Alzheimer’s disease (AD) is a progressive neurodegenerative disorder (Dubois et al., 2021). Ongoing development in the research of AD pathology has expanded the number of fluid (e.g., cerebral spinal fluid [CSF], plasma) biomarkers recognized in the screening, diagnosis, and monitoring of AD (Blennow and Zetterberg, 2018). Current revised 2023 Alzheimer’s Association (AA) criteria differentiates between two broad categories of AD fluid biomarkers related to AD pathogenesis: (1) core AD fluid biomarkers (the CSF ratio of amyloid-β [Aβ42/40], phosphorylated tau-181 and plasma 217 [p-tau181, p-tau217]) and (2) non-specific biomarkers involved in other neurodegenerative disorders pathology, including neurofilament light (Nfl) and glial fibrillary acidic protein (GFAP) (AAIC, 2024). Although not included in the AA core-criteria of the final document, neuroinflammatory/immune biomarkers, interleukin 1b and 10 (IL-1b and IL-10) and tumor necrosis factor α (TNFα), play an important role in other neurodegenerative (Blennow and Zetterberg, 2018; Lyra Silva et al., 2021; Kim et al., 2023; Gulisano et al., 2018; Giacomucci et al., 2022; Babić Leko et al., 2020).

Reference values (RV) of these plasma biomarkers within non-White samples is not yet established, as most studies have been primarily conducted using largely non-Hispanic White (NHW) individuals (Gonzales et al., 2021). There appears to be some evidence of ethnoracial differences in the levels of plasma AD biomarkers and their diagnostic precision based on age and sex (Mohs et al., 2024; Hajjar et al., 2022; Mielke et al., 2022; Hall et al., 2021), emphasizing the need for further research examining these associations. Given the intra- and inter-assay variability, as well as inconsistencies and differences in methods, defining a set of universal cut-offs for plasma AD biomarkers is challenging and may not be possible (Pais et al., 2023). It has been recommended that studies define cut-offs in-house to best represent specific populations, with the recognition that the establishment of contemporary clinical cut-offs will need to be assay-dependent (Pais et al., 2023). Alternatively, calibration equations could be developed to harmonize biomarkers across assays and labs; however, such activities require complicated sample exchanges which can be challenging in cross-national research.

Some studies have established reference intervals, values, and cutoffs based on demographic characteristics such as age and sex. In a sample of healthy Chinese older adults, Chen et al. (2023) identified reference intervals for specific age groups (i.e., 50–59, 60–69, 70–79, 80–89), as well as sex differences. Within their sample, there were no differences between women and men in plasma Aβ42, Aβ40, or Aβ42/40 ratio; however, men had greater plasma p-tau181, p-tau181/t-tau ratio, and p-tau181/Aβ42 ratio than women (Chen et al., 2023). In general, age and sex-specific cut-offs for plasma biomarker diagnostic and prognostic use may be important, particularly given the analytical variability and difference in plasma biomarkers’ concentrations across ethnoracial groups. Plasma biomarker cut-off values can vary not only based on analytical variability or ethnoracial variables, but by demographic factors such as age and sex. Thus, these two factors should also be considered when stratifying plasma AD cutoffs (Mielke et al., 2022; Cooper et al., 2023).

Characterizing plasma biomarkers and their diagnostic precision within Sub-Saharan African (SSA) populations is important, as plasma biomarkers are more cost-effective and easily accessible. Our study aims to provide preliminary RV for plasma protein biomarkers including Aβ42/40 ratio, p-tau181, p-tau217, Nfl, GFAP, IL-1B, IL-10, and TNFα in a sample of adults in the Democratic Republic of Congo with and without dementia to aid in screening, future diagnostic utility, prognostication, and management of AD in the DRC. We hypothesized that in this SSA sample of Congolese, core AD plasma biomarkers concentrations will not be influenced by age or sex. However, we hypothesized that non-specific AD biomarkers show a significant difference across age in this sample. We anticipate that some of these markers, such as p-tau181 or p-tau217, will have potential diagnostic value in this population.

Demographic data, cognitive scores, clinical data, and plasma biomarker concentrations stratified by neurological status are presented in Table 1. As expected, there were significant differences in cognitive screening scores used in distinguishing neurological status, with healthy individuals having better scores than those with dementia. For clinical data, only HbA1c levels showed a significant difference between HC and dementia, with HC having higher levels of HbA1c than suspected dementia. Diagnostic groups differed in mean levels of Nfl and GFAP after controlling for age and sex (Table 1).

Figure 2 shows the distribution [minimum, 25th (q1), 50th (q2), 75th (q3), and the maximum], variability, and the skewness of each plasma biomarker. Aβ42, Aβ40, and p-tau 181 are nearly normally distributed, whereas p-tau 217, GFAP, Nfl, TNFα, IL-1b and IL-10 are right skewed.

The concentrations of plasma biomarkers, except for p-tau181 and GFAP, did not significantly differ by age (Table 2). Plasma p-tau 181 concentrations were significantly higher in participants aged 77 years and older [3.2 (2) pg./mL] than in participants aged 50–69 years [1.8 (1) pg./mL] or 70–76 years [2.9 (2) pg./mL]. In addition, plasma GFAP concentrations were significantly higher in participants aged 70–76 years [238.8 (154) pg./mL] and 77 years and more [234.0 (115.0) pg./mL] than in participants aged 50–69 years [143.0 (94) pg./mL]. The concentration of Nfl was higher in participants aged 70–76 years [46.7 (35) pg./mL] and 77+ [56.31 (39) pg./mL] compared to those aged 50–69 [48.3 (43) pg./mL], though not significant. P-tau 217, TNFα levels and IL-10 were variable and also not significant (Table 2).

Aβ42/40 p-tau 217 values do not increase with older age, while Nfl, GFAP and IL-1B values also increase with age. As age group increases, p-tau 181 values increase. There is a rather large spread of values per age group. IL-10 values decrease as age increases (see Figure 3).

Concentrations of plasma biomarkers were not significantly different between men and women (Table 3).

The distribution of biomarker concentrations by sex is presented in Figure 4. Biomarker concentrations followed a normal distribution or were positively skewed.

Table 4 displays the AUC values for prediction of dementia based on plasma biomarkers. AUC ranges fell between 0.64–0.74 (95% CIs ranging from 0.52–0.86). Aβ42/40, p-tau 217, and IL-10 had higher sensitivity than other plasma biomarkers, followed by IL-1b, GFAP, and TNFα. P-tau 217 [0.72 (0.59, 0.84)], GFAP [0.72 (0.61, 0.83)], and Nfl [0.73 (0.62, 0.84)] had the highest AUC values.

The revised and updated AA criteria have brought major changes to the diagnosis of AD dementia from a purely cognitive diagnosis to a biological clinical diagnostic algorithmic approach (AAIC, 2024). The presence of CSF or plasma amyloid and p-tau 181 or 217 (mostly p-tau 217) has been considered as sensitive and specific to AD with Nfl and GFAP as important non-specific AD biomarkers. The current study has provided preliminary RVs for plasma protein biomarkers, including Aβ42/40 ratio, p-tau181, p-tau217, Nfl, and GFAP, IL-1B, IL-10, and TNFα in a sample of adults in the DRC with and without dementia.

This research explores the clinical performance of established plasma AD and neurodegeneration biomarkers in an African population for which there are no previous biomarker data. Our first hypothesis was partially supported, as age groups did not significantly differ in core AD biomarkers and non-inflammatory/immune AD biomarkers except for p-tau 181 across ages in this SSA sample of Congolese adults. In contrast, non-specific AD biomarkers showed significant age differences, aside from GFAP.

As we predicted, there were no significant differences between women and men in all plasma biomarkers in this sample. Similar findings were reported by a Canadian population-based cohort which also found no significant sex differences in plasma AD biomarkers (Cooper et al., 2023). Similarly, in a sample of healthy Chinese, Chen et al. did not find significant sex differences in plasma Aβ42, Aβ40, or Aβ42/40 ratio; however, men had greater plasma p-tau181, p-tau181/t-tau ratio, and p-tau181/Aβ42 ratio than women (Chen et al., 2023).

One interesting finding is the lack of predictive abilities of core plasma biomarkers to classify subjects as having AD pathology. We found a difference between clinical diagnosis based on cognitive tests and diagnostic classification based on AD core plasma biomarkers (Aβ42/40 and p-tau). However, non-specific biomarkers (e.g., Nfl and GFAP) and p-tau 217 had good AUC. Pais et al. also found discrepancies between cognitive decline and the diagnostic classification based on AD biomarkers in many studies (Pais et al., 2023). Future research should continue to explore the predictive value of plasma biomarkers in various ethnically and culturally diverse samples. Prior research has shown variation in cutoffs by ethnic group. A review by Pais et al. (2023) found that plasma AD biomarker cut-off values can vary not only based on analytical variability, but by demographic factors, which can explain the variation of plasma biomarkers across ethnoracial groups, highlighting the importance of studying the underlying pathophysiology in these groups (Mielke et al., 2022; Pais et al., 2023; Cooper et al., 2023). Thus, in-house cut-offs may better represent specific populations with the understanding that they are assay-dependent (Pais et al., 2023). Establishment of clear cut-off criteria is important for potential future clinical utility.

This study is the first in the SSA to attempt to provide preliminary RVs for core and non-specific AD plasma protein biomarkers in a sample of adults in the DRC with and without dementia. Our findings should be interpreted considering several limitations, such as the cross-sectional nature of the study, limited sample size, and lack of amyloid PET imaging or CSF biomarker measurements confirming AD pathology. These RVs should be further validated in longitudinal studies with larger sample size. Furthermore, this is the first study in the Congo where the population can be phenotyped in biofluids. Future research could benefit from sending the same samples for plasma p-tau217 via LabCorp or C2N (which are commercially available and have established cut points), to compare the performance. The screening measures used (CSID and AQ) have not been validated in the SSA/DRC. To that extent, there have been recent studies looking at these relationships with more commonly used cognitive screeners, such as the Mini Mental Status Exam (MMSE) and Montreal Cognitive Assessment (MoCA) across the globe and in different diagnostic groups (Jiao et al., 2020; Mizutani et al., 2023). This study included only subjects with suspected dementia and healthy controls. Those with cognitive difficulties seen in between these two categories (e.g., MCI, subjective memory complaints) were excluded, leaving only the extremes of the dementia spectrum. Future studies should conduct statistical analyses across all 4 groups (healthy controls, MCI, subjective memory complaint and dementia). In the same vein, this study did not characterize the etiology of the dementia syndrome. AD biomarker performance is best in amnestic or logopenic phenotypes. Thus, if our sample had a mixture of executive, behavioral, or mixed phenotypes, the diagnostic accuracy of the plasma AD biomarkers would be compromised. For example, plasma AD biomarkers correlate well with measures of verbal or visuospatial memory or screening tests that rely heavily on memory (e.g., MMSE, MoCA) (Jiao et al., 2020; Mizutani et al., 2023; Lim et al., 2020; Hanon et al., 2018). Future studies should also aim to replicate our findings using AD biomarkers in CSF, amyloid or tau brain PET, or mass spectrometry of plasma biomarkers. Additionally, Simoa has limitations for the measurements of plasma AD biomarkers (Pais et al., 2023). Thus, continued investigation into racial differences in AD biomarkers and relation to AD-dementia using these gold standard techniques (e.g., brain amyloid PET, CSF) should be conducted. Finally, the findings of this study are exploratory, and we caution that evaluation of these biomarkers in novel populations to support clinical assessments may not be as straightforward as expected. Replication of findings on large scale sample of controls is warranted.