Plasma‐based neurobiological protein biomarkers as predictors of dementia progression: Insights from longitudinal aging study in India – Diagnostic assessment of dementia

Abstract

BACKGROUND

Major neurocognitive disorder (MNCD) represents a critical public health crisis. From the two waves of the Longitudinal Aging Study in India‐Diagnostic Assessment of Dementia, we investigated the predictive power of longitudinal alterations in plasma proteins to predict the progression from mild cognitive impairment (MCI) to MNCD in India.

METHODS

This prospective cohort followed 4635 participants over 55 months. Plasma levels of amyloid beta (Aβ) 42/Aβ40 ratio, neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP), tau phosphorylated at threonine 181 (pTau181), and total tau (t‐tau) were measured via single molecule array technology. Comprehensive geriatric assessments were conducted, and Clinical Dementia Rating (CDR) scores were measured.

RESULTS

As participants progressed from MCI to MNCD, significant declines in cognitive and functional parameters coincided with elevated plasma protein levels. Biomarkers (GFAP, NfL t‐tau, and pTau181) demonstrated strong predictive accuracy, with area under the curve values of 81%, 71%, 74%, and 67%, respectively.

DISCUSSION

Plasma biomarkers effectively estimate dementia progression rates. These findings support using blood‐based markers to monitor and track disease trajectory in the Indian population, offering significant clinical utility for early intervention.

Highlights

• In the progressor group, median HMSE and IADL scores significantly declined.

• Cognitive decline correlates with increased plasma NfL, GFAP, and tau proteins.

• Level of protein markers changed overtime in dementia progressor groups.

• Plasma GFAP, NfL, t‐tau, and pTau181 can predict the progression of MCI to MNCD.

• GFAP showed the greatest predictive accuracy for transitioning from MCI to MNCD.

• These biomarkers may serve as indicators for monitoring MCI across diverse Indian populations.

1. BACKGROUND

In the present global health landscape, the prevalence and incidence of dementia are accelerating dramatically in developing countries as aging populations increase exponentially. The estimated rate of dementia in older people aged ≥60 years in India is 7.4% (8.8 million individuals). ¹ The huge burden of dementia is expected to rise over time, underscoring the need for early diagnosis, monitoring, and effective prevention strategies.

Research has highlighted that the biochemical mechanisms underlying dementia, which reflect Alzheimer's disease (AD) neuropathology, begin to manifest two decades before the onset of clinical symptoms of dementia. ² For the early diagnosis, blood‐based biomarkers have emerged as promising surrogate markers and predictive tools for dementia that are largely accessible in community‐based population studies. ³ , ⁴ Blood‐based protein markers are less expensive, high‐throughput, minimally invasive, and advantageous compared to cerebrospinal fluid (CSF). It can overcome the radiation risk of positron emission tomography (PET) in the clinical management of major neurocognitive disorder (MNCD).

Clinical studies have consistently demonstrated strong associations between blood‐based protein biomarkers and dementia, with accuracy in predicting cognitive decline and the onset of dementia. However, evidence regarding their performance in population‐based settings, particularly in longitudinal studies, remains limited. Longitudinal studies across the dementia continuum are crucial to provide information on the disease progression of mild cognitive impairment (MCI) and MNCD on the basis of changes in protein biomarkers over time. ⁵ Waves 1 and 2 of the Longitudinal Aging Study in India‐Diagnostic Assessment of Dementia (LASI‐DAD) collected longitudinal socio‐clinical, cognitive and neurodegenerative protein markers data, based on population‐representative data, which aid in the study of risk factors of dementia. Previous studies on dementia in India showed non‐representative samples in geographically limited regions. In contrast, LASI‐DAD enrolled a nationally representative, diverse cohort covering the majority of states and union territories.

Studies in the literature mentioned that plasma phosphorylated tau at threonine 181 (pTau181) and amyloid beta (Aβ) were more specific to the brain pathology in dementia compared to neurofilament light (NfL), glial fibrillary acidic protein (GFAP), and total tau (t‐tau). ⁶ Circulating levels of NfL, Aβ42/40 ratio, and pTau181 biomarkers have been reported to reflect AD‐related neuropathological processes such as impaired clearance of Aβ from the brain, disruption of the axonal cytoskeletal structure, and reactive astrogliosis. ⁷ , ⁸ , ⁹ The accumulation of Aβ42 into plaques is signaled by a decrease in soluble Aβ42 in CSF, which is the first sign of a pathological process. ¹⁰ Plasma GFAP, an intermediate filament protein of astrocytes, has been demonstrated to be associated with brain amyloid status and the development of AD and cognitive decline. ¹¹ , ¹² Plasma NfL, a cytoplasmic protein, is a marker of neurodegeneration associated with cognitive decline, brain atrophy, and hypometabolism. ¹³ , ¹⁴ NfL has been mentioned as being among the most crucial AD‐associated biomarkers in the Alz‐biomarker Database, and at the prodromal stage NfL has been found at increased levels. ¹⁵ However, many studies demonstrated that high blood NfL levels were associated with neurodegenerative diseases, and pTau181 can effectively differentiate AD from other neurodegenerative disorders, making it a valuable biomarker for the screening and diagnosis of AD. ¹⁶ Thus, these studies highlight that blood‐based Aβ42, Aβ40, NfL, GFAP, and pTau181 can be considered a core set of biomarkers in dementia research since together they capture different but complementary aspects of neurodegeneration, making them highly valuable for diagnosis, prognosis, and disease monitoring worldwide. Aβ42/Aβ40 ratio has been considered a core AD‐specific pathology marker, whereas NfL has been found to be a strong indicator of ongoing neurodegeneration. GFAP adds an inflammatory dimension to dementia biomarker panels, and pTau181 has been shown to be one of the most clinically transformative blood biomarkers in recent years. These four biomarkers collectively capture the core biological hallmarks of dementia, particularly AD, and may enable scalable and minimally invasive testing of dementia globally.

This study aimed to evaluate longitudinal changes in neurodegenerative plasma protein marker levels of the Aβ42/Aβ40 ratio, NfL, GFAP, pTau181, and t‐tau between the study groups over a period of 55 months of follow‐up in the population‐based cohorts of older Indians as biomarkers of MCI and MNCD progression.

RESEARCH IN CONTEXT

1. Systematic review: MNCD is a critical public health crisis. From LASI‐DAD, we investigated for the first time in India the predictive power of longitudinal changes in plasma proteins to monitor progression from MCI to MNCD and control to MCI over 55 months.

2. Interpretation: This study reports that, as participants progressed from control to MCI and MCI to MNCD, significant declines in cognitive and functional parameters coincided with elevated plasma protein levels. Plasma biomarkers effectively estimate dementia progression rates. The predictive power of the biomarkers was validated by high AUC values.

3. Future directions: These findings support using plasma protein markers (Aβ42/Aβ40 ratio, NfL, GFAP, pTau181, and t‐tau) to monitor and track the dementia trajectory in ethnically diverse Indian population, offering significant clinical utility for early intervention.

2. METHODS

2.1. Study population

This prospective longitudinal cohort study recruited the respondents from Wave 1 and Wave 2 of the harmonized LASI‐DAD on the basis of the Clinical Dementia Rating (CDR) and specific protein markers. During Wave 1 of LASI DAD, 4096 Indian participants aged 60 years and older were recruited from the LASI study (a prospective, population‐based survey) between 2017 and 2020. ¹⁷ The Wave 2 of LASI‐DAD included 4635 participants; among them 2565 participants were followed up from Wave 1, which was conducted across the country between 2022 and 2024. ¹⁸ As for the remaining Wave 1 cohort, 982 respondents were deceased, 324 refused to participate, and 225 could not be located. Participants with missing data were excluded from the final analysis.

The Wave 1 respondents were followed up at Wave 2 with repeated cognitive assessment and blood protein marker assays. The follow‐up period between Waves 1 and 2 was 55 months.

2.1.1. Study execution

For proper coordination and execution of the study, multiple stakeholders were involved. The collaborating centers, All India Institute of Medical Sciences (AIIMS), New Delhi, and University of Southern California (USC), USA, coordinated closely with the regional collaborating hospitals, which played a leading role in data collection.

2.1.2. Ethical approval

The LASI‐DAD Waves 1 and 2 study was approved by the Ethics Committee of the AIIMS, New Delhi, India (IEC‐274/01.04.2022, RP‐05/2202). All investigations were conducted in accordance with the principles outlined in the Declaration of Helsinki in 1975 and its amended guidelines in 2013.

2.2. Cognitive evaluation

Cognitive status was assessed by CDR and Hindi Mini‐Mental State Examination (HMSE). Cognitive scores were derived to reflect memory, executive function, language, and visuospatial domains. All neuropsychological tests were performed at both baseline and follow‐up.

2.2.1. CDR measurements

To rate the participants' cognition on the basis of the CDR scale, the online platform was used to provide standardized data needed by clinicians. On each CDR domain, multiple clinicians independently scored each respondent using standardized data with expert clinical judgment. Algorithmically, the overall CDR score was calculated. For any discrepancy in the ratings, a virtual consensus conference was held for the final rating. CDR is categorized as 0 = cognitively normal, 0.5 = MCI, and 1 or greater = MNCD. ¹⁹

2.2.2. HMSE measurement

A validated Hindi version of the Mini‐Mental State Examination (MMSE) summary score was used for screening the cognitive impairment and dementia in a larger cohort and was useful particularly in a low‐literacy population in rural India. HMSE scores range from 0 to 30 and comprise different cognitive domains. ²⁰ A lower score indicates a higher severity of dementia, with a cut‐off of ≤25 as cognitively impaired.

2.3. Geriatric assessment

2.3.1. Weight, body mass index (BMI), and nutritional status

The weight (kg) and BMI of the study groups were measured in both waves to monitor changes in dementia status. Evaluation of nutritional status was done using the Mini Nutritional Assessment (MNA) scale, a tool consisting of 18 items that makes it possible to group respondents into three categories: normal, at risk of malnutrition, or malnourished. ²¹

2.3.2. Functionality assessment

Functionality was assessed by the Activities of Daily Living (ADL) and Lawton Instrumental ADL (IADL). ²² Impairment of everyday activities is an essential element for the diagnosis of dementia. It impacts the quality of life for both the patient and the caregiver with an increase in socioeconomic burden. The ADL score measures the person's current level of ability to perform basic daily tasks. The total possible score ranges from 0 to 6. A score of less than 6 indicates impairment. The IADL score measures the level of cognition and complex activities. The total possible scores range from 0 to 8, where a score of <8 for women and <5 for men denotes functional impairment, with a lower score indicating greater dependency and disability.

2.3.3. Mental health assessment

2.3.3.1. Center for Epidemiologic Studies Depression (CESD) scale

The CESD scale was used to measure depression. It consisted of 20 self‐report items measuring current depression symptom severity, with a total score range of 0 to 60. A score of ≥16 indicates a person at risk of clinical depression.

2.3.3.2. Beck Anxiety Inventory (BAI) scale

The BAI scale was used to measure the severity of anxiety in dementia patients. It consists of a 21‐item self‐report questionnaire. The cut‐off score of ≥16 is most frequently associated with generalized anxiety disorder. The higher the BAI score, the more anxious the person felt in the past week.

2.4. Quantitative analysis of plasma‐based assays

2.4.1. Blood collection and plasma separation

The details of the procedure for sample collection were reported in a previous paper. ²³ Following the ethical guidelines and protocol of venous blood specimen (VBS) collection, a blood sample was collected from each participant by a trained phlebotomist into K2 EDTA plasma preparation tubes (Catalog No. 362788, BD Biosciences, NJ, USA) for the biomarker assays. The collected blood samples were then shipped to the central Metropolis Laboratory and at 3000 rpm for 10 min in a refrigerated centrifuge. The resulting supernatant (plasma) was divided into multiple aliquots, and then the plasma samples were assigned a unique cryogenic barcode. These samples were shipped to the Department of Biophysics at the AIIMS, New Delhi, for neurodegenerative protein marker assays. The plasma samples were stored at −80°C, until testing.

2.4.2. Biomarker analysis using single molecule array (Simoa) technology

Assays were performed for six protein molecules (NfL, Aβ42, Aβ40, GFAP, pTau181, and t‐tau). Quanterix 96‐well plates were used for sample plating. All measurements were performed on an HD‐X instrument from Quanterix (Quanterix Corp., Billerica, MA, USA). Human Neurology 4‐Plex E (N4PE, Ref: 103670, Simoa, MA, USA), for measuring Aβ42, Aβ40, NfL, and GFAP concentrations, pTau181 V2.1 reagent kit (pTau181, Ref: 104111, Simoa, MA, USA), and Tau 2.0 reagent kit (Tau, Ref: 101552, Simoa, MA, USA) were used according to the manufacturer's instructions. Samples were automatically diluted at 1:4 (four‐fold) with an assay‐specific sample diluent using on‐board dilution on the Quanterix Simoa HD‐X analyzer. The fluorescent signals produced by a single bead were recorded automatically and converted into average enzymes per bead (AEB). In the presence of the proper calibrators, AEB values were further interpolated into SI concentration units. Eight‐point calibration curves and sample measurements were determined on the Simoa HD‐X analyzer software.

2.5. Statistical analyses

Normality of the data was tested using the Shapiro–Wilk test. Qualitative variables were expressed as frequencies and percentages. Quantitative variables were presented as medians and interquartile ranges (first and third quartiles), and comparisons between two groups were made using the Wilcoxon rank‐sum (Mann–Whitney) test. Differences between the groups for categorical variables were analyzed using the χ2 test. Receiver operating characteristic (ROC) analyses were performed to determine the area under the curve (AUC), best cut‐off, sensitivity, and specificity of protein markers between control (Wave 1) to MCI (Wave 2), control (Wave 1) to MNCD (Wave 2), and MCI (Wave 1) to MNCD (Wave 2) conversion groups. Correlations among the plasma protein marker concentrations was estimated via Spearman rank correlation coefficients. The association between protein markers and disease progression groups was analyzed by logistic regression analysis. Two models were constructed: Model 1 was unadjusted values; Model 2 was adjusted for age, sex, BMI, hypertension, diabetes, heart disease, stroke, and head injury. The results were presented as odds ratios (ORs) with 95% confidence intervals (CIs). The significance level was set at p < 0.05. All the analyses were performed using Stata 14 (Stata Corp LP, Texas, USA) and graphs were generated using GraphPad Prism 8 (GraphPad Software, version 8.0, San Diego, USA).

3. RESULTS

3.1. Characteristics of LASI‐DAD study cohorts

Baseline characteristics, demographics, and socio‐clinical data of the study participants were reported. All study participants were aged between 60 and 100 years. Median baseline age was 70 years (IQR 67 to 75) in controls, 71 years (IQR 67 to 76) in MCI, and 73 years (IQR 70 to 78) in MNCD group. The higher proportion (61.46% in MNCD, 60.81% in MCI, and 55.71% in controls) of participants were among the 60‐ to 69‐year‐old category. The percentages of female participants were relatively higher (52.73%). There were differences in the frequency of education and place of residence. Higher percentages of participants were illiterate in the MNCD group followed by MCI. Most of the participants were from a rural setting (67.41%). In the MNCD group, the most common comorbidity was hypertension (43.33%), followed by diabetes (23.33%), stroke (10%), and heart disease (6.67%).

The baseline (Wave 1) and follow‐up (Wave 2) characteristics of the study participants are presented in Table 1. In this study, we classified participants into three progressor groups – control to MCI, control to MNCD, and MCI to MNCD – during the follow‐up period of 55 months. The median HMSE and IADL scores significantly decreased in all three progressor groups. CESD and BAI scores significantly increased in the control study group of Wave 1, which progressed to MCI in the follow‐up period. Overall, the risk of depression and anxiety increased concomitantly with disease progression. Both weight and MNA scores significantly decreased among participants who progressed to the MCI and MNCD stage in Wave 2. While BMI was lower in the follow‐up groups compared to baseline, this decrease did not reach statistical significance.

TABLE 1.

Demographic and clinical attributes of study participants at baseline (Wave 1) and follow‐up (Wave 2).

Attributes	Control (W1) (median [IQR])	MCI (W2) (median [IQR])	p value	Control (W1) (median [IQR])	MNCD (W2) (median [IQR])	p value	MCI (W1) (median [IQR])	MNCD (W2) (median [IQR])	p value
Age	66 (63 to 69)	72 (68 to 78)	<0.0001	67 (63 to 72)	71 (68 to 77)	0.1651	69 (65 to 74)	73 (70 to 79)	0.0001
ADL	6 (5 to 6)	6 (4 to 6)	0.1523	6 (5 to 6)	6 (4 to 6)	0.8356	5 (4 to 6)	5 (3 to 6)	0.7523
IADL	7 (6 to 7)	5 (4 to 6)	<0.0001	6 (5 to 7)	3 (1 to 5)	0.0170	3 (2 to 6)	2 (1 to 4)	0.0026
HMSE	28 (25 to 29)	23 (19 to 26)	<0.0001	23 (21 to 28)	18 (16 to 20)	0.0211	18.5 (16 to 23)	15 (12 to 19)	0.0007
CESD	7 (3 to 10)	10.5 (7 to 14)	<0.0001	9 (8 to 11)	11 (6 to 14)	0.7491	12 (8 to 16)	13 (10 to 16)	0.1837
BAI	1 (0 to 3)	2 (0 to 6)	0.0004	2 (1 to 3)	2 (0 to 4)	0.7010	2 (0 to 6)	3 (0 to 6)	0.7884
Weight	60.75 (52 to 69.6)	55.8 (49.85 to 63.3)	0.0333	51.1 (46.9 to 58.6)	42.35 (37.8 to 49.9)	0.1121	46.5 (39 to 55.3)	47.6 (38.6 to 52.4)	0.7092
BMI	23.96 (21.40 to 27.38)	22.97 (20.02 to 26.20)	0.1034	22.91 (18.30 to 27.40))	20.47 (16.27 to 22.69)	0.4008	19.84 (18.43 to 24.14)	20.09 (17.17 to 23.17)	0.5585
MNA	22 (20 to 24)	20.5 (18 to 22.5)	0.0024	22.5 (21.5 to 24)	16.5 (13.5 to 17.5)	0.0014	18.5 (17 to 20)	15 (12 to 19)	0.5733

Note: p < 0.05 is considered significant.

Abbreviations: ADL, activities of daily living; BAI, Beck Anxiety Inventory; BMI, body mass index; CESD, Center for Epidemiologic Studies Depression Scale; HMSE, Hindi Mini‐Mental State Examination; IADL, instrumental activities of daily living; IQR, interquartile range; MNA, Mini Nutritional Assessment; W1, Wave 1; W2, Wave 2.

Spearman's correlation analysis between the plasma proteins markers is represented in Figure S1. In accordance with the baseline data, among the protein markers, the strongest correlations were observed between NfL and GFAP levels (Spearman's ρ = 0.45), followed by pTau181 and t‐tau levels (Spearman's ρ = 0.32).

3.2. Longitudinal changes in protein markers

Among all the progressor groups from Wave 1 to Wave 2, the differences in the levels of proteins between progressors and non‐progressors were measured for their accuracy in predicting MCI and MNCD as progression markers. Participants progressing from normal cognition to MCI (Wave 2) and from MCI to MNCD (Wave 2) showed significantly elevated plasma NfL, GFAP, pTau181, and t‐tau levels compared to their Wave 1 baselines. The Aβ42/40 ratio remained stable and did not reach statistical significance.

Longitudinal changes in protein markers over time in participants in three different progressor groups from Wave 1 to Wave 2 are represented in Figure 1 and Table 2.

FIGURE 1.

Violin plot showing level of protein; NfL, Aβ42/40 ratio, GFAP, pTau181, and t‐tau in study participants progressing from control to MCI (A), control to MNCD (B), and MCI to MNCD (C) over a period of 55 months. Aβ 42/40 ratio, amyloid beta 42/40 ratio; GFAP, glial fibrillary acidic protein; MCI, mild cognitive impairment; MNCD, major neurocognitive disorder; NfL, neurofilament light chain; pTau181, tau phosphorylated at threonine 181; t‐tau, total tau. W1, Wave 1; W2, Wave 2; p < 0.05 was considered statistically significant.

TABLE 2.

Protein levels of NfL, Aβ 42/40 ratio, GFAP, pTau181, and t‐tau in the study participants progressing from control (W1) to MCI (W2), control (W1) to MNCD (W2), and MCI (W1) to MNCD (W2), represented as median (IQR).

Protein marker (pg/mL)	Control (W1) (median [IQR])	MCI (W2) (median [IQR])	p value	Control (W1) (median [IQR])	MNCD (W2) (median [IQR])	p value	MCI (W1) (median [IQR])	MNCD (W2)(median [IQR])	p value
NfL	23.05 (16.53 to 34.79)	30.51 (21.05 to 43.68)	0.0014	30.24 (7.33 to 44.69)	40.84 (30.71 to 55.48)	0.3095	26.06 (18.60 to 35.47)	50.92 (24.05 to 90.25)	0.0056
Aβ 42/40 ratio	0.061 (0.049 to 0.075)	0.064 (0.052 to 0.071)	0.9016	0.061 (0.048 to 0.096)	0.074 (0.061 to 0.083)	0.6905	0.059 (0.049 to 0.074)	0.065 (0.054 to 0.078)	0.4435
GFAP	107.2 (70.06 to 154.3)	117.3 (86.82 to 185.7)	0.0427	94.07 (32.60 to 133.7)	139.5 (112.3 to 218.2)	0.1508	116.1 (78.67 to 172.8)	243.7 (163 to 296.8)	<0.0001
pTau181	34.95 (24.15 to 49.19)	43.63 (33.58 to 65.63)	<0.0001	34.35 (33.62 to 42.27)	50.56 (40.26 to 78.92)	0.0317	36.68 (27.13 to 36.68)	51.04 (38.93 to 57.77)	0.0177
T‐tau	1.67 (0.91 to 2.70)	2.54 (1.45 to 3.74)	0.0005	0.52(0.509 to 2.604)	1.33 (0.85 to 3.44)	0.3095	1.77 (0.95 to 2.94)	3.24 (2.23 to 6.47)	0.0005

Note: p < 0.05 is considered significant.

Abbreviations: Aβ 42/40 ratio, amyloid beta 42/40 ratio; GFAP, glial fibrillary acidic protein; IQR, interquartile range; MCI, mild cognitive impairment; MNCD, major neurocognitive disorder, NfL, neurofilament light chain; pTau181, tau phosphorylated at threonine 181; t‐tau, total tau; W1, Wave 1; W2, Wave 2.

3.3. ROC analyses

ROC curves were generated for each protein marker, based on the Simoa data. The AUC was constructed to measure the utility of these protein molecules as potential biomarkers for disease progression of MCI and MNCD. In our study, higher levels of the plasma proteins were associated with disease progression (Figure 1). The threshold for detecting MNCD was selected based on the distribution of specificities and sensitivities. GFAP showed the greatest predictive accuracy for transitioning from MCI to MNCD, followed by t‐tau, NfL, and p‐tau181. However, in the ROC analysis on the progression of the control (Wave 1) to MCI (Wave 2) group, the AUC values of all the plasma proteins were below 0.650 with lower sensitivities and specificities (Figure 2).

FIGURE 2.

ROC analysis of plasma proteins NfL, GFAP, pTau181, and t‐tau to distinguish (A) control versus MCI between two waves (Waves 1 and 2) and (B) MCI versus MNCD between two waves (Waves 1 and 2): determining area under the curve (AUC), cut‐off, sensitivity, and specificity. GFAP, glial fibrillary acidic protein; MCI, mild cognitive impairment; NfL, neurofilament light chain; pTau181, tau phosphorylated at threonine 181; ROC, receiver operating characteristic; t‐tau, total tau.

3.4. Protein marker correlation with age, sex, BMI, and comorbidities with disease progression

Participants who progressed from MCI to MNCD showed significantly higher levels of protein markers, including GFAP, pTau181, and t‐tau, even after adjusting for potential confounding factors such as age, sex, BMI, hypertension, diabetes, heart disease, stroke, and head injury by logistic regression analysis.

However, in the control group progressed to MCI, the ORs of NfL, GFAP, pTau181, and t‐tau were not significant after adjusting for the aforementioned confounding factors (Figure 3 and Table 3).

FIGURE 3.

Forest plot showing results of logistic regression association between plasma protein markers (NfL, GFAP, pTau181, and t‐tau) with the progression of MCI to MNCD from Wave 1 to Wave 2. Model 1: pink, unadjusted; Model 2: blue, adjusted with age, sex, BMI, hypertension, diabetes, heart disease, stroke, and head injury. BMI, body mass index; GFAP, glial fibrillary acidic protein; MCI, mild cognitive impairment; MNCD, major neurocognitive disorder; NfL, neurofilament light chain; OR, odds ratio; pTau181, tau phosphorylated at threonine 181; t‐tau, total tau.

TABLE 3.

Protein markers’ correlation with age, sex, BMI, and comorbidities with disease progression (results of logistic regression analysis).

	Model 1		Model 2
Biomarker	OR (95% CI)	p value	OR (95% CI)	p value
A. Progression from control to MCI
NfL	1.41 (0.81 to 2.46)	0.228	0.76 (0.37 to 1.57)	0.472
GFAP	0.84 (0.48 to 1.46)	0.546	0.49 (0.23 to 1.03)	0.062
pTau 181	1.39 (0.82 to 2.39)	0.218	1.59 (0.80 to 3.14)	0.182
T‐tau	1.19 (0.68 to 2.08)	0.526	1.46 (0.71 to 3.00)	0.297
B. Progression from MCI to MNCD:
NfL	6.44 (1.979 to 20.983)	0.002	3.70 (0.84 to 16.26)	0.083
GFAP	10.259 (2.807 to 37.49)	<0.0001	26.52 (3.74 to 187.97)	0.001
pTau 181	4.210 (1.434 to 12.358)	0.009	4.99 (1.22 to 20.39)	0.025
T‐tau	4.909 (1.555 to 15.491)	0.007	5.76 (1.33 to 24.82)	0.019

Note: p < 0.05 is considered as significant.

Abbreviations: CI, confidence interval; GFAP, glial fibrillary acidic protein; IQR, interquartile range; MCI, mild cognitive impairment; MNCD, major neurocognitive disorder; NfL, neurofilament light chain; OR, odds ratios; pTau181, tau phosphorylated at threonine 181; t‐tau, total tau.

Model 1: unadjusted.

Model 2: adjusted for age, sex, BMI, hypertension, diabetes, heart disease, stroke, and head injury.

4. DISCUSSION

Large, population‐based studies with more diverse cohorts and longer follow‐up periods are required to evaluate the clinical validity of these protein biomarkers in real‐world contexts. Such studies would elucidate how the biomarkers are reliable in capturing underlying neurodegenerative or neuroinflammatory processes across heterogeneous populations varying with age, race, genetic background, comorbid profile, cognitive baseline, rate of disease progression, and utility in guiding therapeutic management. People with cognitive impairment are generally admitted to hospital at the MCI stage, so it is important to identify those at risk of progressing to MNCD. Identifying the potential biomarkers that can detect the early stage of cognitive impairment or a higher chance of cognitive decline can be helpful in a clinical setting for management and therapy.

The LASI‐DAD study provides nationally representative data on the ethnically diverse population on late‐life cognition and dementia in India. ²⁴ In this prospective longitudinal study, the plasma protein levels of NfL, Aβ42/Aβ40 ratio, GFAP, pTau181, and t‐tau were evaluated at baseline and over a follow‐up period of 55 months.

Our study revealed significant elevations in NfL, GFAP, p‐tau181, and t‐tau levels during the progression from Wave 1 to Wave 2, from control to MCI, and MCI to MNCD over the period of 55 months. Notably, in the control to MNCD progression group, only t‐tau exhibited a significant longitudinal increase. These findings suggest that plasma biomarkers may serve as strong indicators of dementia disease progression. These longitudinal shifts highlight the utility of these protein markers in monitoring cognitive decline.

Findings from ROC analysis suggest that the plasma protein markers GFAP, NfL, t‐tau, and pTau181 can accurately predict the progression of MCI to MNCD. Thus, they can act as the best plasma predictors (progression markers) in longitudinal studies with cut‐off values to track MCI with accelerated progression throughout the dementia continuum. Findings from our study suggest that predicting progression from control to MCI remains challenging in longitudinal studies, limiting early diagnostic accuracy.

Logistic regression analyses of the association between protein markers and MCI participants who progressed to MNCD from Wave 1 to Wave 2 showed GFAP, pTau181, and t‐tau were independently associated with the disease's progression.

Earlier study reported significant longitudinal increase of GFAP in MCI and AD compared to controls, suggesting a sequence in the progression of biomarkers reflecting the underlying pathological process. ²⁵ Study from Alzheimer's Disease Neuroimaging Initiative (ADNI) suggests longitudinal analyses of plasma pTau181 as a potential non‐invasive biomarker to track disease progression in AD. ²⁶ , ²⁷

In a UK Biobank longitudinal cohort study, elevated baseline plasma GFAP and NfL levels were found to be significantly associated with increased brain atrophy. ²⁸

The ADNI database demonstrated that elevated plasma GFAP and p‐tau217 levels were significantly associated with increased hippocampal atrophy, greater white matter hyperintensity burden, higher cerebral Aβ deposition, and accelerated cognitive decline at baseline as well as during longitudinal follow‐up in AD disease progression. ²⁹

To date, all previous studies’ sampling was done from limited, geographically confined areas, and few studies reported longitudinal changes in the aforementioned plasma marker prognostic value for progression to dementia in clinically classified MCI and MNCD. Study on MEMENTO cohort study from French Research Memory Centers in Order to Improve Knowledge on Alzheimer's Disease and Related Disorders observed higher blood and CSF pTau181 and NfL concentrations associated with accelerated time to AD dementia onset, whereas blood Aβ42/40 was less efficient than CSF Aβ42/40. Blood pTau181 alone was the best blood predictor of 5‐year AD/mixed dementia risk. ³⁰

One longitudinal study reported no differences in plasma concentrations of free Aβ42 between the subjects with normal cognition and preclinical AD. ³¹ Numerous studies have highlighted inconsistencies in findings regarding plasma Aβ, suggesting it may not accurately reflect central Aβ turnover. ³² Measuring Aβ concentrations in the blood remains challenging due to several confounding factors. Specifically, Aβ is produced peripherally, meaning systemic levels do not always correlate with central amyloid accumulation in the brain. ³³ Therefore, for early neurodegeneration and progression, blood Aβ might not be a good candidate biomarker. AD is characterized by an accumulation of extracellular Aβ protein in the brain caused by an imbalance between its production and clearance. Crucial to this clearance is transport across the blood–brain barrier (BBB), which is regulated by specific transporter proteins that become altered in dementia. Specifically, the efflux (removal) of Aβ from the brain into the blood is decreased due to the downregulation of transporters like Lipoprotein receptor–related protein‐1 and P‐glycoprotein. Conversely, the influx (entry) of Aβ into the brain is driven by the upregulation of Receptor for advanced glycation endproducts (RAGE). This abnormal, low influx of Aβ1‐42 into the bloodstream means that the accumulated brain proteins cannot be accurately measured through standard circulatory blood tests. ³⁴ , ³⁵

Demographically, ,   in this study, we observed a higher prevalence of dementia in rural compared with urban habitats. Also, lower education was a considerable risk factor for dementia. Previous surveys similarly reported that older adults residing in rural areas exhibited a higher prevalence of MCI and dementia than those in urban areas. ³⁶ The prevalence of dementia is substantially higher among individuals without formal education compared to educated individuals. ¹

In our study, dementia rating scores, specifically the HMSE and CDR, decreased significantly as the disease progressed, while CESD and BAI scores increased. Studies have shown a higher prevalence of depressive symptoms in those with cognitive impairment, with research indicating a significant increase in depression probability (CESD ≥ 4) in individuals with mild cognitive impairment. ³⁷ Higher BAI scores have been observed in individuals with MCI and AD. A higher baseline BAI score has been reported as an independent predictor for future progression to MCI. ³⁸ Functional decline is a fundamental feature of dementia, substantially reducing the quality of life for affected individuals as well as their caregivers. ³⁹ Previous studies reported that the MCI patients generally preserve functional independence, but a rapid decline occurs in complex IADLs like financial management, medication adherence, and driving, which often occurs 1 to 2 years prior to a dementia diagnosis, signaling its gradual progression to the worst form of dementia. ⁴⁰ Accordingly, in the present large cohort study, a notable decline in functionality (measured by IADL scores) was observed in participants who progressed from MCI to MNCD.

Clinical data suggest a strong link between weight loss and the transition from MCI to AD. ⁴¹ Earlier findings ⁴² noted that 20% to 40% of patients suffer from weight loss during the mild to moderate phases of the disease. Clinically, weight loss in older adults is associated with immune system alterations, muscular atrophy, increased fall risk, and functional dependence. In conjunction with multiple comorbidities, these factors collectively increase the risk of hospitalization and mortality. ⁴³ Likewise, we observed a decline in body weight during the follow‐up period compared to the baseline.

Dietary status, as indicated by MNA scores, also decreased significantly among MCI participants who progressed to MNCD, serving as a further indicator of cognitive decline. Indeed, poor nutritional status has been linked to more rapid cognitive deterioration. ⁴⁴ Collectively, these clinical assessment factors provide a comprehensive overview of the progression of cognitive decline in our participants.

Altogether, the present longitudinal follow‐up study of LASI‐DAD highlights that cognitive decline is a multifactorial process influenced by socio‐demographic factors (such as education), psychological health (depression and anxiety), biological and molecular markers, and modifiable lifestyle components including body weight and dietary patterns. These findings underscore the need for an integrated, multidimensional approach to early identification, prevention, and intervention strategies aimed at promoting healthy cognitive aging in the Indian population.

This longitudinal study investigated the concentrations of specific proteins (NfL, GFAP, pTau 181, and t‐tau) involved in neurodegeneration, correlating these molecular changes with clinical cognitive scores. Despite India's vast ethnic and genetic diversity, our multi‐state cohort study across 28 states revealed consistent, longitudinal changes in plasma protein markers associated with disease progression. These findings suggest that these specific biomarkers may serve as a universal indicator for monitoring MCI across diverse Indian populations. The protein markers GFAP, pTau181, and t‐tau were independently associated with the disease's progression from MCI to MNCD.

The major  strength of this study is that, for the first time in India, it investigated the longitudinal changes in protein biomarkers in two different phases in the progression of the disease from Wave 1 to Wave 2 in three different groups over a period of 55 months.

Beyond the many strengths of this study, it has limitations that should be noted. First, the sample sizes in the MNCD group were relatively small. Second, the study did not determine the associations of protein markers with specific dementia subtypes (i.e., AD dementia, frontotemporal dementia, Lewy body dementia, and vascular dementia), and future studies should investigate such associations.

4.1. Future prospects

MNCD subtypes with larger sample sizes and multiple follow‐up periods could allow for better longitudinal comparisons and greater predictive strength. Protein markers may ultimately prove to be reliable tools for monitoring the effects of disease‐modifying therapies.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflicts of interest. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

Consent to participate in the study for cognitive tests, geriatric assessment, venous blood sample draw, and assays was obtained from all study participants (respondents). For respondents with cognitive impairment, consent was obtained from a legally authorized representative, such as a spouse or adult child. If the respondent was unable to read the consent forms, then the interviewer read them to the respondent. Respondents who were unable to sign the consent forms used a thumb impression in place of a signature. Consent forms were collected and interviews were conducted in the respondents’ language. Also, informants gave consent to participate in informant interviews.

Supporting information

Supporting Information

Supporting Information