Dementia with Lewy bodies (DLB) is a common dementia, clinically characterized by cognitive fluctuations, visual hallucinations, parkinsonism, and REM sleep behavior disorder. Pathologically, it's defined by α-synuclein accumulation in Lewy bodies in the neocortex. The clinical and pathological overlap with Alzheimer's disease (AD) makes differential diagnosis difficult, leading to underdiagnosis. Current biomarkers, such as α-synuclein in CSF (results are conflicting), and core AD CSF biomarkers (amyloid β peptide (Aβ1-42), total tau (tTau), and phosphorylated tau (pTau)) offer limited accuracy in differentiating DLB from AD due to comorbid pathologies. Therefore, additional markers reflecting unique aspects of DLB pathophysiology are needed for improved diagnosis, prognosis, disease monitoring, and clinical trials. Cerebrospinal fluid (CSF) proteome profiling can identify changes across various biological processes, revealing potential biomarker candidates. Previous DLB proteomic studies had limited sample sizes, hindering biomarker identification. This study employed a high-throughput proteomics method (Proximity Extension Assay, PEA) to analyze large cohorts, enabling the development of custom multiplex immunoassays for large-scale validation. The aims were to define CSF proteomic changes in DLB and to identify, develop, and validate multiplex biomarker assays for DLB diagnosis.

The literature review highlights the challenges in diagnosing DLB due to its significant overlap with AD. While studies have explored α-synuclein as a CSF biomarker, results have been inconsistent. Similarly, the use of established AD CSF biomarkers (Aβ1-42, tTau, pTau) has shown limited success in differentiating DLB from AD because of the frequent comorbidity of both pathologies. The need for additional, DLB-specific biomarkers is emphasized to improve diagnostic accuracy and guide clinical trials. A few previous proteomic studies on DLB have been conducted, but their limited sample sizes and the inherent heterogeneity of DLB have hampered the identification of robust biomarkers. This study builds on these limitations by employing a high-throughput method and a larger cohort to increase the likelihood of identifying and validating DLB-specific biomarkers.

The study used a multi-cohort design involving a discovery cohort and three independent validation cohorts. The discovery cohort (n=534) comprised CSF samples from patients with DLB (n=109), AD (n=235), and cognitively unimpaired controls (n=190). CSF proteins were quantified using proximity extension assays (PEA), measuring 665 proteins. Three independent cohorts were used for validation: clinical validation cohort 1 (n=164), clinical validation cohort 2 (n=165), and an AD/DLB autopsy-confirmed cohort (n=76). Demographic characteristics, AD CSF biomarkers (Aβ42, tTau, pTau), and clinical features were recorded. The PEA data underwent statistical analysis using nested linear models to identify differentially regulated proteins between groups, controlling for age and sex. Classification analysis was performed to identify the optimal combination of biomarkers to discriminate DLB from AD and controls. Custom multiplex PEA panels were developed for six of the selected proteins and validated in the independent cohorts. Correlation analyses were conducted to investigate the relationship between the identified biomarkers, cognitive function (MMSE), and AD CSF biomarkers. Furthermore, exploratory analyses examined the association between CSF DDC and DLB-related pathophysiological features (UPDRS-III, post-mortem brain α-synuclein load, Braak stage, DLB stage) in a subset of autopsy-confirmed cases. Publicly available CSF PEA proteomic data from the Parkinson's Progression Markers Initiative (PPMI) were also analyzed to investigate the dysregulation of these biomarkers in Parkinson's disease (PD).

CSF proteome profiling revealed 14 proteins differentially regulated in DLB compared to controls after multiple testing correction. Dopa decarboxylase (DDC) was the most strongly dysregulated protein, significantly increased in DLB. DDC effectively discriminated DLB from controls (AUC=0.91) and AD (AUC=0.81). Classification analysis identified a seven-protein CSF biomarker panel that further improved discrimination of DLB from AD (AUC=0.93). This panel included DDC, along with proteins associated with immune function, extracellular remodeling, and the hypothalamic-pituitary-adrenal axis. A custom multiplex panel for six of these markers showed high correlation and similar diagnostic performance in three independent validation cohorts, including an autopsy-confirmed cohort (AUCs ≥ 0.86). CSF DDC correlated positively with UPDRS-III score, α-synuclein load in specific brain regions, Braak stage, and DLB stage. The DLB-related markers (DDC, FCER2, CRH, MMP3) were also dysregulated in prodromal and symptomatic PD, further supporting their relevance to α-synucleinopathy.

This study identified a panel of CSF biomarkers that can accurately distinguish DLB from AD and controls. The findings strongly support the involvement of dopamine biosynthesis (DDC) and myelination processes in DLB pathophysiology. The high performance of the biomarker panel across multiple independent cohorts, including an autopsy-confirmed cohort, underscores its robustness and potential clinical utility. The identification of DDC as a strong discriminator is noteworthy given its role in dopamine biosynthesis, a key feature of DLB. The overlapping dysregulation of these biomarkers in PD reinforces the common pathological mechanisms in α-synucleinopathies. However, the study also highlighted the influence of medication on some biomarkers. Future studies need to confirm the impact of treatment on CSF FCER2 and assess the potential clinical use of these biomarkers alongside other established markers.

This large-scale CSF proteome study successfully identified and validated a panel of CSF biomarkers for the accurate diagnosis of DLB. The strong performance of the panel, particularly the prominent role of DDC, offers a valuable tool for differentiating DLB from AD and controls. The study's findings provide a promising avenue for improving DLB diagnosis, monitoring disease progression, and guiding clinical trials. Future research should focus on longitudinal studies to assess the panel's utility in tracking disease progression and treatment response and explore the role of these biomarkers in other synucleinopathies.

While this study included a large sample size and multiple cohorts, some limitations should be considered. The influence of Parkinsonian medication on some biomarker levels was observed, although it did not affect the overall discriminative ability of the panel. The clinicopathological overlap between DLB and AD might have led to some misdiagnosis. Despite the use of autopsy confirmation and DAT scans in a substantial number of cases, the possibility of diagnostic error remains. Future studies involving larger, well-characterized cohorts with detailed clinical and neuropathological data are needed to further refine the diagnostic accuracy and explore the clinical utility of the panel in diverse settings.