Abstract
Importance Integrating behavioural assessments with blood-based biomarkers (BBM) could improve diagnostic accuracy for MCI linked to early-stage neurodegenerative disease (NDD).
Objective This study investigates the potential of combining neuropsychiatric symptoms (NPS) with BBM to enhance the differentiation between older adults with Mild Cognitive Impairment (MCI) and those with Normal Cognition (NC) in a multi-ethnic Southeast Asian cohort.
Design, Setting, and Participants This cross-sectional study analyzed baseline data from the Biomarkers and Cognition Study, Singapore(BIOCIS). Data from 678 participants (mean age = 59.16 years, 39.50% males) with NC and MCI were included.
Main Outcomes and Measures Behavioral symptoms were assessed using the Mild Behavioral Impairment Checklist (MBI-C) and Depression, Anxiety, and Stress Scales (DASS). Blood samples were analyzed for amyloid-beta (Aβ40, Aβ42), phosphorylated Tau (p-tau181), neurofilament light (NfL) and glial fibrillary acidic protein (GFAP). Regression models adjusted for age, education, gender, cognitive status (CS) and APOE-ε4 status were used. Discriminative power was evaluated using the area under the curve (AUC) to assess the combined predictive accuracy of behavioral and biological markers for CS, i.e., MCI status over CN.
Results The study included MBI-C scores (total, interest, mood, control) and BBM levels (Aβ40, NfL, GFAP) were significantly higher in MCI group, compared to CN group. Elevated GFAP (OR:3.636, 95% CI:1.959, 6.751, p<0.001) and higher MBI-C-Mood scores (OR:2.614, 95% CI:1.538, 4.441, p<0.001) significantly increased the likelihood of MCI. The combined model, integrating NPS and BBM markers, showed strong discriminative ability for MCI (AUC = 0.786), with 64.7% sensitivity and 84.9% specificity at a threshold of 0.616, compared to NPS markers (AUC: 0.593) or BBM (AUC: 0.697) alone.
Conclusions and Relevance The combined use of BBM and NPS achieved optimal accuracy in distinguishing MCI from NC, with strong associations between GFAP, MBI-C Mood scores, and CS. These findings underscore neuroinflammation and mood disturbances as critical factors in early NDD, supporting the importance of dual-dimension screening strategies. Integrating NPS and BBM represents a novel and effective diagnostic approach for detection of MCI due to AD or other dementias. The integrated framework, leveraging both pathophysiological and neuropsychiatric markers, facilitates earlier diagnosis, potentially improving clinical decision-making and enabling targeted disease-modifying therapies for individuals with neurodegenerative disorders.
Introduction
With the advancement of dementia treatments comes the responsibility to leverage biomarkers to identify cognitive impairment secondary to neurodegenerative disease (NDD) accurately at earlier stages. Plasma-derived and behavioural-biomarkers could play a crucial role for the early detection of NDD-related cognitive impairment, offering both accessibility and a holistic perspective on mechanistic and pathological changes, as well as observable symptoms, that occur during the onset of cognitive decline(CD). Current research demonstrates the value of blood-based biomarkers(BBM) in detection of NDD, enhancing traditional assessment tools such as neuropsychological assessments, positron emission tomography(PET) scans and neuroimaging.
The integration of behavioral-biomarkers—especially those reflecting mood and personality changes—can add significant diagnostic precision for Alzheimer’s disease(AD) and related dementias. Neuropsychiatric symptoms(NPS) are non-cognitive, behavioral, or psychiatric symptoms that often accompany or precede CD1. A notable subset of NPS is Mild Behavioral Impairment(MBI), characterized by the later-life onset of persistent behavioral symptoms that represent change from baseline, which has emerged as a reliable predictor of CD and incident dementia2–4. The Mild Behavioral Impairment Checklist(MBI-C) was developed explicitly to measure MBI, and comprises five domains including decreased motivation, emotional dysregulation, impulse dyscontrol, social inappropriateness, and abnormal perception or thoughts5. Current research demonstrates that the MBI-C effectively distinguishes NPS linked to to NDD from NPS related to psychiatric or stress-related etiologies, thereby enhancing dementia risk assessment6–8. Complementarily, the Depression, Anxiety, and Stress Scales(DASS) 9 provide a comprehensive evaluation of mood symptoms in dementia increasingly recognized as early indicators of CD10,11. Together, the MBI-C and DASS offer a holistic approach, capturing a broad spectrum of behavioural and psychological changes that correlates with cerebrospinal fluid(CSF)12, plasma biomarkers13,14, and cognitive impairment15,16.
In parallel with behavioral assessments, BBM are transforming AD diagnosis by providing a minimally invasive and accessible alternative to traditional CSF and PET biomarkers17–21. Recent research shows that BBM achieve high accuracy in predicting clinical decline in individuals with mild cognitive impairment(MCI), highlighting their potential as early indicators22,23. A Swedish study confirmed that BBM targeting amyloid-beta and phosphorylated tau achieve diagnostic accuracy comparable to CSF and PET, with predictive values exceeding 90%, demonstrating their clinical utility in various care settings24. Compared to CSF biomarkers, BBM offer significant practical advantages as they are highly scalable, less invasive and promote higher patient compliance. BBM are therefore more feasible for large-scale screening and triage in under-resourced settings, while facilitating the integration of anti-amyloid therapies into routine clinical practice25.
Within the Amyloid, Tau, and Neurodegeneration(ATN) framework, amyloid-beta 42(Aβ42), amyloid-beta 40(Aβ40), and the Aβ42/Aβ40 ratio are key markers of amyloid plaque accumulation, a defining feature of AD pathology26,27. For tau pathology, phosphorylated Tau 181(p-tau181) serves as a marker for neurofibrillary tangles associated with AD progression 28–30. Additionally, Neurofilament light(NfL) provides insights into neuroaxonal damage, while Glial fibrillary acidic protein (GFAP) is associated with reactive astrogliosis and neuroinflammation31–33. Alongside these ATN biomarkers, genetic factors such Apolipoprotein E ε4(APOE4) genotype play a critical role in AD risk and biomarker expression. APOE4 is strongly linked to increased amyloid-beta accumulation and is one of the most significant genetic risk factors for late-onset AD. Notably, populations in Southeast Asia exhibit a lower prevalence of APOE4 compared to Western populations 34, potentially explaining differences in amyloid positivity, CD rates, and dementia types observed between these groups 35. Recently, the European Medicines Agency, stipulated that lecanemab could not be prescribed in APOE4 homozygotes, further supporting its importance 36. Together, these BBM and blood-based genetic factors provide a multifaceted view of neurodegeneration, offering deeper insights into CD and AD risk and progression.
Given the capacity to capture diverse aspects of neurodegeneration, BBM hold significant potential for elucidating the relationship between behavioral symptoms and the progression of NDD. The link between NPS and BBMs is still emerging, with research to date yielding mixed results. For instance, in a large population-based study, no difference in AD risk was observed between APOE4 carriers and non-carriers among Americans without baseline depression37. For amyloid-related biomarkers, amyloid-beta has been more extensively studied in relation to behavioral symptoms, particularly depression38. A study in Vienna reported that higher baseline plasma Aβ42 levels predicted incident depression and conversion to AD over five years39. Similarly, the Rotterdam Study reported cross-sectional links between high Aβ40 and depressive symptoms in prodromal dementia, but longitudinally associated lower Aβ40 and Aβ42 with increased depression risk in elderly individuals without dementia, indicating a complex role of Aβ peptides in the etiology of depression40. For tau biomarkers, elevated plasma p-tau181 has been associated with NPS such as appetite changes and disinhibition41. Regarding neurodegeneration biomarkers, NfL has shown promise as a marker for multiple NPS, including aberrant motor behavior, anxiety, sleep disturbances and euphoria42. GFAP, another marker of neurodegeneration, has been associated with depression43.
Focusing on MBI, characterized by late-life emergent and persistent NPS, findings consistently link it to key BBM in the ATN framework. For amyloid-related biomarkers, global MBI status and MBI affective dysregulation demonstrated associations with amyloid-beta44. In terms of tau pathology, recent findings by Ghahremani13 indicate MBI correlates with elevated plasma p-tau181 levels and a nearly fourfold increased risk of dementia, with higher baseline and longitudinal increases in p-tau181 in individuals with MBI compared to those without MBI. Additionally, Gonzalez_Bautista14 linked MBI domains, such as impulse dyscontrol and emotional dysregulation, to p-tau181 and metabolic biomarkers, implicating tau alongside metabolic disruption in preclinical and prodromal AD. In neurodegeneration, findings from the Alzheimer’s Disease Neuroimaging Initiative(ADNI) cohort revealed that in preclinical and prodromal samples, MBI and its interaction with time predict NfL changes, indicating MBI as an early marker of accelerated neurodegeneration and CD45. Collectively, these studies illustrates MBI’s value as an early marker of AD pathology and dementia risk.
The growing research on BBM, MBI, and NPS at early stages of CD underscores their potential for early disease detection. When combined, these markers offer a multifaceted approach to early diagnosis, capturing both pathobiological and psychological dimensions of dementia onset. Despite this prospect, a comprehensive study systematically combining these markers for detecting cognitive impairment has, to our knowledge, not yet been conducted.
This study evaluates the predictive utility of BBM and behavioral-biomarkers in distinguishing persons with MCI from those with normal cognition. By systematically analyzing MBI-C, DASS and BBM in a multi-ethnic Southeast Asian cohort, we assess the combined value of these measures for higher diagnostic accuracy for MCI linked to NDD and to explore underlying mechanisms of cognitive impairment.
Methods
Participants
Participants were drawn from the Biomarkers and Cognition Study, Singapore(BIOCIS), a 5-year longitudinal study with annual assessments. Recruitment is community-based, targeting individuals aged 30 to 95 years in Singapore. To explore how pathology presents with or without minimal clinical symptoms, both cognitively impaired and unimpaired individuals are included. Participants were classified as Cognitively Normal(CN) or as having MCI based on Peterson’s46 criteria and National Institute on Aging-Alzheimer’s Association(NIA-AA) guidelines47 after completion of a validated neuropsychological test battery. Further study design and methodological details are available in the BIOCIS protocol paper48.
Blood based biomarkers
Venepuncture was performed by a certified phlebotomist at baseline study visit. Blood samples were collected in EDTA vacutainers, left at room temperature for 30 minutes, then centrifuged at 2000 g for 10 minutes at 4°C. The resulting plasma was aliquoted and stored at –80°C. Genomic DNA was extracted from whole blood samples, collected into EDTA vacutainers(Becton Dickinson) using the QIAamp DNA Blood Maxi Kit(Qiagen). The concentration and purity of the DNA were assessed using a Nanodrop One spectrophotometer (Thermo Fisher Scientific, United Kingdom). DNA was analysed using either the StepOne plus or QuantStudio 7 Pro Real-Time Polymerase Chain Reaction analyser(Applied Biosystems) to determine the allelic variants of APOE. APOE genotypes were determined using SNPs rs429358 and rs7412(Life Technologies), as per manufacturer’s protocol, using a 96-well MicroAmp Fast(StepOne Plus)/Optical(Quanstudio) reaction plate(Life Technologies), with 10ng of DNA and either one of the 2 SNPs, in a 10ul TaqPath ProAmp master mix reaction(Applied Biosystems)49,50. Results were analysed using the Design and Analysis 2.5.1 Real Time PCR system software(Applied Biosystems) and respective APOE genotype were rated independently by raters. Quanterix’s Single Molecule Array(Simoa) digital biomarker technology platform was used to quantify all plasma biomarkers(NfL, GFAP, Aβ40, Aβ42, p-tau181)(Quanterix, Billerica, MA, USA). The Neurology 4-PlexE(NfL, GFAP, Aβ40, Aβ42) and p-tau181 Advantage V2.1 kits were utilized for expression analysis on the HD-X Analyzer platform(Quanterix, Billerica, MA, USA) in accordance with manufacturer’s protocol51.
Behavioural profiles and Neuropsychiatric symptoms
Self-reported questionnaires were administered including MBI-C and DASS. The MBI-C5 evaluates MBI across five NPS domains: Mood/Anxiety, Apathy/Drive, Impulse Dyscontrol, Social Inappropriateness, and Abnormal Perception/Thought Content, with symptoms required to persist for at least six months and represent a change from longstanding baseline. The DASS9 assesses depression, anxiety, and stress over a one-week timeframe.
Statistical Analyses
Continuous variables were summarized using mean and standard deviation, while categorical variables were described using frequency and percentages. To address skewness, logarithmic transformations were applied to blood and behavioral data, with a constant of 1 added to continuous variables that included values below 1. Two independent sample t tests were employed to compare continuous variables between the CN and MCI groups. Categorical variables were compared between CN and MCI using Chi-square test.
Logistic regression models were performed to investigate the association of baseline characteristics and cognitive diagnosis(MCI vs CN) while adjusting for age, years of education, gender and APOE status. The results were reported as odds ratio(OR), 95% confidence interval(CI), and p value. A dual-regression approach was implemented, consisting of two multivariable logistic regression models with cognitive status(MCI vs CN) as outcome. The first regression(Model 1), included behavioral parameters only as continuous variable predictors and the second(Model 2) included BBM only as continuous variables, both adjusted for demographic variables and APOE status. Within each model, a backward variable selection approach identified variables independently associated with cognitive diagnosis. Subsequently, the shortlisted variables from both models were included in a final multivariable logistic regression model(Combined Model), adjusted for demographic variables and APOE status. Further, overall discriminative power of the final model was assessed via area under the curve(AUC) analyses with model performance evaluated using Hosmer and Lemeshow lack-of-fit test. For all models, variance inflation factor statistics and independence of errors were utilized to confirm the absence of multi-collinearity among predictor variables.
To further understand the association between behavioral parameters, cognitive diagnoses, and BBM, additional logistic regressions were implemented. Predictors were behavioural-biomarkers as continuous variables, adjusting for demographics, APOE status, and cognitive diagnosis (Model 4). Outcomes were blood biomarker status(i.e., high vs low), based on dichotomization using a median split 52,53.
All statistical analyses were conducted using IBM SPSS Statistics, Version 29.0(Armonk,NY:IBM Corp.), with statistical significance set as p<0.05 unless otherwise specified.
Results
Demographics
678 participants were recruited. The MCI group was older(p<.001) and had fewer years of education(p<0.001). For behavioral assessments, MBI-C-Total scores were higher in the MCI group(p=0.006), with significant differences in the interest (p=0.005), mood(p=0.005), and control(p=0.009) subdomains. Additionally, BBM analyses showed higher mean levels of Aβ40(p< 0.001), NfL(p< 0.001), and GFAP(p< 0.001) in the MCI group compared to the CN. Table 1 summarizes the BIOCIS demographic variables alongside behavioral and BBM data.
Behavioral and BBM as Independent Predictors of MCI Over CN
Table 2 summarizes the results of the multivariable logistic regression analyses, detailing significant behavioral and blood-based predictors of MCI over CN. In the behavioural model(Model 1), for each 1-point increase in the MBI-C-Mood subdomain, the odds of having MCI over being CN increased by approximately 59%(OR:1.587,95%CI:1.127–2.236,p=0.008). Similarly, for each 1-point increase in the DASS Depression score, the odds of MCI over CN increased by 25%(OR:1.250,95%CI:1.013–1.542,p=0.037). In contrast, higher DASS-Stress scores were inversely associated with cognitive impairment, with each 1-point increase in DASS-Stress reducing the odds of MCI by 20%(OR:0.800,95%CI:0.660–0.970,p=0.023). The AUC for behavioral-markers alone was 0.593(95%CI:0.55–0.636).
Among BBM(Model 2), elevated NfL and GFAP levels were strongly associated with the likelihood of being classified as MCI over CN. Each unit increase in NfL corresponded to a 110% increase in MCI likelihood over CN(OR:2.096,95% CI:1.365–3.217,p=0.001). GFAP showed an even stronger association, with each unit increase raising the odds of MCI over CN by approximately 199%(OR:2.987,95%CI:1.913–4.664,p<0.001).
Combined Model of Predictive Value of Behavioral Scores and Blood Biomarker Levels for Cognitive Status
In the behavioural and blood model(Combined Model), variables from Table 2(i.e.,variables from the Model 1 and 2 meeting statistical significance) were included along with demographics and APOE status as clinical confounders. Higher GFAP(OR:3.636,95%CI:1.959–6.751,p< 0.001) and MBI-C-Mood(OR:2.614,95%CI:1.538–4.441,p< 0.001) significantly increased the likelihood of MCI by 263.6% and 161.4% respectively. The model AUC was 0.786(95%CI:0.744–0.827), surpassing separate models based on either behavioral-markers(AUC:0.593,95%CI:0.55–0.636) or BBM(AUC:0.697,95% CI:0.658–0.736). At an optimal threshold of 0.616, the Combined model achieved 64.7% sensitivity and 84.9% specificity, demonstrating strong discriminatory ability in distinguishing between CN and MCI individuals(Figure 1).
Receiver Operating Characteristic (ROC) Curves Comparing Different Models for Distinguishing Between Cognitively Normal and Mild Cognitive Impairment Groups
Behavioural and Cognitive Diagnoses in Association with Dichotomized BBM Levels
For Model 4, with continuous behaviour scores as predictors and biomarker status as binary outcome, a higher MBI-C-Total score was significantly associated with odds of high Aβ42 status(OR:1.437,95%CI:1.127–1.831,p=0.003). Specific MBI-C subdomains were significantly linked to likelihood of high Aβ42: MBI-C-Interest(OR:1.800,95%CI:1.139–2.845,p=0.012), MBI-C-Mood (OR:1.587,95% CI:1.068–2.359,p=0.022), and MBI-C-Control(OR:1.782,95%CI:1.242–2.557,p=0.002). Higher DASS-Stress scores were significantly associated with likelihood of high Aβ40 levels (OR:1.252,95%CI:1.039–1.508,p=0.018). MBI-C-Total score was associated with likelihood of high p-tau181(OR:1.279,95%CI:1.001–1.634,p=0.049). Additionally, higher DASS-Depression scores(OR:1.302,95%CI:1.061–1.598,p=0.012) and higher DASS-Stress scores (OR:1.238,95%CI:1.020–1.503,p=0.031) significantly increased the odds of high p-tau181. Lastly, higher DASS-Anxiety scores were marginally associated with probability of high GFAP levels(OR:1.260,95%CI:1.000–1.587,p=0.05). Table 3 presents a detailed summary of these results.
Discussion
Integration of Biomarkers for Early Detection of Cognitive Impairment
The results demonstrate that integrating blood-based and behavioral measures enhances prediction of MCI status compared to using either measure alone. Specifically, models that combine both BBM and behavioral data demonstrate the highest accuracy in identifying MCI status, with BBM alone outperforming behavioral measures when used in isolation. These findings present three key insights: behavioral measures contribute to cognitive status prediction and correlate with BBM, BBM alone serve as robust predictors of MCI, and the combined model provides superior predictive power over individual approaches.
Behavioural-biomarkers as Indicators of MCI status
Higher scores on the MBI-C-Total and subdomains—specifically Interest, Mood, and Control—were linked to increased likelihood of MCI. The robust association between mood disturbances and cognitive impairment is particularly compelling, as mood symptoms often precede CD and reflect underlying disruptions in neural circuits involved in emotional regulation54,55. Interestingly, lower DASS Stress scores were inversely associated with cognitive impairment, possibly because moderate stress enhances attention and memory, consistent with the Yerkes-Dodson law56. Collectively, these findings underscore that behavioral and mood-related symptoms may not only serve as early predictors but also actively contribute to the acceleration of CD. This assertion is well-supported by prior research on the prodromal stages of AD, which suggests early NPS play a significant role in modulating the disease’s trajectory57–59.
Behavioural-Biomarkers as Early Indicators of Neurodegeneration
Higher MBI-C scores were significantly associated with elevated Aβ42 levels, particularly within the subdomains of Interest, Mood, and Control, supporting the role of early behavioral disturbances as clinical indicators of amyloid-related pathology12,44,60,61. Moreover, higher DASS-Stress scores were strongly associated with elevated Aβ40 levels, reinforcing evidence of an association between stress and amyloid pathology62,63. The inconsistent findings for Aβ42 and the Aβ42/Aβ40 ratio in our study and throughout literature highlight variability likely attributable to differences in study cohorts and methodological heterogeneity, underscoring the complexity of utilizing amyloid biomarkers for prediction of CD.
Furthermore, elevated MBI-C Total scores, DASS-Depression and DASS-Stress scores were significantly linked to higher p-tau181 levels, aligning with emerging evidence that behavioral symptoms may correlate with tau pathology12–14,64,65. This relationship illustrates the potential interplay between stress, depression, and tau-related neurodegeneration, which is critical for understanding the broader neuropathological mechanisms underlying CD. Finally, the marginal association between elevated GFAP levels and DASS-Anxiety scores suggests a tentative link between anxiety and neuroinflammatory processes, potentially reflecting glial activation in response to chronic psychological distress.
Plasma Biomarkers and MCI status
Our results demonstrate the potential of BBM to distinguish MCI from CN participants, reinforcing their value in early detection of AD and related cognitive disorders.
Amyloid Pathology and Early Detection
Although the Aβ42/40 ratio did not significantly differ between the CN and MCI groups, the elevated Aβ40 levels in MCI participants suggest early amyloid dysregulation. Aβ40, which is increasingly implicated in amyloid plaque formation and neurotoxicity, may serve as an early biomarker of amyloid pathology despite historically receiving less attention than Aβ4266. This pattern aligns with recent research indicating that Aβ40 elevations may precede Aβ42 changes67, supporting its potential as an early biomarker of amyloid-related pathology.
Tau
In contrast, p-tau181 did not differ significantly between cognitive groups nor predict MCI status. Although tau pathology is central to AD, plasma p-tau181 may have lower sensitivity to early cognitive impairment compared to markers of amyloid dysregulation or neuroinflammation68. This corresponds with recent findings that p-tau181 becomes more relevant in later AD stages as tau pathology accumulates in the medial temporal cortex and spreads to other cortical regions69,70.
Neurodegeneration
NfL emerged as a robust marker of neurodegeneration, strongly associated with MCI status. Its high sensitivity to axonal damage makes it valuable for detecting preclinical neurodegeneration and tracking the MCI-to-AD transition71–73. This study supports NfL’s role as a core biomarker for monitoring disease progression and evaluating neuroprotective therapies74.
Neuroinflammation
The significant elevation of GFAP levels in MCI participants reinforces its role as a biomarker of neuroinflammation, specifically astrocytic activation75. Our results suggest that neuroinflammation, a key driver of AD pathology, may precede or exacerbate AD-related changes, contributing to CD via oxidative stress, cytokine release, and neuronal damage76,77. Detecting neuroinflammation through GFAP may enable early intervention strategies aimed at modulating these pathways before significant neuronal loss occurs.
Enhanced Predictive Power through Integrated Models
Our results show that integrating BBMs with behavioral assessments provides a comprehensive view of early pathological changes in CD, offering insights beyond traditional cognition-based models. The combined model demonstrated notably superior discriminative power (MCI vs. CN) compared to models using behavioral or BBM alone, emphasizing the value of capturing both pathobiological changes and NPS. Notably, GFAP and MBI-C Mood emerged as the most robust predictors of MCI, supporting addressing both neuroinflammatory processes and mood-related symptoms in early detection strategies.
The Role of Neuroinflammation and Behavioral Changes
The prominent role of GFAP in our findings emphasizes neuroinflammation as a key process in early pathogenesis of AD and other dementias. Elevated GFAP levels, reflecting astrocytic activation, are increasingly recognized not only as markers of neurodegeneration but as contributors to disease progression via cytokine release and reactive gliosis, actively drive neuroinflammatory responses that exacerbate neuronal injury and accelerate CD78,79. Consequently, GFAP emerges as a critical biomarker for early detection, particularly in individuals with subtle cognitive changes at high risk for progression.
The significant predictive value of MBI-C-Mood scores emphasizes the role of mood disturbances as early indicators of neurodegeneration. Rather than psychological consequences of cognitive impairment, mood disturbances may reflect underlying neuropathological changes80–82. The association between elevated mood scores, p-tau181, and GFAP links emotional dysregulation to biological markers of neurodegeneration, particularly in the prefrontal cortex and hippocampus— regions vulnerable to early AD pathology83. These findings affirm the interplay between mood disturbances and the biological processes driving CD.
Limitations
As with most studies, there are limitations. The exclusion of participants with severe psychiatric diseases may limit the generalizability of our findings, as this group could provide crucial insights into early cognitive changes. Importantly, severity of psychiatric symptoms is often used as a proxy for psychiatric disorders, potentially overlooking neurodegeneration underlying some NPS. The cross-sectional design further restricts our ability to infer causality or track changes over time. The relatively small sample size also limits statistical power and generalizability to larger populations. Future longitudinal studies, such as those planned within the expanded BIOCIS cohort, will be essential to validate these associations, elucidate temporal relationships between NPS, BBM and CD. Finally, translating these findings into clinical practice will require research focused on validating biomarker mechanisms, assessing predictive value for disease progression, and ensuring the scalability, cost-effectiveness, and accessibility of biomarker assays for real-world implementation.
Conclusion
This study accentuates the value of integrating biological markers with behavioural-biomarkers to deepen our understanding of early neurodegeneration and CD. Elevated GFAP and MBI-C Mood scores emerged as robust indicators of cognitive impairment, spotlighting the vital roles of neuroinflammation and mood disturbances. This integrative approach holds significant potential to enhance early detection of cognitive impairment, refine intervention strategies for at-risk populations, and improve outcomes through timely intervention.
The combination of self-reported MBI assessments with BBM provides enhanced diagnostic utility, enabling a more nuanced understanding of NDD progression. As precision medicine gains prominence, such integrative models can inform clinical decision-making and enable targeted disease-modifying therapies for individuals with neurodegenerative disorders. By establishing a framework that incorporates NPS, MBI, and BBM into diagnostic protocols, this research supports targeted therapeutic interventions and has the potential to transform dementia diagnosis, monitoring, and care.
Question
Do behavioral and blood-based biomarkers independently predict mild cognitive impairment (MCI) compared to cognitively normal (CN) individuals, and does their integration enhance predictive accuracy?
Findings
In this cohort study of 678 participants, higher plasma GFAP and NfL levels, as well as elevated MBI-C-Mood and DASS-Depression scores, were significantly associated with MCI over CN status. A combined model integrating blood-based and behavioral markers demonstrated superior discriminatory power (AUC: 0.786) compared to models using behavioral (AUC: 0.593) or blood-based (AUC: 0.697) biomarkers alone.
Meaning
The integration of behavioral and blood-based biomarkers enhances the early detection of cognitive impairment, highlighting the roles of neuroinflammation and mood disturbances as key contributors to neurodegenerative processes.
Article Information
Corresponding Author
Nagaendran Kandiah, Lee Kong Chian School of Medicine, Nanyang Technological University, 11 Mandalay Rd, Singapore 308232 (nagaendran_kandiah@ntu.edu.sg)
Author Contributions
Yi Jin Leow had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Yi Jin Leow
Acquisition, analysis, or interpretation of data: Yi Jin Leow, Zahinoor Ismail, Seyed Ehsan Saffari, Gurveen Kaur Sandhu, Pricilia Tanoto, Faith Phemie Hui En Lee, Smriti Ghildiyal, Shan Yao Liew, Gursimar Bhalla, Sim Xin Ying, Adnan Azam Mohammed, Ashwati Vipin, Bocheng Qiu, Chao Dang, Nagaendran Kandiah
Drafting of the manuscript: Yi Jin Leow, Zahinoor Ismail, Seyed Ehsan Saffari, Nagaendran Kandiah
Statistical analysis: Yi Jin Leow, Seyed Ehsan Saffari
Supervision: Nagaendran Kandiah
Funding
This study received funding support from the Strategic Academic Initiative grant from the Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, National Medical Research Council, Singapore under its Clinician Scientist Award (MOH-CSAINV18nov-0007), Ministry of Education Start-up Grant, Ministry of Education Academic Research Fund Tier 1 (RT02/21) and Ministry of Education Science of Learning Grant (MOESOL2022-0002).
Conflict of Interest Disclosures
All authors declare no competing interests.
Data Availability
Data may be made available upon reasonable request to the corresponding author.
Availability of data and materials
Data may be made available upon reasonable request to the corresponding author.
Ethics approval and consent to participate
Informed consent was obtained from all participants according to the Declaration of Helsinki and local clinical research regulations, and procedures used in the study were in accordance with ethical guidelines. This study has been approved by the NTU Institutional Review Board (NTU-IRB-2021-1036), and procedures used in the study were in accordance with ethical guidelines.
Consent for publication
All authors have reviewed and approved the contents of the final manuscript contents and provided consent for its publication.
Acknowledgements
The authors express their gratitude to all individuals who are currently or will be participating in research at the Dementia Research Centre (Singapore).
References
- (1).↵
- (2).↵
- (3).
- (4).↵
- (5).↵
- (6).↵
- (7).
- (8).↵
- (9).↵
- (10).↵
- (11).↵
- (12).↵
- (13).↵
- (14).↵
- (15).↵
- (16).↵
- (17).↵
- (18).
- (19).
- (20).
- (21).↵
- (22).↵
- (23).↵
- (24).↵
- (25).↵
- (26).↵
- (27).↵
- (28).↵
- (29).
- (30).↵
- (31).↵
- (32).
- (33).↵
- (34).↵
- (35).↵
- (36).↵
- (37).↵
- (38).↵
- (39).↵
- (40).↵
- (41).↵
- (42).↵
- (43).↵
- (44).↵
- (45).↵
- (46).↵
- (47).↵
- (48).↵
- (49).↵
- (50).↵
- (51).↵
- (52).↵
- (53).↵
- (54).↵
- (55).↵
- (56).↵
- (57).↵
- (58).
- (59).↵
- (60).↵
- (61).↵
- (62).↵
- (63).↵
- (64).↵
- (65).↵
- (66).↵
- (67).↵
- (68).↵
- (69).↵
- (70).↵
- (71).↵
- (72).
- (73).↵
- (74).↵
- (75).↵
- (76).↵
- (77).↵
- (78).↵
- (79).↵
- (80).↵
- (81).
- (82).↵
- (83).↵
