Exploratory Blood Biomarker Patterns in a Mixed Dementia Cohort

Alzheimers disease (AD) diagnosis is challenging due to overlapping symptoms with other dementias. Current diagnostic methods are invasive and costly, highlighting the need for accessible biomarkers. This study investigates the diagnostic performance and pathophysiological implications of a novel plasma biomarker panel in a mixed dementia cohort, aiming to enhance diagnosis and elucidate underlying pathogenic mechanisms. 120 plasma biomarkers were analyzed using the NULISA platform in a well-characterized mixt dementia. CSF biomarkers were measured via ELISA. Statistical analyses employed ANOVA, and Kruskal-Wallis tests for group comparisons. Spearman correlations assessed relationships between CSF and plasma biomarkers. Diagnostic accuracy was evaluated using regression models and ROC curves. Feature importance and selection were performed using random forest analysis. Protein interactions assessed with GO enrichment analysis. We evaluated 248 subjects (130 females, 118 males) with 117 AD, 50 MCI, 39 FTD, 25 DLB, and 17 other dementias. Plasma pTau were significantly elevated in AD compared to other groups, and in DLB compared to MCI. Plasma A{beta}42 was highest in DLB, while NfL was highest in FTD. Plasma GFAP was highest in AD and elevated in DLB compared to MCI and FTD. Plasma pTau levels showed a negative correlation with CSF A{beta}42 and a positive correlation with CSF pTau in the entire cohort. CSF and plasma NfL levels were also highly correlated. These correlations were stronger in DLB and amyloid-positive MCI groups but weaker or absent in the AD group. Plasma pTau, GFAP, and NfL were negatively correlated with MMSE in AD, while GFAP showed a negative correlation with MMSE in FTD and DLB. Plasma pTau217 demonstrated the best diagnostic accuracy for AD, DLB, FTD diagnosis and CSF amyloid positivity (AUCs 0.9, 0.84, 0.79, and 0.87, respectively). pTau181, pTau217, pTau231, total-tau and GFAP had lower odds in DLB and FTD compared to AD. AGRN, CXCL1, SCNB, TEK, and UCHL1 had higher odds in DLB compared to AD. SNAP25 had lower odds ratio in FTD compared to AD and DLB. pTau181, pTau217, pTau231, GFAP, MAPT, SNAP25 and PGF is related to AD progression. Random forest analysis incorporating all plasma biomarkers, age, and gender yielded an AUCs of 0.85 for AD, 0.84 for FTD, and 0.75 for DLB. Refining the model by including biomarkers identified as significant in regression model improved performance, resulting in AUCs of 0.88 for AD, 0.87 for FTD, and 0.81 for DLB. This demonstrates potential for enhancing diagnostic accuracy through targeted biomarker panel refinement.