The Target48 Neurodegeneration panel: A novel tool for profiling protein signatures in neurodegenerative disorders.

The Target48 Neurodegeneration panel: A novel tool for profiling protein signatures in neurodegenerative disorders HOK‐A‐HIN et al.

Abstract

INTRODUCTION

Novel tools for absolute quantification of established and emerging fluid neuro‐biomarkers are required to advance diagnostic studies and improve biological insights.

METHODS

We conducted an extensive analytical and clinical validation of the Olink™ Target 48 Neurodegeneration panel (T48 Neuropanel) in 352 paired cerebrospinal fluid (CSF) and plasma samples from cognitively unimpaired controls (CU), Alzheimer's dementia (AD), frontotemporal dementia (FTD), and dementia with Lewy bodies (DLB), n = 44 per group. Comparisons with benchmark assays were performed.

RESULTS

Good detectability (CSF: 31 out of 42 assays; plasma: 38 out of 42 assays) and technical performance was observed. Benchmark assays showed good correlations, supporting transformation formula's. Next to emerging biomarkers (MMP10, ITGB2), discriminative performance was excellent in AD: CSF pTau217: area under the curve (AUC) = 1; FTD: plasma neurofilament light chain (NfL): AUC = 0.952; and DLB: CSF dopa decarboxylase (DDC): AUC = 0.901.

DISCUSSION

This analytical and clinical validation of the T48 Neuropanel highlights initial cutoffs and emerging biomarkers to aid clinical studies for the diagnosis, prognosis, and monitoring of neurodegenerative diseases.

Highlights

The T48 Neuropanel validates established (i.e., pTau217, Aβ42, NfL, and GFAP) and emerging biomarkers (i.e., DDC, MMP10, ITGB2, ITGAM, NPTX2, NPTXR, SMOC1, sTREM1, and sTREM2) in CSF and plasma.

CSF NfL, GFAP, ITGB2, and ITGAM and plasma GFAP were dysregulated across AD, FTD, and DLB dementias.

The multiplex design of the T48 Neuropanel enables rich biological interpretation by simultaneously quantifying established and emerging biomarkers. The inclusion of absolute quantification facilitates the establishment of cutoffs, supporting its potential for clinical translation.

1INTRODUCTION

Neurodegenerative dementias, including Alzheimer's disease (AD), frontotemporal dementia (FTD), and dementia with Lewy bodies (DLB), affect millions of people globally.1, 2 These conditions are defined by the accumulation of protein aggregates in the brain: AD by amyloid‐β (Aβ) plaques and tau tangles, FTD by tau or TAR DNA‐binding protein 43 (TDP‐43) aggregates, and DLB by α‐synuclein aggregates.3, 4, 5 Quantification of brain aggregate pathology and all aspects of neurodegeneration in its full complexity during life remains a challenge. For AD, advanced cerebrospinal fluid (CSF) and blood‐based biomarkers (e.g., pTau217) are incorporated into the Alzheimer's Association criteria for identification of the core AD pathologies.6 However, for other neurodegenerative dementias, fluid biomarkers for clinical use are not yet available. For FTD, only biofluid measurements of neurofilament light chain (NfL) are incorporated into the clinical guidelines to help distinguish FTD from primary psychiatric disorders.7 In DLB, CSF α‐synuclein seed amplification assay (αSyn‐SAA) has undergone extensive validation and can reliably detect α‐synuclein aggregates and is, therefore, included in the neuronal α‐synuclein disease–integrated staging system research framework.8, 9 In addition, recent studies highlight dopa decarboxylase (DDC) in CSF as a potential biomarker of Lewy body disorders, likely reflecting dopaminergic dysfunction.10, 11, 12, 13, 14 In view of the advancement of disease‐modifying therapies, the need for early, accurate, and pathology‐specific diagnosis is becoming increasingly critical. Furthermore, there is a need for fluid biomarkers that can support clinical trials by capturing diverse biological treatment effects.15 To comprehensively assess similarities and differences between neurodegenerative dementias, the use of protein assay panels that capture diverse underlying brain pathology and biology prove valuable. Recent studies therefore support the added value of multiplex biomarker panels for detecting distinct AD, FTD, and DLB pathogenesis. Such biomarker panels provide added value by capturing the complexity of disease processes in dementias.10, 16, 17, 18, 19, 20 Furthermore, as biomarker profiles are known to shift across disease stages, multiplex panels may have particularly valuable in tracking these dynamic changes along the disease continuum.21, 22

The Target 48 Neurodegeneration Panel (T48 Neuropanel) is a novel multiplex test based on the proximity extension assay (PEA) technology. This biomarker panel simultaneously quantifies 42 proteins from a low sample volume (1 µL). It is designed to cover a broad range of biological mechanisms often dysregulated in neurodegenerative dementias, such as synaptic (dys)function, neuroinflammation, myelin integrity, axon‐glia interactions, neuroprotection, oxidative stress, and vascular functioning. One key advantage of this panel is that it measures proteins in absolute concentrations. This facilitates the development of robust cutoffs and enables external model validation and prospective validation studies, ultimately supporting more accurate biological interpretation and clinical translation. Here, we present the first assessment of this panel in paired plasma and CSF samples from cognitively unimpaired individuals and patients with AD, FTD, and DLB. The aim was to conduct an analytical and clinical validation of the T48 Neuropanel. Analytical performance was assessed by evaluating cross‐reactivity between the assays in the T48 Neuropanel, the detectability in CSF and in the optimized matrix (plasma), and by comparing T48 Neuropanel measurements with results from orthogonal methods. The clinical validation included comparison of the T48 Neuropanel protein levels in AD, FTD, and DLB to the levels in controls, assessment of the discriminative power through receiver operating characteristics curve (ROC) curve analysis, with the Youden cutoffs. Lastly, we assessed which biomarker abnormalities are shared across these neurodegenerative dementias.

2METHODS

2.1Human CSF and plasma samples

Three hundred fifty‐two paired CSF and plasma samples from cognitively unimpaired controls (n = 44) and individuals diagnosed with AD (n = 44), FTD (n = 44), and DLB dementia (n = 44) from the Amsterdam Dementia Cohort (ADC) were included.23, 24 Among these, eight samples were unpaired (six AD, two FTD, and one DLB) because either the CSF or plasma sample was unavailable. Cohort characteristics are shown for the paired samples in Table 1. All participants underwent standard neurological screening and cognitive testing in the ADC. Clinical diagnosis were made in a multidisciplinary consensus meeting based on the diagnostic criteria for each type of dementia.6, 25, 26, 27, 28 CSF was collected and stored according to consensus biobanking guidelines.29 Ethylenediaminetetraacetic acid (EDTA) plasma samples were obtained through venipuncture, centrifuged at 1800 x g for 10 min within 2 hours of collection, and stored at ‐80°C until analysis, following standard operating procedures (SOP) guidelines.30 Concentrations of the core AD CSF biomarkers were used to support the diagnosis of AD and to determine AD copathology in DLB patients, measured by commercially available enzyme‐linked immunosorbent assay (ELISA) kits (Innotest Aβ(1–42) and Innotest hTAUAg assays, Fujirebio, Ghent, Belgium) or by the Elecsys platform (Elecsys β‐Amyloid (1–42) CSF II and Elecsys Phospho‐Tau (181P) CSF assay, Roche Diagnostics, Gmbh, Almere, The Netherlands). A positive AD profile was defined as a tTau/Aβ42 ratio >  0.46 (ELISA) or pTau181/Aβ42 >  0.020 (Elecsys)[31, 32]. CSF Aβ42 concentrations measured by ELISA were corrected for the drift that occurred over the years, as described previously.33 The individuals with FTD included in this study were clinically diagnosed with behavioral variant FTD.34 Individuals with subjective cognitive decline who had normal clinical and cognitive test results and a negative core AD CSF biomarkers (ELISA: tTau/Aβ42 ratio < 0.46; Elecsys: pTau181/Aβ42 ratio < 0.020) were included as controls.31, 32 Informed consent was obtained from all participants or their authorized representatives, in accordance with the ethical approval of the Amsterdam UMC IRB and in agreement with the Helsinki Declaration of 1975.

RESEARCH IN CONTEXT

Systematic review: We found no prior studies in the scientific literature that used the multiplex Target48 Neurodegeneration biomarker panel (T48 Neuropanel) to identify biofluid biomarkers in neurodegenerative dementias such as Alzheimer's disease (AD), frontotemporal dementia (FTD), or dementia with Lewy bodies (DLB). Therefore, we conducted an extensive analytical and clinical validation of this biomarker panel in cerebrospinal fluid (CSF) and plasma samples.

Interpretation: Our study demonstrates the utility of this multiplex biomarker panel for more comprehensively characterizing the underlying pathophysiology of diverse neurodegenerative dementias in both CSF and plasma. Furthermore, the use of absolute quantification enables the development of transformation formulas for benchmarked assays and facilitates the derivation of clinically relevant cutoff values.

Future directions: Our study shows that the T48 Neuropanel multiplex assays could be useful in the investigation of established and emerging fluid biomarkers for neurodegenerative dementias. The lot‐specific Passing–Bablok transformation formulas and clinical cutoff values should be evaluated in independent cohorts to confirm their generalizability, assess cross‐laboratory variability, and establish their robustness across different analytical settings.

2.2Protein analysis

2.2.1Target 48 Neurodegeneration Panel

Forty‐two proteins were simultaneously measured using the multiplex T48 Neuropanel (detailed in Table S1, kit lot: #LC00406, Olink Proteomics, Uppsala, Sweden). Measurements were performed on the Q100 instrument  (Olink Proteomics) in the Neurochemistry Laboratory (Amsterdam UMC, Amsterdam, The Netherlands). All experimental procedures were performed by technicians blinded to the diagnostic information of the samples. CSF and plasma samples were randomized based on clinical diagnosis across multiple plates, separated by matrix. Each plate additionally included three quality controls (QCs), two negative controls, and three single‐point calibrators. Precision (intra‐ and inter‐assay coefficients of variation [CVs]) of the assays was calculated using three QC samples provided by the manufacturer. For each protein, a 24‐point standard curve was established by the manufacturer using four‐parameter logistic (4PL) curve fitting. The median values of the single‐point calibrators were used to adjust to the 24‐point predefined calibration curve. The lower and upper limits of quantification (LLOQ and ULOQ) were defined by the manufacturer during panel development. Assays were excluded from further analysis if protein concentrations were below the LLOQ or above the ULOQ in more than 15% of the samples. Protein concentrations were reported as absolute concentrations (pg/mL).

2.3Specificity test

To determine cross‐talk between the 42 proteins within the T48 Neuropanel, specificity tests were performed as part of standard panel development by Olink Proteomics, and data were kindly shared with the authors for use. Recombinant antigens were serially diluted to achieve the desired target concentrations (detailed in Table S1). The target concentration for each recombinant antigen was determined based on the signal intensity observed in EDTA plasma samples analyzed in‐house for each assay. All 42 antigens were assayed in duplicate, and a pooled mixture of all 42 antigens was included as a positive control in each assay. The contribution of nonspecific signal was calculated by dividing the nonspecific signal by the true signal.

2.4NULISA CNS panel

The same CSF and plasma samples were analyzed using the NULISAseq CNS Panel (NULISA120‐CNS Panel), targeting 120 proteins associated with a broad spectrum of neurodegenerative disorders, as part of our previous work (B. Ariöz & Y.S. Hok‐A‐Hin et al., under review). All samples were randomized prior to analysis. Next‐generation sequencing data were processed using the NULISAseq algorithm (Alamar Biosciences), and intra‐plate and inter‐plate normalization were performed as described previously (B. Ariöz & Y.S. Hok‐A‐Hin et al., under review). The CV% were based on two sample controls included on each plate and were < 15%. The limit of detection (LOD) for each assay was determined as 3 standard deviations above the negative control. Assays were excluded from further analysis if protein levels were below the LOD in more than 15% of the samples. Interpolated control‐normalized counts were log2‐transformed and referred to as NULISA Protein Quantification (NPQ) units.

2.5Simoa neurology 4‐PLEX

Concentrations of plasma NfL, GFAP, Aβ42, and Aβ40 (Simoa [single molecule array] Neurology 4‐PLEX E Advantage kit; Quanterix Corp, kit lot: #503143) were previously determined in a subset of the cohort (n = 67) on the Simoa HD‐X Analyzer (Quanterix, Billerica, USA).35 Plasma samples were tested in single measurements with on‐board automated sample dilution, according to the manufacturer's instructions. Two quality control samples were included in all Simoa runs, showing good inter‐assay coefficients of variation < 15% for all biomarkers.

2.6Statistical analysis

All statistical analyzes were performed in Rstudio (version 2023.12.1). Patient demographics were tested using the Kruskal–Wallis test followed by Bonferroni correction for continuous variables or Pearson's chi‐squared test for categorical variables. Absolute values measured with the T48 Neuropanel were Log2‐transformed for all analysis where normal distribution was required.

The T48 Neuropanel (Log2) values were compared to protein levels measured by NULISA for the overlapping proteins (n = 16, Table S2) using Pearson correlation analysis. In addition, overlapping proteins (Table S2) from the T48 Neuropanel were compared to protein levels measured by larger PEA multiplex panels detailed previously.19, 22 Spearman correlation analysis was performed to compare directly between protein concentrations measured by the T48 Neuropanel (pg/mL) and Elecsys (i.e., Aβ42 assay); and between T48 Neuropanel (pg/mL) and Simoa (NfL, GFAP, Aβ42, and Aβ40 assays). In addition, Passing–Bablok regression and Bland–Altman analyzes were performed to assess platform agreement for the T48 Neuropanel with Elecsys and Simoa assays.

Differences between groups in CSF and plasma log2‐transformed protein levels were assessed using analysis of covariance (ANCOVA), corrected for age and sex. The p‐values were false discovery rate (FDR) ‐corrected for the number of proteins included in the analyzes. For the significant differentially regulated proteins, post‐hoc analyzes included the comparison of each dementia group to the control group, corrected for the total number of group comparisons using the Bonferroni method. Furthermore, to test the discriminative performance, receiver operating characteristics (ROC) analysis were performed using the raw biomarker values. Optimal biomarker cutoffs (pg/mL) were determined based on Youden's Index for each comparison (AD vs controls, FTD vs controls, and DLB vs controls). Lastly, linear regression models including age and sex were used to test the association of individual protein levels with global cognition as measured by Mini‐Mental State Examination (MMSE) scores. P < 0.05 was considered significant.

Spearman correlation analysis was used to assess the correlation for each protein concentration between CSF and plasma matrix. Correlations coefficients Spearman's r or Spearman's ρ < 0.3 are considered weak, between 0.3 and 0.5 moderate and > 0.5 are considered strong.

3RESULTS

3.1The T48 neuropanel shows good detectability and analytical performance in CSF and plasma

We first determined the analytical performance of the novel T48 Neuropanel. Specificity testing revealed that, among the 42 assays, cross‐reactivity was observed only for the OMG assay in plasma, which showed a 10% increase in signal when combined with the ITGB2 antigen (Figure S1).

Next, the detectability of the assays in CSF and plasma in our cohort was assessed. The results show that 31 proteins (74%) were successfully detected in CSF (Figure S2). Eight assays had results below the LLOQ (i.e., LRRK2, GDNF, WWOX, FOXO3, NGF, EIF2AK2, TP53, and SNCAIP) and three assays had results above the ULOQ (i.e., OMG, MOG, and NPTXR) in at least 26 out of 176 samples (15%). The results show that 38 proteins (90%) were successfully detected in plasma (Figure S3). Four assays had results below the LLOQ (i.e., S100B, LRRK2, NLGN1, and SNCAIP) in at least 26 out of 176 samples (15%).

The precision of the T48 Neuropanel was good, with a median inter‐assay CV of 7.4% and an median intra‐assay CV of 6.5% measured in the quality control samples (n = 30). Additional exploration of the 42 individual assays revealed that all assays showed good inter‐ and intra‐assay CVs (all < 15%), indicating reproducible analytical performance across runs for all assays (Table S3).

3.2Cross‐platform comparison shows moderate to strong correlations across multiple methods

We next assessed correlations between the T48 Neuropanel and other immune‐assay based platforms. A strong correlation between concentrations obtained with the CSF Elecsys Aβ42 and the T48 Aβ42 assays was observed (n = 28, ρ = 0.91, p < 0.001; Figure 1 A‐C). Passing‐Bablok analysis revealed no platform agreement with higher levels measured by the T48 Neuropanel (Figure S4A). After transforming the T48 Neuropanel values to Elecsys using the Passing‐Bablok transformation formula, values were comparable between methods (Figure S4B). For the comparison in CSF with NULISA(120‐CNS panel), 14 of the 16 overlapping assays had a strong correlation with the T48 Neuropanel (r between 0.58 and 0.95, all p < 0.001; Figure 1A), of which pTau217 and NfL showed the strongest correlation (both: r = 0.95, p < 0.001). Passing‐Bablok analysis was not performed because the NULISA(120‐CNS Panel) does not generate absolute quantification of the protein measurements. Comparison of the overlapping assays in CSF within the PEA technology revealed a moderate to strong correlation between assays (r between 0.44 and 0.90, all p < 0.001; Figure S5A).

In plasma, strong correlations were observed between the T48 Neuropanel and Simoa NfL (ρ = 0.90, p < 0.001), GFAP (ρ = 0.84, p < 0.001), Aβ42 (ρ = 0.55, p < 0.001), and Aβ40 (ρ = 0.57, p < 0.001) concentrations (Figure 1D‐F). None of these assays showed platform agreement, and thus Passing‐Bablok transformation formulas were calculated (Figure S6). After transforming the T48 Neuropanel values to the Simoa scale, no significant bias between methods was observed, indicating that the protein values between platforms are on the same scale (Supplementary Figure 6). Visual inspection of the Bland‐Altman plots of Aβ42 and Aβ40 showed a minor trend towards increasing underestimation of the biomarker values at higher concentrations. This suggests that the biomarker values are not fully aligned across the full analytical range. Furthermore, results for 11 assays showed a strong correlation between the T48 Neuropanel and the NULISA(120‐CNS Panel) platforms (r between 0.52 and 0.91, all p < 0.001; Figure 1D). Noteworthy, three assays had a weak to moderate correlation in plasma between these platforms (SMOC1, BACE, and NGF; r between 0.18 and 0.45, all p < 0.001; Figure 1D). Comparison of the overlapping assays in plasma within the PEA technology revealed a moderate to strong correlation between assays (r between 0.33 and 0.95, all p < 0.001 besides SMOC1 p < 0.05; Supplementary Figure 5B).

Overall, the T48 Neuropanel results correlated well with other immunoassay‐based methods, particularly for the established biomarkers NfL, GFAP, Aβ42, Aβ40, and pTau217, which showed slightly stronger correlations in CSF than in plasma. Passing‐Bablok formulas to convert NfL, GFAP, Aβ42, and Aβ40 on the T48 Neuropanel aligned well with Elecsys and Simoa measurements.

3.3CSF shows strong protein dysregulations and discriminative performance across neurodegenerative dementias

We next determined the protein profiles of AD, FTD, and DLB dementias compared to controls. Concentrations of 18 CSF proteins (17 increased and one decreased) were dysregulated in AD. The three CSF proteins with the largest fold changes were pTau217, Aβ42, and ITGB2 (Figure 2A). These proteins showed excellent discriminative performance compared to controls (pTau217: AUC = 1.00, 95% confidence interval [CI]: 1.00–1.00; Aβ42: AUC = 0.98, 95% CI: 0.97–1.00; ITGB2: AUC = 0.97, 95% CI: 0.94–1.00; Figure 2D). We additionally determined their cutoffs using Youden's Index (Table 2).

In FTD, seven CSF proteins (six increased and one decreased) were dysregulated, of which NfL, MMP10, and GFAP concentrations showed the strongest increase compared to controls (Figure 2B). These proteins additionally showed good to excellent performance in differentiating FTD from controls (NfL: AUC = 0.95, 95% CI: 0.89–1.00; MMP10: AUC = 0.90, 95% CI: 0.83–0.97; GFAP: AUC = 0.86, 95% CI: 0.78–0.94; Figure 2E). NfL, MMP10 and GFAP cutoffs were determined to classify FTD from controls (Table 2).

Concentrations of nine CSF proteins (seven increased and two decreased) were dysregulated in DLB. The largest protein changes were observed for DDC, Aβ42, and pTau217 proteins (Figure 2C), which additionally showed good to excellent discriminative performance (DDC: AUC = 0.90, 95% CI: 0.84–0.97; Aβ42: AUC = 0.82, 95% CI: 0.74–0.91; pTau217: AUC = 0.82, 95% CI: 0.74–0.91; Figure 2F). Their cutoffs were determined to classify DLB from controls (Table 2).

Noteworthy, fourCSF proteins (i.e., GFAP, NfL, ITGB2, and ITGAM) were significantly altered across AD, FTD, and DLB dementia (Figure S7A). Linear regression models were performed to further evaluate the associations of these biomarkers with cognitive performance, as they are dysregulated in all comparisons. CSF NfL concentrations were not associated with MMSE scores (β  =  ‐0.732, 95% CI:−1.616 to 0.152, p  =  0.106; Figure S8A). In contrast, CSF GFAP, ITGB2, and ITGAM were all significantly associated with MMSE scores (GFAP: β  =  ‐2.166, 95% CI: –3.435 to ‐0.895, p  =  0.001; ITGB2: β  =  ‐2.645, 95% CI: ‐3.878 to ‐1.412, p  <  0.001; ITGAM: β  = ‐3.262, 95% CI ‐4.800 to ‐1.724, p  <  0.001; Supplementary Figure 8B‐D). Eight CSF proteins were uniquely dysregulated in AD (i.e., VSNL1, SDC4, DDAH1, SYT1, GLRX, ENO2, BACE1, and STX1B), whereas no uniquely dysregulated CSF proteins were identified in FTD or DLB (Figure S7A).

3.4Plasma protein dysregulation is limited across neurodegenerative dementias, with moderate to strong discriminative performance

We next evaluated protein changes in plasma for AD, FTD, and DLB dementias. Concentrations of two plasma proteins were increased (i.e., pTau217 and GFAP) and of one protein were decreased (NPTXR) in AD compared to controls (Figure 3A). pTau217 and GFAP showed excellent performance to discriminate AD from controls, while the performance of NPTXR was poor (pTau217: AUC = 0.97, 95% CI: 0.92–1.00; GFAP: AUC = 0.94, 95% CI: 0.90–0.99; NPTXR: AUC = 0.60, 95% CI: 0.47–0.72; Figure 3D). In addition, their cutoffs were determined to classify AD from controls using Youden's Index (Table 2).

Three proteins were elevated in plasma of FTD patients compared to controls (i.e., NfL, GFAP, and Aβ42; Figure 3B). ROC curve analysis showed excellent discriminative performance for NfL and GFAP, whereas the performance of Aβ42 was relatively poor (NfL: AUC = 0.95, 95% CI: 0.91–0.99; GFAP: AUC = 0.89, 95% CI: 0.82–0.96; Aβ42: AUC = 0.66, 95% CI: 0.54–0.78; Figure 3E). Cutoffs for these plasma proteins were additionally determined that classify FTD from controls (Table 2).

Concentrations of three plasma proteins were increased in DLB compared to controls (i.e., GFAP, pTau217, and NfL; Figure 3C). These proteins showed moderate to good performance for discriminating DLB from controls (GFAP: AUC = 0.84, 95% CI: 0.76–0.93; pTau217: AUC = 0.74, 95% CI: 0.63–0.84; NfL: AUC = 0.87, 95% CI: 0.79–0.95; Figure 3F). Their cutoffs were determined to classify DLB from controls (Table 2).

Plasma GFAP was the only plasma protein significantly altered across AD, FTD, and DLB (Figure S7B). There was a negative association between plasma GFAP and MMSE scores (β  =  ‐1.762, 95% CI: ‐2.587 to 0.937, p  <  0.001; Figure S8E). NPTXR was uniquely dysregulated in AD, and Aβ42 was uniquely dysregulated in FTD, while no proteins were uniquely dysregulated in DLB (Figure S7B).

3.5Moderate to strong protein correlations between CSF and plasma

Comparison of concentrations between CSF and plasma was possible for 29 proteins. Concentrations of 45% of these proteins correlated significantly between CSF and plasma (ρ between 0.15 and 0.79). The strongest correlations were observed for NfL (ρ = 0.79), pTau217 (ρ = 0.71), and AGER (ρ = 0.64; Figure 4). Interestingly, many proteins that were not correlated are related to synaptic function and dysfunction (i.e., NPTX2, SYT1, KLK8, STX1B, SCG2, and CLSTN3). Noteworthy, KLK8 concentrations are higher in plasma, NPTX2, SCG2, and CLSTN3 has similar concentration ranges between CSF and plasma, while SYT1 and STX1B concentrations are 10‐fold lower in plasma compared to CSF (Supplementary Figure S9). Upon stratifying the cohort by diagnosis, some proteins correlated between matrices only in the controls (e.g., VSNL1), whereas others correlated specifically in the dementia groups (e.g., GFAP and sTREM2).

4DISCUSSION

In this study, we present the first analytical and clinical evaluation of the T48 Neuropanel in paired CSF and plasma from patients with different neurodegenerative dementias. The T48 Neuropanel shows good detectability (CSF: 74%; plasma: 90% of the 42 assays) and correlates well with other immune‐based methods. The panel contains established (i.e., pTau217, Aβ42, NfL, and GFAP) and emerging biomarkers (i.e., ITGB2, MMP10, and DDC), which showed good discriminative performance between groups. We additionally provide transformation formulas and clinical cutoffs for use in future studies. Lastly, we established correlations between CSF and plasma and showed the extent to which these correlations depended on the presence of neurodegenerative disease.

Fluid biomarkers have revolutionized the detection, diagnosis, and screening of neurodegenerative dementias, especially for AD.36, 37 The potential of the T48 Neuropanel lies in its capability to measure the concentrations of 42 proteins in a small volume of sample (1 µL), thereby reducing the time required for analysis and minimizing the amount of sample needed. Considering the novelty of this panel, we first assessed the detectability in CSF and plasma and performed a cross‐platform comparison with established immunoassay technologies, namely Elecsys, Simoa, and NULISA methods. Good detectability of the 42 proteins included in the T48 Neuropanel in CSF (76%) and plasma (90%) was observed. The reduced detectability in CSF relative to plasma is due to the assays being specifically optimized the plasma matrix. In addition, in CSF, three proteins (OMG, MOG, and NPTXR) surpassed the upper quantification limit, which further decreased the proportion of proteins that could be reliably measured. However, these proteins could became detectable when the CSF samples are diluted. LRRK2 and SNCAIP were not detected in either CSF or plasma, which is likely due to the cohort selection, as these markers have been associated with Parkinson's disease.38, 39 The data showed an overall good correlation between platforms and between assays within the PEA platform. Noteworthy, given that absolute quantification is a key advantage of the T48 Neuropanel, platform agreement was assessed to evaluate whether measurements across methods fall within a comparable range. No platform agreement was observed, which is likely due to differences in antibodies and assay setup between these technologies, which is consistent with literature showing similar observations for many plasma‐based assays.40, 41, 42 Therefore, the current study additionally provides lot‐specific Passing‐Bablok formula's and cohort‐specific cutoffs for established biomarkers (i.e., Aβ42, Aβ40, GFAP, and NfL) as a reference for future studies. The observed bias detected specifically for the Aβ42 and Aβ40 concentrations measured with the T48 Neuropanel are not directly interchangeable with those from established platforms, which is particularly significant for borderline cases and early‐stage AD where values may lie near clinical cutoffs. Consequently, assay‐specific reference ranges and decision thresholds will be required for clinical use of the T48 Neuropanel but this warrants further investigation. Furthermore, as batch‐to‐batch variation between kits may influence the transformation formula's, and cutoffs are inherently cohort‐specific, external validation across independent cohorts will be necessary to confirm their generalizability and evaluate cross‐laboratory variability.

To further explore disease‐specific protein differences and better understand the underlying pathological mechanisms across neurodegenerative disorders, we next assessed the differential protein abundance in AD, FTD, and DLB across matrices. Within CSF, we observed a wide range of protein changes across these neurodegenerative disorders. Many of the strongly dysregulated proteins in CSF are well‐established dementia biomarkers (e.g., pTau217, Aβ42, GFAP, and NfL). In addition, emerging CSF biomarkers were also detected (e.g., ITGB2, MMP10, and DDC). ITGB2 forms a heterodimeric receptor with ITGAM and is involved in cell‐surface–mediated signaling, with high expression in immune cells including microglia.43 Multiple proteomic studies have detected increased ITGB2 levels in AD, as well as in earlier clinical stages of AD.19, 21, 43, 44 The additional relationship with cognitive decline suggests its potential to serve as a disease progression markers. MMP10 was recently shown to be increased in patients with FTD, additionally detecting particularly high levels in FTD MAPT carriers.20, 45 Furthermore, DDC showed the strongest fold change in DLB, which corroborates recent work that shows the clinical value of DDC as a diagnostic biomarker for DLB and PD.10, 14 This, in turn, underscores the value of the T48 Neuropanel as a novel multiplex assay capable of quantitatively detecting DDC, complementing the existing DDC Simoa and DDC Ella assays.14

Blood‐based biomarkers offer a less invasive and more cost‐effective approach than CSF, making them highly accessible and effective for tracking disease‐related pathological processes. Some of the strong CSF protein changes were also observed in plasma (e.g., pTau217, GFAP, and NfL), in line with observations on other platforms36. Noteworthy, the strong increase of pTau217 in DLB is likely due to the inclusion of DLB patients with AD co pathology (36%). Exclusion of these cases would decrease these AD‐specific effects and shift the plasma cutoffs for distinguishing DLB from controls. Furthermore, plasma NPTXR, a marker reflecting synaptic dysfunction, was one of the few plasma proteins that was significantly decreased in AD compared with controls. This is in agreement with literature using other technologies, which showed that higher levels of NPTXR are associated with better cognitive function.46, 47, 48 This highlights the potential of NPTXR as an AD staging biomarker or as a prognostic biomarker.

Due to the paired design of this study, we could evaluate protein levels between CSF and plasma matrices. Protein changes were generally more pronounced in CSF than in plasma, which is likely due to the close association of CSF to the brain pathology. In addition, 45% of the proteins were significantly correlated between CSF and plasma. Interestingly, proteins that did not correlate between matrices were enriched in proteins related to synaptic function and dysfunction. For these proteins, different protein profiles between CSF and plasma were observed. Specifically, KLK8 concentrations were higher in plasma compared to CSF, which is likely due to additional expression within the pancreas.49 Overall, it is likely that peripheral sources or peripheral metabolism may mask CNS‐specific changes in blood.50 When stratifying the analyzes by clinical diagnosis, we found that certain proteins correlated between CSF and plasma specifically within the dementia groups (e.g., GFAP and sTREM2). The increased protein levels observed in CSF likely reflect disease‐related processes, including enhanced release from reactive astrocytes and increased cleavage of extracellular domains, as described for GFAP and sTREM2, respectively.51, 52 The positive correlation between CSF and plasma levels suggests that increased CSF concentrations may also contribute to higher plasma concentrations. One potential mechanism is impaired blood–brain barrier integrity in dementia, enabling brain‐derived proteins to enter the peripheral circulation.53 Together, these findings reiterate that CSF remains an important matrix for studying brain diseases.

This study is not without limitations. Specifically, not all proteins from the T48 Neuropanel could be compared with those assessed on other platforms (e.g., pTau217), highlighting the need for future comparative analyzes. In addition, other biological matrices such as serum or urine were not evaluated, however, in biomarker studies of neurodegenerative dementias, biobanked samples typically include CSF and plasma. Furthermore, comorbidities such as body mass index (BMI) and chronic kidney disease could influence plasma biomarker levels in clinical disease groups54, 55 However, such analyzes were outside of the scope of the current study. Lastly, some of our analysis may be limited by the relatively small sample sizes within the diagnostic groups, underscoring the need for validation in larger, independent cohorts. One of the strengths of this study is the workflow we applied. We comprehensively assessed the analytical performance of the T48 Neuropanel and calculated cross‐platform transformation formulas. In addition, the paired design of this study allowed us to simultaneously assess protein changes in both CSF and plasma, thereby testing both well‐characterized and emerging fluid biomarkers.

5CONCLUSION

The T48 Neuropanel represents a novel tool for gaining biological insights into various neurodegenerative diseases. Its multiplex design, combined with absolute protein quantification of established and emerging CSF (Aβ42, pTau217, GFAP, NfL, ITGB2, MMP10, and DDC) and plasma biomarkers (Aβ42, pTau217, NfL, GFAP, and NPTXR), enables a rich biological interpretation and supports its potential for clinical translation. We have performed analytical and clinical evaluation of the T48 Neuropanel, additionally providing transformation formulas and clinical cutoffs for future studies.

AUTHOR CONTRIBUTIONS

Yanaika S. Hok‐A‐Hin, Lisa Vermunt, and Charlotte E. Teunissen conceived and designed the study. Y.H. performed the statistical analysis. Yanaika S. Hok‐A‐Hin, Lisa Vermunt, Marta del Campo Milan, Everard Vijverberg, Afina W. Lemstra, Argonde C. van Harten, Yolande Pijnenburg, and Charlotte E. Teunissen recruited participants and collected clinical data and samples. Natascha de Jong performed laboratory experiments. Yanaika S. Hok‐A‐Hin, Lisa Vermunt, and Charlotte E. Teunissen drafted the manuscript. All authors contributed to the revision and editing of the manuscript.

CONFLICT OF INTEREST STATEMENT

Y.H., L.V., N.J., E.V., A.L., Y.P., A.C. have nothing to disclose. M.C. has been an invited speaker at Eisai and Novonordisk. She is an associate editor at Alzheimer´s Research & Therapy and has been an invited writer for Springer Healthcare. C.T. is listed as an inventor on a patent application concerning the use of AcTau174 as a biomarker (EPO application EP26151547.2). C.T. has “research contracts” with Acumen, ADx Neurosciences, AC Immune, Alamar, Aribio, Axon Neurosciences, Beckman‐Coulter, BioConnect, Bioorchestra, Brainstorm Therapeutics, C2N diagnostics, Celgene, Cognition Therapeutics, EIP Pharma, Eisai, Eli Lilly, Fujirebio, Instant Nano Biosensors, Merck, Muna, Nitrase Therapeutics, Novo Nordisk, Olink, PeopleBio, Quanterix, Roche, Sysmex, Toyama, Vaccinex, Vivoryon. She is editor‐in‐chief of Alzheimer Research and Therapy, and serves on editorial boards of Molecular Neurodegeneration, Alzheimer's & Dementia, Neurology—Neuroimmunology & Neuroinflammation, Medidact Neurologie/Springer, and is a committee member to define guidelines for Cognitive disturbances, and one for acute Neurology in the Netherlands. She has ‘consultancy/speaker contracts’ for Aribio, Biogen, Beckman‐Coulter, Cognition Therapeutics, Danaher, Eisai, Eli Lilly, Janssen, Merck, Neurogen Biomarking, Nordic Biosciences, Novo Nordisk, Novartis, Olink, Quanterix, Roche, Sanofi and Veravas. The other authors declare no competing interests. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All participants provided written Informed consent was obtained from all participants or their authorized representatives, in accordance with the ethical approval of the Amsterdam UMC IRB and in agreement with the Helsinki Declaration of 1975.

Supporting information

REFERENCES

1WHO
, Global action plan on the public health response to dementia .: World Health Organization, 2017 .

2Alzheimer's disease facts and figures . Alzheimers Dement , 2023 ;19 (4 ): p. 1598 ‐1695 .10.1002/alz.13016 36918389

3Scheltens P , De Strooper B , Kivipelto M , et al. Alzheimer's disease . Lancet . 2021 ;397 (10284 ):1577 ‐1590 . doi:10.1016/S0140-6736(20)32205-4
PMC8354300 33667416

4Del Campo M , et al. New developments of biofluid‐based biomarkers for routine diagnosis and disease trajectories in frontotemporal dementia . Alzheimers Dement . 2022 ;18 (11 ):2292 ‐2307 .10.1002/alz.12643 PMC9790674 35235699

5Outeiro TF , Koss DJ , Erskine D , et al. Dementia with Lewy bodies: an update and outlook . Mol Neurodegener . 2019 ;14 (1 ):5 . doi:10.1186/s13024-019-0306-8
PMC6341685 30665447

6Jack CR , Andrews JS , Beach TG , et al. Revised criteria for diagnosis and staging of Alzheimer's disease: alzheimer's Association Workgroup . Alzheimers Dement . 2024 ;20 (8 ):5143 ‐5169 . doi:10.1002/alz.13859
PMC11350039 38934362

7Ducharme S , Dols A , Laforce R , et al. Recommendations to distinguish behavioural variant frontotemporal dementia from psychiatric disorders . Brain . 2020 ;143 (6 ):1632 ‐1650 . doi:10.1093/brain/awaa018
PMC7849953 32129844

8Rossi M , Candelise N , Baiardi S , et al. Ultrasensitive RT‐QuIC assay with high sensitivity and specificity for Lewy body‐associated synucleinopathies . Acta Neuropathol . 2020 ;140 (1 ):49 ‐62 . doi:10.1007/s00401-020-02160-8
PMC7299922 32342188

9Simuni T , Chahine LM , Poston K , et al. A biological definition of neuronal alpha‐synuclein disease: towards an integrated staging system for research . Lancet Neurol . 2024 ;23 (2 ):178 ‐190 . doi:10.1016/S1474-4422(23)00405-2
38267190

10Del Campo M , Vermunt L , Peeters CFW , et al. CSF proteome profiling reveals biomarkers to discriminate dementia with Lewy bodies from Alzheimer s disease . Nat Commun . 2023 ;14 (1 ):5635 .10.1038/s41467-023-41122-y PMC10499811 37704597

11Paslawski W , Khosousi S , Hertz E , Markaki I , Boxer A , Svenningsson P . Large‐scale proximity extension assay reveals CSF midkine and DOPA decarboxylase as supportive diagnostic biomarkers for Parkinson's disease . Transl Neurodegener . 2023 ;12 (1 ):42 . doi:10.1186/s40035-023-00374-w PMC10476347 37667404

12Pereira JB , Kumar A , Hall S , et al. DOPA decarboxylase is an emerging biomarker for Parkinsonian disorders including preclinical Lewy body disease . Nat Aging . 2023 ;3 (10 ):1201 ‐1209 . doi:10.1038/s43587-023-00478-y
PMC10570139 37723208

13Rutledge J , Lehallier B , Zarifkar P , et al. Comprehensive proteomics of CSF, plasma, and urine identify DDC and other biomarkers of early Parkinson's disease . Acta Neuropathol . 2024 ;147 (1 ):52 . doi:10.1007/s00401-024-02706-0
PMC10927779 38467937

14Bolsewig K , Bellomo G , Hok‐A‐Hin YS , et al. A quantitative DOPA decarboxylase biomarker for diagnosis in Lewy body disorders . Nat Med . 2026 ;32 (3 ):1073 ‐1084 . doi:10.1038/s41591-026-04212-0
PMC13004688 41699128

15Cummings JL , Zhou Y , Lee G , et al. Alzheimer's disease drug development pipeline: 2025 . Alzheimers Dement (N Y) . 2025 ;11 (2 ):e70098. doi:10.1002/trc2.70098
PMC12131090 40463637

16Gezegen Haş , Alaylıoğlu M , Şahin E , et al. Unravelling the plasma proteome: pioneering biomarkers for differential dementia diagnosis . Alzheimers Dement . 2025 ;21 (7 ):e70162. doi:10.1002/alz.70162
PMC12231214 40613333

17Higginbotham L , Shantaraman A , Guo Q , et al. Multiplex Proteomics of Lewy Body Dementia Reveals Cerebrospinal Fluid Biomarkers of Disease Pathology and Progression . bioRxiv . 2025 :2025.06.10.658994. doi:10.1101/2025.06.10.658994

18Wang Yi‐T , Ashton NJ , Therriault J , et al. Identify biological Alzheimer's disease using a novel nucleic acid‐linked protein immunoassay . Brain Commun . 2025 ;7 (1 ):fcaf004. doi:10.1093/braincomms/fcaf004
PMC11753389 39845736

19Del Campo M , Peeters CFW , Johnson ECB , et al. CSF proteome profiling across the Alzheimer's disease spectrum reflects the multifactorial nature of the disease and identifies specific biomarker panels . Nat Aging . 2022 ;2 (11 ):1040 ‐1053 .10.1038/s43587-022-00300-1 PMC10292920 37118088

20Hok‐A‐Hin YS , Vermunt L , Peeters CFW , et al. Large‐scale CSF proteome profiling identifies biomarkers for accurate diagnosis of frontotemporal dementia . Mol Neurodegener . 2025 ;20 (1 ):93 . doi:10.1186/s13024-025-00882-5
PMC12392567 40866991

21Del Campo M , Quesada C , Vermunt L , et al. CSF proteins of inflammation, proteolysis and lipid transport define preclinical AD and progression to AD dementia in cognitively unimpaired individuals . Mol Neurodegener . 2024 ;19 (1 ):82 .10.1186/s13024-024-00767-z PMC11552178 39523360

22Bellomo G , Vermunt L , in ′t Veld S , et al. Plasma proteome profiling identified biomarkers for the differential diagnosis and molecular staging of neurodegenerative dementias . Nature Aging . 2026 ;6 (8 ):1717 ‐1732 . doi:10.1038/s43587-026-01162-7 . Forthcoming.PMC13472907 42477094

23van der Flier WM , Pijnenburg YA , Prins N , et al. Optimizing patient care and research: the Amsterdam Dementia Cohort . J Alzheimers Dis . 2014 ;41 (1 ):313 ‐327 .10.3233/JAD-132306 24614907

24van der Flier WM , Scheltens P . Amsterdam dementia cohort: performing research to optimize care . J Alzheimers Dis . 2018 ;62 (3 ):1091 ‐1111 .10.3233/JAD-170850 PMC5870023 29562540

25McKhann GM , Knopman DS , Chertkow H , et al. The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging‐Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease . Alzheimers Dement . 2011 ;7 (3 ):263 ‐269 . doi:10.1016/j.jalz.2011.03.005
PMC3312024 21514250

26Rascovsky K , Hodges JR , Knopman D , et al. Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia . Brain . 2011 ;134 (Pt 9 ):2456 ‐2477 . doi:10.1093/brain/awr179
PMC3170532 21810890

27McKeith IG , Boeve BF , Dickson DW , et al. Diagnosis and management of dementia with Lewy bodies: Fourth consensus report of the DLB Consortium . Neurology . 2017 ;89 (1 ):88 ‐100 .10.1212/WNL.0000000000004058 PMC5496518 28592453

28Jessen F , Amariglio RE , Buckley RF , et al. The characterisation of subjective cognitive decline . Lancet Neurol . 2020 ;19 (3 ):271 ‐278 . doi:10.1016/S1474-4422(19)30368-0
PMC7062546 31958406

29Hok‐A‐Hin YS , Willemse EAJ , Teunissen CE , Del Campo M . Guidelines for CSF Processing and Biobanking: impact on the Identification and Development of Optimal CSF Protein Biomarkers . Methods Mol Biol . 2019 ;2044 :27 ‐50 . doi:10.1007/978-1-4939-9706-0
31432404

30Verberk IMW , Gouda M , Antwi‐Berko D , et al. Evidence‐based standardized sample handling protocol for accurate blood‐based Alzheimer's disease biomarker measurement: results and consensus of the Global Biomarker Standardization Consortium . Alzheimers Dement . 2025 ;21 (10 ):e70752. doi:10.1002/alz.70752
PMC12481164 41025225

31Duits FH , Teunissen CE , Bouwman FH , et al. The cerebrospinal fluid “Alzheimer profile”: easily said, but what does it mean . Alzheimers Dement . 2014 ;10 (6 ):713 –723 e2. doi:10.1016/j.jalz.2013.12.023
24721526

32van Harten AC , Wiste HJ , Weigand SD , et al. Detection of Alzheimer's disease amyloid beta 1‐42, p‐tau, and t‐tau assays . Alzheimers Dement . 2022 ;18 (4 ):635 ‐644 .10.1002/alz.12406 PMC9249966 34310035

33Tijms BM , Willemse EAJ , Zwan MD , et al. Unbiased Approach to Counteract Upward Drift in Cerebrospinal Fluid Amyloid‐beta 1‐42 Analysis Results . Clin Chem . 2018 ;64 (3 ):576 ‐585 . doi:10.1373/clinchem.2017.281055
29208658

34Rascovsky K , Hodges JR , Knopman D , et al. Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia . Brain . 2011 ;134 (9 ):2456 ‐2477 . doi:10.1093/brain/awr179
PMC3170532 21810890

35Doecke JD , Bellomo G , Vermunt L , et al. Diagnostic performance of plasma Abeta42/40 ratio, p‐tau181, GFAP, and NfL along the continuum of Alzheimer's disease and non‐AD dementias: an international multi‐center study . Alzheimers Dement . 2025 ;21 (6 ):e14573. doi:10.1002/alz.14573
PMC12185245 40551285

36Teunissen CE , Verberk IMW , Thijssen EH , et al. Blood‐based biomarkers for Alzheimer's disease: towards clinical implementation . Lancet Neurol . 2022 ;21 (1 ):66 ‐77 . doi:10.1016/S1474-4422(21)00361-6
34838239

37Zetterberg H , Bendlin BB . Biofluid biomarkers in Alzheimer's disease and other neurodegenerative dementias . Nature . 2026 ;650 (8100 ):49 ‐59 . doi:10.1038/s41586-025-10018-w
41639575

38Magalhães P , Lashuel HA . Opportunities and challenges of alpha‐synuclein as a potential biomarker for Parkinson's disease and other synucleinopathies . NPJ Parkinsons Dis . 2022 ;8 (1 ):93 . doi:10.1038/s41531-022-00357-0 PMC9307631 35869066

39Rui Q , Ni H , Li Di , Gao R , Chen G . The Role of LRRK2 in Neurodegeneration of Parkinson Disease . Curr Neuropharmacol . 2018 ;16 (9 ):1348 ‐1357 . doi:10.2174/1570159X16666180222165418
PMC6251048 29473513

40De Meyer S , Schaeverbeke JM , Verberk IMW , et al. Comparison of ELISA‐ and SIMOA‐based quantification of plasma Abeta ratios for early detection of cerebral amyloidosis . Alzheimers Res Ther . 2020 ;12 (1 ):162 .10.1186/s13195-020-00728-w PMC7719262 33278904

41Ashrafzadeh‐Kian S , Figdore D , Larson B , et al. Head‐to‐head comparison of four plasma neurofilament light chain (NfL) immunoassays . Clin Chim Acta . 2024 ;561 :119817. doi:10.1016/j.cca.2024.119817
38879065

42Arslan B , Rembeza E , Rasch S , et al. Method comparison of plasma and CSF GFAP immunoassays across multiple platforms . Clin Chem Lab Med . 2025 ;63 (12 ):2447 ‐2459 . doi:10.1515/cclm-2025-0667
40887826

43Kamalian A , Ho SG , Patel M , et al. Exploratory Assessment of Proteomic Network Changes in Cerebrospinal Fluid of Mild Cognitive Impairment Patients: a Pilot Study . Biomolecules . 2023 ;13 (7 ):1094. doi:10.3390/biom13071094
PMC10377001 37509130

44van der Ende EL , In 't Veld SGJG , Hanskamp I , et al. CSF proteomics in autosomal dominant Alzheimer's disease highlights parallels with sporadic disease . Brain . 2023 ;146 (11 ):4495 ‐4507 .10.1093/brain/awad213 PMC10629764 37348871

45De Houwer JFH , Dopper EG , van Buuren R , et al. Distinct proteomic CSF profiles in genetic frontotemporal lobar degeneration . Brain . 2026 ;149 (8 ):2816 ‐2830 .10.1093/brain/awaf457 PMC13431664 41343108

46Huang Y‐N , Liu S , Park T , et al. Plasma proteomic analysis identifies proteins and pathways related to Alzheimer's risk . Alzheimers Dement . 2025 ;21 (8 ):e70579. doi:10.1002/alz.70579
PMC12351396 40810263

47Afshar S , Dammer EB , Bian S , et al. Plasma proteomic associations with Alzheimer's disease endophenotypes . Nat Aging . 2025 ;5 (10 ):2104 ‐2124 . doi:10.1038/s43587-025-00965-4
PMC12532695 40931114

48Ali M , Erabadda B , Chen Y , et al. Shared and disease‐specific pathways in frontotemporal dementia and Alzheimer's and Parkinson's diseases . Nat Med . 2025 ;31 (8 ):2567 ‐2577 . doi:10.1038/s41591-025-03833-1
PMC12404994 40665052

49Uhlén M , Fagerberg L , Hallström BM , et al. Tissue‐based map of the human proteome . Science . 2015 ;347 (6220 ):1260419.10.1126/science.1260419 25613900

50Kvartsberg H , Portelius E , Andreasson U , et al. Characterization of the postsynaptic protein neurogranin in paired cerebrospinal fluid and plasma samples from Alzheimer's disease patients and healthy controls . Alzheimers Res Ther . 2015 ;7 (1 ):40 . doi:10.1186/s13195-015-0124-3
PMC4487851 26136856

51Gogishvili D , Honey MIJ , Verberk IMW , et al. The GFAP proteoform puzzle: how to advance GFAP as a fluid biomarker in neurological diseases . J Neurochem . 2025 ;169 (1 ):e16226. doi:10.1111/jnc.16226
PMC11658191 39289040

52Yang J , Fu Z , Zhang X , Xiong M , Meng L , Zhang Z . TREM2 ectodomain and its soluble form in Alzheimer's disease . J Neuroinflammation . 2020 ;17 (1 ):204 . doi:10.1186/s12974-020-01878-2 PMC7341574 32635934

53Alkhalifa AE , Al‐Ghraiybah NF , Odum J , Shunnarah JG , Austin N , Kaddoumi A . Blood‐Brain Barrier Breakdown in Alzheimer's Disease: mechanisms and Targeted Strategies . Int J Mol Sci . 2023 ;24 (22 ):16288. doi:10.3390/ijms242216288 PMC10671257 38003477

54Syrjanen JA , Campbell MR , Algeciras‐Schimnich A , et al. Associations of amyloid and neurodegeneration plasma biomarkers with comorbidities . Alzheimers Dement . 2022 ;18 (6 ):1128 ‐1140 . doi:10.1002/alz.12466
PMC8957642 34569696

55Daniilidou M , Öhlund‐Wistbacka U , Hagman G , et al. Enhancing diagnostic precision in Alzheimer's disease: impact of comorbidities on blood biomarkers for clinical integration . Alzheimers Dement . 2025 ;21 (12 ):e70931. doi:10.1002/alz.70931
PMC12668905 41326959