HAND in Africa: prevalence, diagnostic challenges and the search for objective markers

Abstract

HIV Associated Neurocognitive Disorder (HAND) affects 20–50% of people with HIV (PWH) on suppressive antiretroviral therapy, with about 72% of the 16 million global cases residing in sub-Saharan Africa. Despite this disproportionate burden, HAND remains very challenging to diagnose. The Frascati diagnostic criteria which are considered the mostly used criteria albeit with recent challenges require specialist to administer the neuropsychological batteries, making it inaccessible, literacy-dependent, and culturally biased in resource-limited African settings. This creates a significant diagnostic capacity gap. Identifying and developing biomarker-based HAND diagnostic assays are critical to the early detection and effective management of HAND among people with HIV in Africa and other resource-limited settings. This narrative review used 80 articles to critically evaluate HAND neuropathogenesis, existing diagnostic tools, and the emerging peripheral blood biomarker landscape. HIV establishes a persistent CNS reservoir that drives neuroinflammation through infected microglia, dysfunctional activation of astrocytes, and neurotoxic viral proteins mainly gp120, and Tat, independently of immune-virologic state of the PWH. Among candidate biomarkers, serum ICAM-5, serum ATP, and plasma neurofilament light chain (NfL) appear to be among the most promising candidates; validating such markers in a multicenter approach to complement Africa friendly neuro-psychological test tools will help improve on HAND diagnosis. Once some biomarkers are validated, we propose that a 3-tier multi-marker diagnostic framework that could be used in complementing validated tools, spanning primary care serum-based screening through to specialist CSF biomarker profiling could enable objective, regional friendly, and early HAND detection at the point of HIV care across Africa.

Introduction

Human immunodeficiency virus (HIV) infection is still a global pandemic with an estimated (40.8) million people living with HIV/AIDS by the end of 2024 (UNAIDS 2025a). The magnitude of the infection is most prominent in Sub-Saharan Africa, region with a 65% of the global HIV population (UNAIDS 2025b).

According to the Ghana AIDS commission, the country’s HIV population stands at 334,721 in 2024, with an estimated prevalence of 1.49%, down from the 1.6%, reported in 2023 (Commission 2025), an improvements in prevention and treatment coverage.

The main challenge PWH face is the gradual reduction in their CD4 T-Lymphocyte numbers and exhaustion of T-cells, leading to a suppressed immune system, and making the infected person susceptible to opportunistic infections and immune-mediated complications (Afam et al. 2014; Govindaraj et al. 2023). The dynamics involved in this T-Cell depletion has been shown to involve the expression of CCR5 on CD4 T cells (Afam et al. 2014; Govindaraj et al. 2023).

Beyond the immune system, HIV exerts systemic effects across multiple organ systems (León-rivera et al. 2021), with the central nervous system (CNS) among the earliest and most persistently affected compartments. HIV enters the CNS within weeks of primary infection and establishes a protected viral reservoir behind the blood-brain barrier (BBB) that persists even under virologically suppressive ART (León-rivera et al. 2021; Maria et al. 2016; Frank and Brew 2025).

Once established, HIV productively infects microglia and perivascular macrophages, induces astrocyte dysfunction, and triggers a sustained neuroinflammatory cascade involving TNF-alpha, IL-1beta, CXCL10, and CCL2, ultimately compromising neuronal integrity (Frank and Brew 2025; Smith et al. 2018; Wang et al. 2020). The neurological sequela is HIV-associated neurocognitive disorder (HAND), formalized at the Frascati consensus in 2004 as a spectrum of progressive cognitive, behavioral, and motor impairments (Shao and Li 2023; Gartner et al. 1986). HAND encompasses three sub-syndromes: Asymptomatic Neurocognitive Impairment (ANI), Mild Neurocognitive disease (MND), and HIV Associated Dementia (HAD); the last representing the most severe and historically dominant form in the pre-ART era (Koenig et al. 1986). The viral proteins Tat and gp120 are the major HIV proteins responsible for the neuropathological cascade, and also, higher CNS viral reservoir burdens are associated with greater HAND risk (Pope et al. 1994).

The advent of cART transformed HIV from a rapidly fatal to a chronic manageable condition, substantially reducing HAD incidence (Maria et al. 2016; Kamerman et al. 2012). However, milder HAND forms now predominate, affecting 20–50% of PWH on suppressive cART (Smith et al. 2018). Ageing compounds this burden: in the USA, approximately 72% of PWH are now over 50 years old, and ageing synergizes with HIV-driven CNS pathology to amplify HAND risk (Maria et al. 2016). In Ghana, Sarfo et al. (2021) reported a combined HAND prevalence of 44.7% — consistent with the Africa-wide meta-analytical estimate of 46.34% (Russell et al. 2017; Cui et al. 2021). Globally, approximately 72% of an estimated 16 million HAND cases reside in sub-Saharan Africa (Wang et al. 2020).

Despite this burden, diagnosis of HAND remains deeply challenging. Neuropsychological test batteries which require trained specialists and culturally appropriate normative data, are biased against low-literacy populations, and lack standardization across settings (He et al. 1997). These limitations translate directly to missed diagnoses, delayed interventions, and reduced quality of life (Levy et al. 1995; Liang et al. 2017). The identification of peripheral blood-based biomarkers that correlate well with cognitive decline in HIV could help standardized HAND diagnosis, particularly in sub-Saharan Africa where these constraints are most pronounced, however more than a decade of research in this area has so far not yielded the desire results.

This narrative review synthesizes current evidence on HAND neuropathogenesis, diagnostic tools, and the emerging biomarker landscape up to 2025, with particular emphasis on translational potential in resource-limited settings. We propose a biomarker-assisted tiered diagnostic framework suited to primary and secondary HIV care in Africa.

Methodology

Search strategy

This is a narrative review. A comprehensive literature search was performed across PubMed/MEDLINE, Scopus, Web of Science, Google Scholar, Science Direct and the Cochrane Library. Searches were conducted at the least, once a month between January 1st and December 31 st 2025.

The following Boolean search strings were applied:

“HIV-associated neurocognitive disorder”, OR “pathophysiology of HIV-associated neurocognitive disorder”, OR “biomarkers of HIV-associated neurocognitive disorder”, OR “molecular mechanisms of HIV-associated neurocognitive disorder”, OR “HAND Africa” OR “diagnosis of HAND” OR “HIV and “Neurofilament” OR “ICAM-5” Or “ATP” OR “Exosomes” OR “sCD163” OR “GFAP” OR “HAND and sub-Sahara Africa”.

We also applied other search words like ‘HIV associated neurocognitive disorder’, ‘pathophysiology of HIV associated neurocognitive disorder’, ‘biomarkers of HIV associated neurocognitive disorder’, ‘molecular mechanisms of HAND’, HAND in Africa in our google and google scholar searches to retrieve article from PubMed, ScienceDirect and other sources.

Inclusion and exclusion criteria

Inclusion criteria

Peer-reviewed original research, systematic reviews, meta-analyses, and narrative reviews published in English between 2010 and 2025; studies involving PWH; studies reporting on at least one of the following; HAND prevalence, diagnostic tools for HAND, biomarker evaluation for HAND, or neuropathological mechanisms. Key foundational papers before 2010 were included where they established core concepts (e.g., Frascati nosology, Trojan horse hypothesis).

Exclusion criteria

Non-English articles (as French and other language translation was not feasible and beyond the scope of this manuscript); grey literature; conference abstracts without full-text; animal studies with no direct human clinical translation.

Study selection

Titles and abstracts were screened by two reviewers (PKA, MEH), with full texts retrieved for all records meeting inclusion criteria. Reference lists of included reviews were hand-searched for additional relevant articles. A total of 84 articles were used for this review.

Neuropathogenesis of HAND

Mechanism of entry

HIV-1 uses its enveloped glycoprotein 120 and 41 to attach and eventually infect target cell that expresses CD4, especially CD4 T-lymphocytes. It also requires a co-receptor which is either CCR5 or CXCR4 for this attachment. Thus, HIV can infect all cells that express these cell surface receptors (CD4 and CCR5 or CXCR4), including T-lymphocytes, microglia cells monocytes/macrophages, dendritic cells (Shao and Li 2023; Gartner et al. 1986; Koenig et al. 1986; Pope et al. 1994). With the BBB acting as a very strong defense against entry of pathogens, many hypotheses have been put forward about how the virus enters the central nervous system (CNS), during the early stage of infection.

1. “Trojan horse” theory, suggesting that the virus gains access to the CNS using infected CD4 + T lymphocytes as a means of transport into perivascular spaces while evading immune surveillance (León-rivera et al. 2021; Sorrells et al. 2018; Churchill and Nath 2017).

2. Direct BBB disruption: HIV-1 Tat inhibits occludin expression and cleaves it via MMP-9, compromising tight junctions between endothelial cells within blood vessels (Pu et al. 2003).

3. Transcytosis via gp120: cellular transfer mediated by the viral envelope protein facilitates movement of free virus across the BBB (Gerhart et al. 1992).

4. LPS-mediated pathway: HIV-1 infection also compromises the structural integrity of the intestinal tract and can cause leakage of bacteria into the blood stream. Such microbial translocation results in elevated plasma levels of bacterial lipopolysaccharide (LPS) (Shao and Li 2023). In response to LPS, the BBB becomes more vulnerable, facilitating the transmigration of peripheral monocytes/macrophages. HIV is also associated with increased monocyte activation and dementia. The LPS is thought to also use an alternate mechanism by enhancing the trans-cellular transport of HIV-1 across the BBB via a p38 MAPK-dependent pathway (León-rivera et al. 2021; Churchill and Nath 2017).

5. Retrograde movement: It is also suggested that HIV may directly infect peripheral nerve fibers and subsequently travel in a retrograde manner to reach the CNS (León-rivera et al. 2021; Wang et al. 2020).

Cellular targets and HIV pathology within the CNS

Once inside the CNS, HIV invades various cell types, including astrocytes, microglia, oligodendrocytes (León-rivera et al. 2021), affecting these cells differently and having different levels of neurological damage (Shao and Li 2023; Russell et al. 2017; Cui et al. 2021; He 1 et al. 1997; Levy et al. 1995; Liang et al. 2017; Sorrells et al. 2018; Nath et al. 2000; Guinovart 2013; Churchill and Nath 2017; Pu et al. 2003; Gerhart et al. 1992; Lauritzen et al. 2013; Tang 2020; González-scarano and Martín-garcía 2005; Bansal et al. 2000; Eugenin et al. 2011).

Microglia, the principal reservoir for productive HIV infection within the brain when chronically infected release inflammatory chemokines and cytokines (TNF-alpha, IL-1beta), and produce neurotoxic viral proteins (gp120, Tat, Vpr), and neurotoxic factors such as PDGF, quinolinic acid (QUIN), and nitric oxide (González-scarano and Martín-garcía 2005; Bansal et al. 2000). Astrocytes, which can be infected through endocytosis, and maintain viral latency (León-rivera et al. 2021); when activated by neurotoxic viral proteins and cytokines, make BBB leaky, disrupts glutamate release and uptake, and increases intracellular Ca2 + release, causes CNS energy metabolism dysregulation, thus contributing to excitotoxicity (Eugenin et al. 2011; Zou et al. 2019). Oligodendrocytes are targeted by HIV-1 Tat via NMDA and AMPA receptor mechanisms, as well as CaMKIIβ–GSK3β interactions, leading to demyelination and white matter damage (Lauritzen et al. 2013; Nass et al. 2022). Neurons sustain indirect damage through viral proteins that disrupt Ca2 + homeostasis, create excitatory/inhibitory imbalance, and drive apoptosis and synaptodendritic degeneration (Paris et al. 2016).

The cell-specific effects of HIV are summarized in Table 1.

Table 1.

Effects of HIV on CNS Cell Populations

Cell type	Function in CNS	How HIV infect Cells	Pathology caused by HIV
ASTROCYTES	Provide vital support to control water and ion balance Regulate glutamate concentration Release cytokines Promote myelination Wahl et al. (2023; Sofroniew 2020) Wrap around endothelial cells to preserve the blood-brain barrier (Wahl et al. 2023).	1. Endocytosis and maintain viral latency Lutgen et al. 2020; Li et al. 2021 2. CXCR4 dependent, CD4 independent mechanism Li et al. 2021	1. Energy metabolism disorders (León-rivera 2021; Paris et al. 2016 2. Formation of glial scars 3. Influence the BBB León-rivera 2021; Sorrells et al. 2018; Guinovart 2013; Churchill and Nath 2017; Pu et al. 2003; Gerhart et al. 1992 4. Increases the release of intracellular León-rivera 2021 Ca2 + and glutamate; Decreases the glutamate uptake León-rivera 2021 Increases neurotoxins production León-rivera 2021; Wahl et al. 2023)
MICROGLIA	Major CNS immunological defense Cui et al. 2021; He et al. 1997 Maintain neuronal viability and synaptodendric viability (Menassa et al. 2018)	Direct HIV infection He et al. 1997	Release inflammatory chemokines and cytokines Wesselingh et al. 1997; Sreeram et al. 2022 Induces the release of viral proteins (gp120, Tat, Vpr) Scutari et al. 2017 Induces neurotoxins production (Inflammatory mediators, PDGF, QUIN) Scutari et al. 2017 Activates viral replication (Scutari et al. 2017
OLIGODENDROCYTES	Form the myelin sheath of the axon León-rivera 2021; Scutari et al. 2017; Menassa et al. 2018	Production and release of viral proteins León-rivera 2021; Wesselingh et al. 1997
NEURON	Fundamental functional units León-rivera 2021; Russell et al. 2017	1. Direct HIV infection 2.Viral proteins disrupt the Ca2 + homeostasis (León-rivera 2021; Cui et al. 2021	1.E/I imbalance 2.Neuronal dysfunction, injury and death Shao and Li 2023; Eugenin et al. 2011

Certain cortical and subcortical regions of the brain, including the medial prefrontal cortex (MPFC), basal ganglia, and hippocampus, which play crucial roles in cognition and emotion, are susceptible to the impact of HIVand HIV proteins, leading to functional and structural alterations within these regions, resulting in the cognitive, behavioral and motor manifestation in HAND (Smit et al. 2001; Giatsou et al. 2020; Wesselingh et al. 1997; Sreeram et al. 2022; Hildebrand et al. 1993; Foerster et al. 2024).

Chronic-phase CNS pathology and the HAND-Alzheimer’s disease intersection

The acute infection period in HIV is very brief (Shao and Li 2023), and the increase viremia and a return to steady state happens within 4–5 weeks (Robb and Ananworanich 2016) During this period, the massive viral replication causes CD4 T-cell numbers to decrease rapidly. With the advent of cART and early initiation, a chronic phase of the infection ensues where HIV patients on cART can maintain CD4 T-cell numbers at above 250 cells/mL. This chronic phase of HIV infection is characterized by the induction of premature aging of the immune system including senescence. with release of pro-inflammatory cytokines such as TNF-α and IL-1B, which in turn activate astrocytes and microglia (Shao and Li 2023) leading to neuronal death, cognitive decline and synaptodendritic loss (León-rivera et al. 2021; Sofroniew 2020; Li et al. 2021).

An important emerging paradigm is the overlap of HAND with Alzheimer’s disease-like neurodegeneration (ADLD). As PWH age on ART, the clinical presentations of HAND increasingly resemble cortical dementia, with accumulation of amyloid-beta and hyperphosphorylated tau, a shift attributed to chronic neuroinflammation, HIV-driven epigenetic changes, and APOE e4 susceptibility (Scutari et al. 2017). This overlap has direct implications for biomarker selection: AD markers (CSF tau, amyloid-β, EV-associated tau/Aβ42) are now entering the HAND biomarker evidence base, interestingly p tau 181 looks promising as a good differentiator of HAND and AD (Clifford and Holtzman 2009).

Importantly, HAND severity does not correlate reliably with plasma viral load or CD4 count, but is more strongly associated with duration of untreated HIV infection, highlighting that cumulative neurotoxic burden can persist even after viraemia is controlled (Levy et al. 1995). This dissociation between standard HIV markers and CNS disease is a central argument for dedicated HAND biomarkers.

Diagnosis of HAND by the Frascati framework

HAND is diagnosed through neuropsychological testing and functional status assessment, guided by the Frascati criteria formalized by Antinori et al. in 2007 (Wesselingh et al. 1997). A comprehensive neurocognitive battery evaluates motor function, verbal fluency, executive function, attention and working memory, information processing speed, learning, and memory across at least five cognitive domains (He et al. 1997; Hildebrand et al. 1993). Performance is classified relative to age- and education-matched normative standards and integrated with functional assessment to assign one of three HAND sub-diagnoses.

Asymptomatic Neurocognitive Impairment (ANI)

ANI requires impairment in at least two cognitive domains (> 1 SD below the normative mean on standardized testing) without observable functional impairment in daily activities. Cognitive impairment must be attributable to HIV and not to another aetiology (Wesselingh et al. 1997). ANI is the most prevalent yet clinically least detectable sub-syndrome especially when using the Frascati criteria. The diagnosis of ANI by the Frascati criteria has had controversies over the years; clinically since the condition is asymptomatic its relevance has been questioned as it does not present with any overt signs and if the potential for developing mild or severe cognitive disorder later cannot be established (Torti et al. 2011), also studies have intimated that socioeconomic cultural and education can so influence the diagnosis of ANI such that as high as 20 of false positive exist (MS et al. 2024). However a longitudinal study by Grant and colleagues in 2014 showed that PWH diagnose with ANI had a 2-fold to 6 fold risk of developing symptomatic HAND (Grant et al. 2014).

Mild Neurocognitive Disorder (MND)

MND requires mild to moderate impairment in at least two cognitive domains with at least mild interference in one or more activities of daily living — work efficiency, social functioning, or homemaking — attributable to HIV and not another primary cause (Wesselingh et al. 1997).

HIV-Associated Dementia (HAD)

HAD requires performance impairment of two or more standard deviations below normative data in at least two cognitive domains, with marked day-to-day functional impairment not explained by a virus-independent pre-existing condition or age-related disease (Wesselingh et al. 1997).

The Frascati criteria as used above have recently been found to be too liberal (Campbell et al. 2020) and overly diagnose up to 20% of cognitively healthy individuals as having cognitive defect in HIV (Sam Nightingale et al. 2023), because of the lack of appreciation of educational and socio-cultural dynamics in it, suggesting the inclusion of clinical assessment, the use of neuroimaging, and the use of the global deficit score in the classification of HAND, albeit with limitations. There is a suggestion from a consensus statement currently published (Sam Nightingale et al. 2023) advocating the separation of HIV brain injury from the HAND diagnosis criteria (Sam Nightingale et al. 2023), while others are advocating for reclassification into biotypes based on immune reaction and viral pathogenesis of the PWH (Johnson and Nath 2023). The diagnostic and classification challenges of HAND is still an area that needs more research.

Challenges in HAND diagnosis and the Africa diagnostic gap

Procedural differences of assessment tools

There is no agreement on a standardized neuropsychological battery for HAND diagnosis. Studies in Ghana and across Africa as well as in the global space have employed different assessment tools including the Trail Making Test A/B (TMT-A/B), Stroop Color-Word Test (SCW), International HIV Dementia Scale (IHDS), Montreal Cognitive Assessment (MoCA), Stick Design Test (SDT), 10-Item Word List Learning, Memorial Sloan-Kettering Scale (MSK), and Karnofsky Performance Status Scale, in the diagnosis of HAND (Russell et al. 2017; He et al. 1997; Hildebrand et al. 1993; Foerster et al. 2024). Table 2 provides a comparative analysis some of the tools used.

Table 2.

Comparative Analysis of Neuropsychological Assessment Tools Used in HAND Diagnosis

Assessment tool	Cognitive domain assessed	time to administer	English literacy required	Validation in Africa	Limitations in RLS application	Brief on the Procedure used
International HIV Dementia Scale (IHDS)	Motor speed, memory registration, psychomotor speed	~ 5 min	minimal	Yes (Ghana and Uganda) Hildebrand et al. 1993	Low sensitivity for mild HAND (ANI, MND); does not capture full Frascati spectrum	Four words to recall, finger tapping, and other maneuvers of the non-dominant hand
Montreal Cognitive Assessment (MoCA)	Attention, memory, language, visuospatial, executive, orientation	~ 5–10 min	moderate	Partial (Ghana) Russell et al. 2017	Education bias; Western normative data; under-detects subcortical features of HAND	Connect numbers and letters in ascending order draw abstract things call names of animal spell words backward ect
Trail Making Test A & B (TMT-A/B)	Processing speed (A); executive function and cognitive flexibility (B)	~ 5–10 min	high	Limited	Requires literacy and education ≥ 10 years; high cultural bias; inapplicable in low-literacy cohorts He et al. 1997	Connecting numbers randomly arranged & numbers to alphabets
Stroop Colour-Word Test (SCW)	Executive function, response inhibition, processing speed	~ 5 min	high	Limited	Colour-naming requires language fluency; inapplicable where formal education is absent He et al. 1997	Rapidly read colors, name colors and name the color of a printed incongruent color
Stick Design Test (SDT)	Visuospatial construction and executive function	~ 10 min	none	Yes (West Africa) Hildebrand et al. 1993	Single domain; insufficient alone for Frascati classification	Replicating 4 geometric shapes with match sticks
10-Item Word List Learning	Verbal learning and episodic memory	~ 10 min	Minimal (oral)	Yes (several African settings) Russell et al. 2017	Limited to memory domain; requires oral instruction translation for non-English speakers	Memorizing 10 -item names and recall after 5–10 min
Memorial Sloan-Kettering Scale (MSK)	Global neurological staging (motor, cognitive)	~ 10 min	Minimal	Limited	Coarse staging; insufficient for ANI/MND differentiation; originally designed for HAD	Clock drawing to show time correctly, word recall after 3–5 min, several other assessments
Karnofsky Performance Status Scale (KPS)	Functional/physical performance status	~ 5 min	None (clinician-rated)	Yes (HIV clinics) Russell et al. 2017	Not a cognitive test; no direct neuropsychological domain assessment; proxy only	Uses an algorithm to determine how a person can live independently (ability to perform daily activities)
Brief Neurocognitive Screen (BNCS)	Processing speed, attention, executive function	~ 15 min	Low–Moderate	Limited	Limited African normative data; requires trained administrator	Use of tools like Mini-Cog, MoCA, GPCOG RUDAS for quick screen
HIV Dementia Scale (HDS)	Memory recall, psychomotor speed, construction	~ 5 min	Low	Partial	Poor sensitivity for mild HAND; derived from HAD era; less applicable to cART-treated cohorts
International HIV Dementia Scale (IHDS)	Processing speed, attention, executive function	~ 15 min	Low–Moderate	Limited	Limited African normative data; requires trained administrator

Variations in prevalence estimates and the Africa diagnostic gap

In a global meta-analysis of 123 studies involving 35,513 participants, from 32 countries, the overall HAND prevalence of 42.6% (95% CI: 39.7–45.5%) and did not significantly differ by diagnostic criteria used (Zou et al. 2015). However, the Africa-specific meta-analysis by Mekuriaw et al. (2023) found a pooled prevalence of 46.34% (95% CI: 40.32–52.36%) with marked heterogeneity (I2 = 98.5%, p = 0.001) with prevalence ranging from 14% to 88% across African studies (Mekuriaw et al. 2023), with viabilities in the prevalence and the approaches used, including the study design, the types of neurocognitive assessment tools used (Mekuriaw et al. 2023). There is therefore a diagnostic dilemma when it comes to HAND, for example one study used the Trail Making Test Part A (TMT A), Stroop Colour-Word (SCW), and the Inter-national HIV Dementia Scale (IHDS) (Asiedu et al. 2020), Sarfo et al. used the Montreal Cognitive Assessment (MOCA), Stick Design Test (SDT), 10-Item Word List Learning, TMT A, Karnofsky Performance Status Scale (Sarfo et al. 2021) and another study used the Memorial Sloan-Kettering Scale (MSK) (Yuan et al. 2017) to assess cognitive function and diagnose HAND.

A review by Olajide et al. (2024), that focused on HAND only in Africa demonstrated that, the highest pooled prevalence (49.33%) is from Central and Southern Africa followed by East Africa (45.04%) and then West Africa (42.40%) (Olajide et al. 2024), with differences in ART coverage, TB co-infection rates, and healthcare infrastructure influencing the prevalence (Robb and Ananworanich 2016). This is somehow a confirmation of the work done earlier by Zenebe et al. in 2022, which calls for consistency in measurement approaches for the diagnosis of HAND (Milicic et al. 2025).

Thus, the critical diagnostic gap in Africa stems from the findings of Olajide and his group which when expanded, covers issues like; late ART initiation, frequent non-adherence, low awareness in neurocognitive complication which are usually attributed to spiritual attacks, absent specialist neurology services, and cultural mismatch between imported assessment tools and local populations and the literacy gap (Levy et al. 1995; Hildebrand et al. 1993; Robb and Ananworanich 2016). In Ghana, an estimated 17% or more of HIV clinic attendees do not have any formal education, rendering tools requiring literacy, such as TMT-A/B, MoCA and the SCW essentially inapplicable (He et al. 1997). Others such as IHDS and SDT require less literacy demands and have been validated in several African settings, but none of them constitutes a comprehensive cognitive battery, alone (Hildebrand et al. 1993).

Frascati criteria limitations in resource-limited settings

The Frascati criteria in its current form were calibrated with normative that data does not contain contribution from sub-Saharan African populations and hence its applicability is problematic. Including an African data would have accounted for the confounding effects of common African co-infections like TB, malaria, and CMV, that independently impair cognition, it would also consider the ADLD-like presentations increasingly observed in ageing African PWH (Scutari et al. 2017; Robb and Ananworanich 2016). This limitation has been recognized by a 2023 Nature Reviews Neurology consensus which proposed an updated criteria less reliant on specialist-administered batteries, the implementation of these updated Frascati criteria in African clinical practice remains incipient (Clifford and Holtzman 2009). The situation is critical in resources limited settings (RLS) where late initiation of ART, opportunistic infections, lower awareness of HIV/AIDS treatments, frequent non-adherence to antiretroviral therapy (ART), and the poor health care systems compounds HAND diagnosis dilemma. Consequently, the patients’ quality of life, functionality, treatment adherence, viral load suppression, and the overall survival rate of people with HIV is badly affected (Mekuriaw et al. 2023; Gras and Kaul 2010).

Other confounders of HAND diagnosis in Africa

Nutritional deficiencies are common in resource-limited populations: B-vitamins especially vitamin B6 deficiency is known to cause intractable epilepsy and cognitive decline (Kerac et al. 2014), and micronutrients such as iron and iodine also impair cognition independently (Kerac et al. 2014). Maternal malnutrition, which is very common in Africa also affect the brain development of the child and leads to a life time cognitive defect, for example folate deficiency (Kerac et al. 2014; Kirolos et al. 2022). Tropical infections like helminthiasis, malaria and neglected tropical disease like Leprosy and Buruli ulcer have all been implicated in cognitive decline in Africa (Fagan 2023). These co-factors necessitate careful covariate adjustment in African biomarker studies and informed clinical interpretation of any biomarker results. There is also the high cost of acquisition and maintenance of laboratory equipment for measuring these biomarkers.

Biomarkers of HAND: from discovery to clinical translation

Rationale and classification

Biomarkers are the objective, measurable indicators of normal biological processes, pathogenic processes, developmental changes or pharmacological responses to therapeutic intervention (Zou et al. 2015; Eliseo Eugenin, Jonathan Geiger et al. 2021). For the diagnosis of HAND in sub-Sahara Africa, the ideal biomarker should be detectable in peripheral blood to avoid invasive CSF sampling procedures. The ideal biomarker should be sensitive to subclinical neurodegeneration for early detection, specific to HIV-driven CNS injury, have a strong quantitative correlation with HAND severity and realistic in resource-limited clinical laboratories.

Currently no single marker yet fulfils all these criteria for an ideal biomarker. The evidence supports a multi-marker strategy in which panels of peripheral blood analytes from different biological dimensions of HAND, namely neurodegeneration, glial activation, immune activation, and synaptodendritic injury are combined to improve diagnostic accuracy. We group current HAND biomarkers into four main groups: (i) those that are secreted following immune activation and monocyte/macrophage; (ii) neuronal and glial cells injury markers; (iii) synaptic and extracellular vesicle markers; and (iv) genetic signatures. As presented in Table 3.

Table 3.

Summary of HAND Biomarkers: Evidence, Mechanism, and RLS Applicability

Biomarker(s)	Biomarker category	CNS Specificity	Correlation with HAND	Biological Mechanism	RLS Feasibility
CD4⁺ count; plasma viral load (HIV RNA)	Traditional HIV Markers	Low	Weak (HAND persists despite virological suppression) Zou and Fuss 2015	Reflects systemic immune status, not CNS-specific pathology	High (standard of care in all HIV clinics)
sCD163; sCD14; CXCL10; CCL2; sCD25; LPS	Monocyte/Macrophage Activation	Moderate	Moderate–Strong (inflammatory panel (CXCL10, CCL2, sCD163, sCD25) independently associated with worse cognition in African cohorts) Grant et al. 2014	Shed by activated monocytes/macrophages; reflects BBB breakdown, microbial translocation, and macrophage-driven neuroinvasion	Moderate (ELISA-based; established research assays; AFRICOS experience)
Neurofilament Light (NfL); GFAP	Neuronal Injury / Axonal Degeneration	High	Strong (plasma NfL consistently associated with HAND (p = 0.039); only CNS blood marker identified by systematic review) Mekuriaw et al. 2023	NfL: released from damaged axons into CSF/blood; GFAP: released from injured/activated astrocytes	Moderate (SIMOA platform required for plasma NfL; ELISA available for GFAP; increasingly accessible
sICAM-5 (Telencephalin)	Synaptic Injury	High (exclusively expressed in telencephalic neurons)	Strong (elevated in HAND; validated in virologically suppressed PWH) Li et al. 2021; Foerster et al. 2024	Shed from somatodendritic membrane during synaptic injury; reflects dendritic degeneration and neuroimmune crosstalk	High (standard ELISA from venous blood; no specialist neurological infrastructure required)
Serum ATP	Purinergic Signalling / BBB Integrity	Moderate	Strong — correlates with HAND severity; reflects BBB compromise and CNS neuroinflammation Gras and Kaul 2010	Pannexin-1 channel opening triggered by HIV–CD4/CCR5 binding releases ATP from T-lymphocytes; drives purinergic cascade and BBB disruption	High (detectable by luminescence or colorimetric assay from serum; no specialist equipment)
EV-Aβ42; EV-tau/Aβ42 ratio; Neuron-derived exosome cargo	Extracellular Vesicle (EV) Markers	High	Strong (CSF EV tau/Aβ42 elevated in HAND; not replicated for soluble markers)(Guha et al. 2024)	CNS-derived EVs carry neuronal/glial proteome into blood/CSF; EV cargo reflects amyloid and tau dynamics analogous to Alzheimer’s disease	Low (currently requires CSF sampling or ultracentrifugation; research setting only)
APOE ε4; hsa-miR-16-5p; hsa-miR-320a; RHOA; DLG4; APP	Genomic / miRNA Signatures	Moderate	Moderate — APOE ε4 associated with worse cognitive outcomes and lower CSF Aβ42 in HIV⁺ individuals Scutari et al. 2017; Kirolos et al. 2022	Genetic risk stratifier for ADLD overlap; synaptic gene dysregulation and microRNA-mediated post-transcriptional changes in neuronal networks	Low — PCR genotyping feasible for APOE in reference labs; miRNA panels require NGS
TMT-A/B; MoCA; IHDS; SDT; HDS	Traditional Neuropsychological Batteries	High — direct measurement of functional cognitive output	Gold standard — Frascati classification requires battery performance	Measures downstream functional consequences of CNS damage across multiple cognitive domains	Low — requires trained specialists; high cultural/educational/linguistic bias; inapplicable in low-literacy settings He et al. 1997

Immune activation and monocyte/macrophage markers

Inflammatory protein panel: CXCL10, CCL2, sCD163, sCD25

In a landmark study from the African Cohort Study (AFRICOS) presented at CROI 2024, Wilson and colleagues evaluated a four-protein inflammatory panel (CXCL10, CCL2, sCD163, sCD25) (Milicic et al. 2025). In 2,737 PWH across Nigeria, Kenya, Uganda, and Tanzania. They found that greater inflammation as reflected by this panel was independently associated with worse cognitive performance even in virologically suppressed PWH, while none of the markers associated with cognition in HIV-negative controls (Grant et al. 2014). This represents the most direct evidence to date that blood-based inflammatory biomarkers specifically track cognitive impairment attributable to HIV in an African cohort, a finding that could be urgently incorporated into African HAND diagnostic programs if validated and widely accepted by the scientific community especially in Africa.

Plasma LPS, and IFN-alpha

Microbial translocation secondary to HIV-driven intestinal barrier disruption elevates circulating LPS, which independently enhances HIV neuroinvasion via the p38 MAPK pathway and sustains monocyte activation (Tang 2020). Plasma IFN-alpha reflects innate immune activation. These markers represent indirect measures of CNS-directed immune activation, neuroinflammation, BBB permeability and reservoir burden (Campbell et al. 2020; Asiedu et al. 2020; Sarfo et al. 2021), but do not directly relate to HAND. However, because of their impact on the CNS when other causes of cognitive decline in HIV are eliminated these makers can aid in the diagnosis of HAND.

Soluble CD163 and soluble CD14

CD163 and sCD14 are shed from the surface of activated monocytes and macrophages and reflect systemic macrophage activation. a central event in HIV neuroinvasion. CSF sCD163 and sCD14 correlate significantly with CSF NfL across all HAND stages, demonstrating that peripheral macrophage activation is mechanistically linked to neuronal injury (Torti et al. 2011). The loss of CCR2 + non-classical monocytes has additionally been associated with cognitive impairment in ART-naive PWH (MS et al. 2024), reinforcing the monocyte axis as a key peripheral indicator of CNS disease.

Neuronal and glial injury markers

Neurofilament Light Chain (NfL)

NfL is a structural axonal component released into the CSF and detectable in peripheral blood following neuroaxonal damage (Brandenberg et al. 2015; Arslan 2023; Barro et al. 2020). Advances in ultrasensitive SIMOA platforms have enabled reliable plasma NfL quantification (Truffi et al. 2023; Revendova et al. 2025).

In the specific context of HAND, plasma NfL was significantly associated with HAND diagnosis (p = 0.039, Cohen’s d = 0.45) in a 2023 study (Rocha et al. 2024). The pathophysiology by which plasma NfL release from neuroaxonal degeneration is distinct from the inflammatory dimension by which sCD163/CCL2-class markers are released, justifying their combined use (Zeng 2023). NfL also correlates with EV-based CSF biomarkers (Guha et al. 2024), making it a useful bridge between blood-accessible and CSF-accessible biomarker dimensions. It is however a late stage disease marker for HAND (Eliseo Eugenin, Jonathan Geiger et al. 2021) and also Nfl is also elevated in many disease that results from neuroaxonal injury like demyelination diseases (Abdelhak et al. 2025) and even covid-19 (Zingaropoli et al. 2022) Alzheimer’s disease (Khalil et al. 2024), hence not very specific to HAND.

Glial Fibrillary Acidic Protein (GFAP)

GFAP, an intermediate filament released from activated and damaged astrocytes, is elevated in HIV+ individuals compared with HIV-negative controls and correlates with plasma NfL in PWH, suggesting coordinated neuronal and glial injury. Guha et al. (2024) found plasma GFAP associated with lower neurocognitive test scores, though in an adjunct rather than primary biomarker role (Guha et al. 2024). GFAP can be measured on the same ultrasensitive platforms as NfL, supporting their concurrent assessment. It is however not specific for HAND as it can be found in other diseases such as Alzheimer’s disease (Agnello and Maria 2025) and must be used cautiously.

Serum ICAM-5 (Telencephalin)

Cell adhesion is of the utmost importance in normal development and cellular functions (Yang 2012). Intercellular adhesion molecule-5, telencephalin, TLN (ICAM-5), a member of the ICAM family of adhesion proteins (Robb and Ananworanich 2016; Yang 2012). ICAM-5 (telencephalin), has been shown to be expressed only in the telencephalic neurons (Tian et al. 2008).

The developmental appearance of ICAM-5 parallels the time of dendritic elongation and branching, and synapse formation in the telencephalon (Yang 2012). As a somatodendrite-specific adhesion molecule, ICAM-5, not only participates in immune-nervous system interactions, it could also participate in neuronal activity, dendrites targeting signals, and cognition (Robb and Ananworanich 2016; Yang 2012). ICAM-5 suppresses the TCR-mediated activation of T cells as indicated by the decreased expression of CD69, CD40L, and CD25 (IL-2R) on T cells, especially CD45RO Low naive T cells. Moreover, sICAM-5 promotes the expression of transforming growth factor-1 (TGF-1) and IFN- gamma, but not TNF. This suggests that ICAM-5 is involved in immune privilege of the brain and acts as an anti-inflammatory agent (Tian et al. 2008).

A study by Yuan et al. (2017) and a review by Bandera et al. (2019) have shown that, sICAM5 might be a good biomarker for HAND (Yuan et al. 2017; Bandera et al. 2019). Serum sICAM-5 is detectable by ELISA from standard venous blood, requiring no specialist laboratory infrastructure, making it particularly attractive for resource-limited setting deployment. This biomarker however still remains to be validated in a multicenter African study, as the current studies are few and outside sub Sahara Africa (SSA).

Serum ATP and the pannexin-1 Axis

Another marker identified was Pannexin 1 protein. Pannexin 1 protein forms a large cell membrane channel which usually remains closed under normal physiological conditions and opens under certain pathological conditions including HIV infection.

In HIV infection, the binding of HIV to CD4 receptor and CXCR4 and/or CCR5 co-receptors, induces opening of Panx-1 channels, resulting in local ATP release through the pore and subsequent autocrine and paracrine purinergic receptor activation, which accelerates HIV entry into immune cells (Eliseo Eugenin, and Jonathan Geiger et al. 2021; Velasquez et al. 2019).

Studies by Velasquez et al. 2019, have shown that, the opening of pannexin 1 channel in PBMC leads to the movement of ATP into circulation and that the level of ATP correlate with HAND as the release of the ATP and other inflammatory mediators results in a compromised blood brain barrier BBB (Velasquez et al. 2019) They concluded that ATP is a good biomarker for HAND (Velasquez et al. 2019). Even though the study methodology was rigorous, the sample size and the site will require that such experiments are repeated in other settings so that it can be validated for global acceptance. Like sICAM-5, serum ATP is detectable from standard venous blood via luminescence or colorimetric assays without specialist neurological infrastructure. Together, sICAM-5 and serum ATP constitute a practical, blood-based biomarker pair well-suited to resource-limited clinical settings.

Extracellular vesicle-associated biomarkers

Extracellular vesicles (EVs), particularly exosomes, are nano-sized membranous particles carrying cellular proteins that mediate intercellular communication. In HAND, CSF-derived EVs carry a proteome of neuronal, astrocyte, myeloid, and choroid plexus markers reflecting CNS cellular status. In a review by Mu et al. (2025) there was evidence that neuron-derived exosomes (NDEs) accessible via peripheral blood serve as direct indicators of CNS pathological alterations, supporting non-invasive longitudinal CNS monitoring (Mu et al. 2025).

Guha et al. (2024) reported that in 98 virologically suppressed PWH on ART, CSF EV-associated Abeta42 was decreased and the CSF EV tau/Abeta42 ratio was increased in HAND compared to non-HAND, an association not observed for soluble (non-EV-bound) forms of these markers (Guha et al. 2024). Plasma NDE cargo may eventually provide a peripheral window into CNS amyloid and tau dynamics without CSF sampling, though this awaits validation.

Genetic signatures

Nguyen and Kim, in 2025 integrated GWAS, single-cell RNA sequencing of CSF, and brain tissue datasets to identify key HAND biomarkers including APOE, RHOA, DLG4, APP, and GAPDH — consistently altered across datasets (Nguyen and Kim 2025). Synapse-related gene products were significantly decreased while immune system biomarkers were increased, reflecting the dual neurodegenerative and neuroinflammatory nature of HAND. MicroRNAs hsa-miR-16-5p, hsa-miR-320a, and hsa-miR-335-5p were implicated, alongside transcription factors THRA and NEUROD6 (Kirolos et al. 2022). The APOE e4 genotype specifically associates with worse cognitive outcomes in HAND and moderates CSF amyloid-beta levels in HIV+ individuals (Scutari et al. 2017; Manu 2026). Given that APOE genotyping is feasible by PCR in sub-Saharan African reference laboratories, it may serve as a genetic stratifier for HAND risk in future African cohort studies, pending population-specific validation.

HAND epidemiology in sub-Saharan Africa: specific considerations

Prevalence landscape

Sub-Saharan Africa carries approximately 72% of the global HAND burden of approximately 11.6 million cases (Zou et al. 2015). Regional variation within Africa is substantial, with Central and Southern Africa reporting the highest pooled prevalence (49.33%), followed by East Africa (45.04%) and West Africa (42.40%) (Robb and Ananworanich 2016; Asiedu et al. 2020; Sarfo et al. 2021) These differences reflect the differences in ART coverage, TB, malaria, CMV, cryptococcal meningitis co-infection burden (Robb and Ananworanich 2016; Asiedu et al. 2020; Sarfo et al. 2021) baseline CD4 nadir distributions, and neurological diagnostic infrastructure. In Ghana, Sarfo et al. (2021) reported a combined HAND prevalence of 44.7%, and Asiedu et al. (2020) linked HAND in Accra-based PWH on ART to psycho-behavioral factors including depression and non-adherence (Asiedu et al. 2020; Russell et al. 2017; Hildebrand et al. 1993). A comprehensive 2025 review by Adhikary et al. (2025) evaluated HAND management globally and highlighted that early neurological assessment for individuals with poor educational attainment and social isolation is a priority (Adhikary et al. ; Jiang et al. 2021).

HIV-1 subtype heterogeneity

HIV-1 non-B subtypes (A, C, D, G, AG, and recombinant forms) predominate in sub-Saharan Africa but represent a minority of participants in most biomarker and neuropathological studies (Strimbu and Tavel 2011). Subtype C exhibits differential neurotropism, Tat-mediated transactivation, and co-receptor usage compared with subtype B, potentially yielding distinct neuropathological profiles (Strimbu and Tavel 2011). The extent to which biomarkers calibrated in predominantly B-subtype Western cohorts perform equivalently across African HIV-1 subtypes is unknown and constitutes a major research gap.

Proposed tiered diagnostic framework for resource-limited settings

Given the barriers to conventional HAND diagnosis and the emerging evidence for accessible blood-based biomarkers when validated, we propose a three-tier diagnostic framework for resource-limited HIV clinics. This frame work will however require much investment in the area of laboratory infrastructure and personnel training to tackle this problem.

• Tier 1: Routine Screening – (primary care HIV clinics):

  • Administration of brief, low-literacy cognitive screens (IHDS or SDT; 3–5 min; no specialist required). Measurement of serum ATP and sICAM-5 from routine venous blood draw. Abnormal results trigger Tier 2 evaluation.
• Tier 2: Enhanced Evaluation

  • More comprehensive neuropsychological battery appropriate for educational level (MoCA + SDT + 10-item word list). Plasma NfL assay (ELISA or SIMOA, reference laboratory). Inflammatory protein panel (sCD163, CXCL10) where available. Findings classified against Frascati criteria with local normative data.
• Tier 3: Specialist Referral

  • Full neuropsychological battery with locally validated normative data. MRI brain. CSF EV tau/Abeta42. APOE genotyping. Exclusion of confounding CNS co-infections and nutrition.

This framework prioritizes accessibility at Tier 1, designed for the primary care HIV clinics where the majority of PWH in sub-Saharan Africa receive ART. The inclusion of biomarkers at all tiers provides an objective anchor that reduces dependence on specialist expertise at first clinical contact — the level at which most HAND currently goes undetected.

Future directions

Point-of-care biomarker platforms

Translation of HAND biomarkers to resource-limited practice requires point-of-care (POC) formats. Lateral flow immunoassay formats for sICAM-5 and serum ATP have not yet been clinically validated, but could be technical feasibility. Portable electrochemical biosensors for ATP from whole blood have been demonstrated in proof-of-concept studies (Nishiyama et al. 2021). Compact SIMOA platforms are reducing infrastructure requirements for plasma NfL. Integration with existing HIV viral load monitoring infrastructure (e.g., Cobas 4800/6800) could provide a cost-effective implementation route.

Digital cognitive assessment

Tablet-based and smartphone-adapted cognitive tests calibrated for low-literacy populations represent a complementary approach. The AFRICOS study employed brief digital assessments across four African countries (Grant et al. 2014). These tools reduce administrator bias, enable standardized data collection, and can generate Africa-specific normative data at scale. Their integration with biomarker panels could constitute the next generation of Tier 1 HAND screening.

Machine learning and multi-marker panels

The diagnostic power of individual HAND biomarkers is modest. Multi-marker models combining NfL, sICAM-5, sCD163, CXCL10, and ATP in machine learning algorithms trained on African cohort data offer the prospect of substantially improved sensitivity and specificity. Multi-domain marker integration (neurodegeneration, glial activation, immune activation, synaptic injury) substantially outperforms any single marker in comparable neurological conditions. The AFRICOS data provide the largest existing African platform for developing such models.

Research priorities

Based on this review, the following research priorities are identified for HAND in sub-Saharan Africa:

1. Prospective validation of sICAM-5, serum ATP, and plasma NfL in large, well-characterized African HIV cohorts with comprehensive cognitive phenotyping using locally validated assessment tools.

2. Calibration of biomarker reference ranges across HIV-1 non-B subtypes (particularly C, A, D and CRF) prevalent in sub-Saharan Africa.

3. Development of locally validated normative neuropsychological data for Ghana and West Africa across literacy and education strata.

4. Integration of APOE genotyping and inflammatory protein panels into existing African cohort platforms (AFRICOS, H-CRIS, NIMH NeuroAIDS Africa network).

5. Head-to-head comparison of emerging POC biomarker platforms against ELISA-based standards under African resource-limited conditions.

6. Studies examining the effect of TB, malaria, and other African-prevalent co-infections on HAND biomarker performance.

7. Long-term prospective studies examining whether early biomarker-guided HAND detection and intervention improve ART adherence, viral load suppression, and quality of life in African PWH.

Conclusion

HAND remains a highly prevalent and clinically challenging complication of HIV infection in the ART era, with a disproportionate impact in sub-Saharan Africa where approximately 11.6 million of the world’s 16 million HAND cases reside. The current diagnostic standard; specialist-administered neuropsychological testing classified by the Frascati criteria is neither scalable to resource-limited African clinical settings nor free from the cultural, educational, and infrastructural biases that produce the wide variability in Africa-specific prevalence estimates.

The biomarker landscape has advanced substantially. Serum ICAM-5 and serum ATP, detectable from standard venous blood by ELISA and colorimetric/luminescence assays respectively provide complementary peripheral signals of synaptic injury and BBB compromise. Plasma NfL has emerged as the most consistently validated peripheral neurodegeneration marker in HAND. Inflammatory protein panels validated in African cohorts (CXCL10, CCL2, sCD163, sCD25) provide an immune-activation dimension that complements structural markers. Emerging EV-associated amyloid/tau biomarkers and genomic signatures (APOE, miR-16-5p) further enrich the toolkit, though their current RLS applicability is limited to research settings.

No single biomarker yet provides sufficient diagnostic accuracy to replace clinical assessment. From the evidence instead, we propose a tiered, multi-marker diagnostic framework in which peripheral blood biomarkers serve as objective, accessible anchors at every level of HIV care. from primary care screening using serum ATP and sICAM-5 to tertiary-Centre CSF EV biomarker profiling and APOE genotyping. Such a framework could improve HAND diagnosis from a late diagnosis of exclusion made by specialists to an early, objective, clinic-level diagnosis made where ART is dispensed.

Achieving this requires urgent investment in Africa-specific biomarker validation, locally adapted cognitive assessment tools with validated normative data, and point-of-care platform development suited to the infrastructure realities of sub-Saharan Africa.

Author contributions

PKA, CO, and GBK conceptualized the manuscript; PKA and MEH wrote the draft manuscript. EYB and JOA reviewed and edited the manuscript and made intellectual inputs. PKA and MEH revised the draft manuscript for further review and proofreading. JOA, CO, GBK and EYB reviewed and proofread the manuscript. All authors reviewed the edited manuscript and approved the final version.

Funding

This article is part of the H-Cure-RT project supported by the Fogarty International Center of the National Institutes of Health under Award Number D43TW012774. EYB was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Number R21AI174880. This work was also supported in part by the European Union/EDCTP (Grant No. TMA2017SF-1955, H-CRIS) awarded to GBK. The content is solely the responsibility of the authors and does not represent the official views of the funders.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval

Ethical approval was not required for this review article.

Competing interests

The authors declare no competing interests.