized overall and by HIV serostatus.Covariates of interest included sociodemographic, behavioral, clinical and neurological factors.Among WLWH, HIV viral load, CD4 T cell count, and history of AIDS were characterized.We calculated t tests for continuous variables and χ 2 tests for categorical variables to assess differences by HIV status at baseline.Primary independent variables were baseline blood-based AD biomarker levels and 1-year change in biomarker levels.
One-year change was calculated as baseline biomarker level subtracted from follow-up biomarker level.The primary outcomes were domains of NP and global NP.Multivariable linear regression analyses were performed to examine cross-sectional baseline and 1-year change in level of each plasma AD biomarker in association with the outcome and NP T-scores among all women (adjusted for HIV serostatus) and stratified by HIV serostatus.In analyses of WLWH only, we additionally adjusted for baseline HIV viral load, CD4 T cell count category, and history of AIDS.
Covariates were selected on the basis of previous WIHS analyses and included income, BMI, CES-D score, history of diabetes, use of hypertensive medication, alcohol, tobacco, marijuana, and other drug use (crack, cocaine, or heroin). 14,29,32We also adjusted for eGFR because kidney function may influence some of these biomarker levels. 33

Results
The Baseline and 1-year change in biomarker characteristics are presented in Table 2.At baseline, median plasma levels of Aβ40, GFAP, and NFL were higher among WLWH compared with WLWOH.
There were no differences in plasma Aβ42, ratio of Aβ42 to Aβ40, t-tau, or p-tau 231 .There were no differences in 1-year biomarker level change by HIV serostatus.Scatterplots of each biomarker at baseline and follow-up are provided as eFigures 1-6 in Supplement 1.
One-year biomarker change calculated as baseline subtracted from follow-up (Table 4), indicated the following among all women, adjusted for HIV status and other covariates: a greater average increase in Aβ40 was associated with worse learning, memory, and motor performance; greater increase in Aβ42 with worse learning and motor performance; greater increase in p-tau 231 with worse motor performance; and greater increase in NFL level with worse processing speed, verbal fluency, and motor performance.When stratified by HIV status, associations of biomarkers with NP predominated in WLWH.Among WLWH, a greater average increase in Aβ40 level was associated with worse learning, memory, and global NP; a greater increase in t-tau with worse executive function; and a greater decrease in NFL level with worse processing speed.Among WLWOH, greater increases in Aβ40 and Aβ42 levels were associated with poorer memory performance, and a greater increase in NFL level with poorer motor performance.

Discussion
In this cohort study, we found that among WLWH and WLWOH with similar sociodemographic characteristics, 1-year increases in plasma Aβ40, Aβ42, p-tau 231 , and NFL were associated with worse motor performance.The association of 1-year increase in plasma Aβ40 with memory performance was consistent among both WLWH and WLWOH, and the magnitude of the regression coefficient was largest for NP associations with Aβ42 followed by t-tau and p-tau 231 .In addition, WLWH performed more poorly on learning, memory, verbal fluency and global NP compared with WLWOH, as previously reported in the WIHS. 14,29,30,34e plasma biomarkers measured reflect the ATN criteria for AD 10 and differentiate AD from other types of dementia. 8,35However, traditionally, cerebrospinal fluid (CSF) measures of these biomarkers have been used to make this differentiation.Associations of CSF with some plasma biomarkers are not robust and none reach a perfect association. 9 In addition, other hypotheses related to AD risk and pathology have been proposed including those related to vascular pathologies, inflammation, neuroimmune dysregulation, synaptic dysfunction, blood-brain barrier alternations, and communicable disease reservoirs. 11,12The construct of multiple-etiology dementias needs to be addressed, and understanding these hypotheses in underrepresented communities with health disparities is underscored. 36e majority of our associations spanning 5 cognitive domains were observed with 1-year change in 2 plasma Aβ peptides (Aβ40 and Aβ42) and NFL in the entire sample adjusted for HIV status and other covariates.In addition, worse motor performance was consistent with 1-year increase in these 3 biomarkers, as well as p-tau 231 .Aβ peptides are cleavage products of amyloid precursor protein, a transmembrane protein present in almost every cell of the human body.Aβ deposits (amyloid plaques) in the brain are a neuropathological hallmark of both clinical and neuropathological AD. 10 Lower levels of Aβ42 and Aβ40 in CSF and plasma and decreases in those levels are associated with AD, 37 cognitive decline, 38 and psychiatric and behavioral symptoms in aging individuals without HIV. 39Some studies have reported no association with either plasma Aβ42 or Aβ40 levels in PLWH, specifically men who have sex with men and transgender women. 40 observed no association of NP with the Aβ42-to-Aβ40 ratio, which has been associated with an increased incidence of AD in HIV-uninfected aging individuals. 9However, no association of plasma Aβ42-to-Aβ40 ratio in people aged older than 40 years with AD and without HIV compared with similar-aged PLWH who were cognitively normal or who had AD was also reported. 41u biomarkers in CSF, notably t-tau and p-tau, are prognostic and diagnostic for AD, which is characterized by neurofibrillary tangles. 42Recently, tau positron emission tomography has evolved for clinical diagnosis of AD and differentiates from other neurodegenerative tauopathies. 43P-tau 231 may be the preferred CSF discriminatory tau protein for an AD diagnosis. 43,44observed 1 association of p-tau 231 with motor performance.To our knowledge, there are no published reports on this blood-based protein in PLWH.
Neurofilaments are neuronal cytoskeletal proteins released into the CSF or blood in response to neurological injuries and neurodegeneration and have been associated with dementia, stroke, and traumatic brain injury.HIV crosses the blood-brain barrier, activates microglia, and leads to neuroinflammation and neurodegeneration, resulting in the release of NFL. 45 Published studies 46 in cross-sectional samples of PLWH have reported higher NFL levels in association with lower NP, and no association between baseline levels of NFL and subsequent cognitive decline over 24 months or more in PLWH, including men who have sex with men and transgender women. 40,47Some reports indicate that impaired kidney function, more commonly observed among PLWH, may be associated with changes in blood NFL levels. 33However, since the molecular weight of NFL is more than 60 kDa and NFL is negatively charged, its biokinetics may be unaffected in people with impaired kidney function as measured via eGFR. 48 observed no associations of NP with GFAP, even though astrocytes make up approximately30% of the cells in the central nervous system and are constituents of the blood-brain barrier. 49GFAP is an astrocyte protein measurable in CSF and plasma and is a biomarker of both central neurological disorders and injuries. 9Our analyses have many strengths.First, the WIHS cohort was active with a consistent research protocol for over 25 years.Second, the WIHS comprises WLWH and WLWOH who are underrepresented in research.The cohort consisted of women of many races and ethnicities and predominately included women who have lower educational attainment and lower annual income.
Third, we used state-of-the-art Simoa technology to analyze our biomarker samples.Single molecule protein detection is based on the isolation of individual immunocomplexes on paramagnetic beads using standard enzyme-linked immunosorbent assay reagents and is 1000 times more sensitive than traditional enzyme-linked immunosorbent assays.However, differences in laboratory methods and evolution in sensitivity and specificity over time in this research field may contribute to differences in published results.Fourth, to our knowledge, there are no other publications presenting data based on a cohort of WLWH and comparison group of similar women assessing 6 blood-based AD biomarkers repeated over a 1-year period in association with NP assessments.Fifth, Aβ isoforms, such as Aβ42 and Aβ40, and t-tau and p-tau isoforms have traditionally been measured in the CSF.
However, CSF collection is not practical in many underrepresented communities.Finally, due to extensive clinical data available in the WIHS, we were able to adjust for eGFR; however, this adjustment did not influence the results in all essential aspects.

Limitations
As with any epidemiologic observational study, there are limitations to our analyses.First, although the study demonstrated several associations of NP with blood-based biomarkers, external validity may be compromised.WIHS participants in these analyses comprised a well-studied group of women who were followed for 6 to 7 years or more.Second, since WIHS was an intent-to-treat cohort, these women may have been healthier than WLWH who did not participate in a research study.In 2018 and 2019, at the time of these blood measures, more than 90% reported being adherent to ART. 13 Although ARTs are associated with lower blood levels of NFL, 50 the WLWH in our study had higher  NFL levels compared with WLWOH.Third, some of the biomarkers measured are sensitive but not specific.For example, higher blood levels of NFL and GFAP have been observed in several neurological conditions and accompany peripheral and central neurological injury or events. 13urth, other blood-based biomarkers of inflammation (eg, chemokines, interleukins, and tissue inhibitor of metalloproteinases-1) and altered metabolism (eg, leptin, adiponectin, ghrelin, amylin, and gastric inhibitory peptide) have been associated with NP in WLWH in the WIHS at different times over the adult life course. 14,29,32We cannot consider their contribution here.However, other biomarker study results have been consistent with HIV pathophysiology and associated comorbidities, such as obesity, and are potential biomarkers that may be useful in the creation of a biomarker panel among WLWH with promising clinical applications.Fifth, our sample is composed primarily of Black women predominantly living in urban environments; thus, extrapolation to other racial and ethnic groups, men, and living environments is limited.Sixth, EDTA plasma is required for all AD and related neurodegenerative disorder biomarkers measured via Simoa except for NFL.In our cohort, we did not collect EDTA plasma until 2018.In combination with NP assessments being given once every 2 years, we can only evaluate a 1-year biomarker change in association with 1 NP assessment.Seventh, due to the exploratory nature of our analyses, we chose not to adjust for multiple comparisons.Therefore, some of our results may represent type 1 error.Eighth, data on other neurological symptoms, signs, and outcomes (eg, stroke) that may be associated with NP were not routinely collected in this cohort or were rare.

Conclusion
This study found that increases in plasma Aβ42 and Aβ40 peptides, p-tau 231 , and NFL over time were associated with NP.These findings suggest that they are promising preclinical biomarkers of NP and brain health in underrepresented samples of WLWH and WLWOH.
WIHS began in 1994 and initially enrolled 2054 WLWH and 1712 WLWOH who were sociodemographically similar across 6 sites in San Francisco, California; Los Angeles, California; Chicago, Illinois; Washington, DC; and Brooklyn and the Bronx in New York, New York.Two additional enrollment waves occurred in 2001 to 2002 and 2011 to 2014.In 2012, 4 additional sites were added in Atlanta, Georgia; Birmingham and Jacksonville, Alabama; Chapel Hill, North Carolina; and Miami, Florida; and 1 site was dropped (Los Angeles, California).Semiannual WIHS core visits included sociodemographic, behavioral, and clinical measures.A standardized comprehensive NP battery was administered every 2 years starting in 2009.A subset of WIHS participants who completed 1 NP assessment between 2017 and 2019 had batch testing of fasting plasma Aβ40, Aβ42, t-tau, p-tau 231 , GFAP, and NFL levels in 2018 (designated as baseline) and 1 year later in 2019.
NP domains: (1) attention and working memory (outcomes: total correct on the Letter-Number Sequencing control and experimental conditions), (2) executive function (outcomes: time to completion on the Trail Making Test Part B and Stroop color word test [interference] trial), (3) processing speed (outcomes: total correct on the Symbol Digit Modalities Test and time to completion on the Stroop word-reading trial), (4) memory (outcome: HVLT-R delayed recall score),

JAMA Network Open | Infectious Diseases Alzheimer
Race, ethnicity, WRAT score, education, and age were not included as covariates because they were already included in the demographically adjusted NP T-scores.Data analyses were conducted using R software version 4.3.2(R Project for Statistical Disease Biomarkers in Women With and Without HIV JAMA Network Open.2023;6(11):e2344194. doi:10.1001/jamanetworkopen.2023.44194(Reprinted) November 29, 2023 4/17 Downloaded from jamanetwork.comby guest on 12/04/2023 Computing).Results were considered significant at 2-sided P < .05.Data analysis was conducted from April to December, 2022.

Table 1 .
SD] T-score 48.3 [9.8] vs 50.7 [8.5]), and global (mean [SD] T-score 49.2 [6.8] vs 51.1 [5.9]) NP assessments.Baseline survey and clinical characteristics of participants are presented in No participants had a diagnosis of AD or other type of dementia.

Table 1 .
Baseline Demographic, Alzheimer Disease Biomarker, and Neuropsychological Performance Characteristics for All Women and by HIV Serostatus (continued) a Variables reported as No. (%) were analyzed with χ 2 tests.b Other race was defined as American Indian or Native Alaskan, Asian, Native Hawaiian or Pacific Islander, and multiracial.c Body mass index was calculated as weight in kilograms divided by height in meters squared.d T-scores are demographically adjusted for age, education, Wide Range Achievement Test reading subtest score, race (African American vs White or other race), and ethnicity (Hispanic vs non-Hispanic).

Table 3 .
Linear Regression Models Associating Cross-Sectional Baseline Plasma AD Biomarker Measures With Neuropsychological Performance by Domain in All Women and Stratified by HIV Serostatus a (continued) Depression scale score, estimated glomerular filtration rate, history of hypertension, diabetes, tobacco use, marijuana use, alcohol use, and other drug use (crack, cocaine, or heroin).β represents an adjusted, unstandardized coefficient.Neuropsychological performance assessments were reverse scored, and a higher β indicates better neuropsychological performance.
Abbreviations: Aβ, amyloid β; AD, Alzheimer disease; GFAP, glial fibrillary protein; NFL, neurofilament light chain; P-tau, phosphorylated tau; T-tau, total tau; WLWH, women living with HIV; WLWOH, women living without HIV.aAll models are adjusted for HIV status, income, body mass index, Center for Epidemiological Studies-b WLWH models were additionally adjusted for baseline CD4 T cell count, HIV viral load, and history of AIDS.

Table 4 .
Linear Regression Models Associating 1-Year Change in AD Biomarker Measures With Neuropsychological Performance by Domain in All Women and by HIV Serostatus a (continued) Abbreviations: Aβ, amyloid β; AD, Alzheimer disease; GFAP, glial fibrillary protein; NFL, neurofilament light chain; P-tau, phosphorylated tau; T-tau, total tau; WLWH, women living with HIV; WLWOH, women living without HIV.All models are adjusted for HIV status, income, body mass index, Center for Epidemiological Studies-Depression scale score, eGFR estimated glomerular filtration rate, history of hypertension, diabetes, tobacco use, marijuana use, alcohol use, and other drug use (crack, cocaine, or heroin).β represents an adjusted, unstandardized coefficient.Neuropsychological performance assessments were reverse scored and a higher β indicates better neuropsychological performance.Anderson AM, Jang JH, Easley KA, et al.Cognitive and neuronal link with inflammation: a longitudinal study in people with and without HIV infection.J Acquir Immune Defic Syndr.2020;85(5):617-625.doi:10.1097/QAI.0000000000002484 48.Alagaratnam J, De Francesco D, Zetterberg H, et al.Correlation between cerebrospinal fluid and plasma neurofilament light protein in treated HIV infection: results from the COBRA study.J Neurovirol. 2022;28 (1):54-63.doi:10.1007/s13365-021-01026-349.Sweeney MD, Sagare AP, Zlokovic BV.Blood-brain barrier breakdown in Alzheimer disease and other neurodegenerative disorders.Nat Rev Neurol.2018;14(3):133-150.doi:10.1038/nrneurol.2017.18850.Hagberg L, Edén A, Zetterberg H, Price RW, Gisslén M. Blood biomarkers for HIV infection with focus on neurologic complications-a review.Acta Neurol Scand.2022;146(1):56-60.doi:10.1111/ane.13629Association Between Baseline and Follow-Up Levels of Aβ40 in the Entire Sample eFigure 2. Association Between Baseline and Follow-Up Levels of Aβ42 in the Entire Sample eFigure 3. Association Between Baseline and Follow-Up Levels of GFAP in the Entire Sample eFigure 4. Association Between Baseline and Follow-Up Levels of NFL in the Entire Sample eFigure 5. Association Between Baseline and Follow-Up Levels of T-tau in the Entire Sample eFigure 6. Association Between Baseline and Follow-Up Levels of P-tau231 in the Entire Sample a b WLWH models were additionally adjusted for baseline CD4 T cell count, HIV viral load, and history of AIDS.47.