This is the first reported meta-analysis of the CSF NSE levels of AD, PD, DLB, and MSA patients, and it provided evidence about the significance of CSF NSE levels in AD and PDD/DLB.
1. CSF NSE levels in AD
This study detected significantly elevated CSF NSE levels in AD patients, which are considered to reflect the neurodegenerative processes that occur in AD, and the sensitivity analysis of the AD-related data confirmed the consistency of the results. Therefore, this study indicated that the CSF NSE level might be useful as an objective surrogate biochemical marker of AD-related neuronal damage. T-tau is widely accepted as a biomarker of AD-related neurodegeneration [27], and the current study suggested that NSE could also be used for such purposes.
Meta-regression analysis revealed that age contributed to the high heterogeneity of the data, and this point should be considered when interpretating the CSF NSE levels of AD patients. However, a previous study reported CSF NSE levels were not correlated with age or sex in a normal population [28]. Other possible causes of the heterogeneity include sex, the spatial distribution of pathological changes, disease activity/rapidity, disease stage, genotypes (e.g., apolipoprotein ε4), and confounding vascular risk factors (hypertension, diabetes mellitus, dyslipidemia, etc.). Indeed, a previous study detected a significant correlation among NSE, Aβ42, and total tau levels [13]. Generally, NSE is seen as a marker of neurodegeneration, whereas amyloid and tau are regarded as markers of upstream changes in AD. Further studies are needed to examine these points.
The high heterogeneity observed in the CSF NSE levels of the AD patients in this study raises other possibilities, such as the effects of diagnostic accuracy. The clinical diagnosis of AD is based on the relevant criteria, but post-mortem pathological verification has demonstrated the difficulty of achieving an accurate pre-mortem diagnosis and differentiating AD from non-Alzheimer’s dementias, such as argyrophilic grain disease, primary age-related tauopathy, and other AD-mimicking disorders, pre-mortem [29]. A previous set of diagnostic criteria (NINCDS-ADRDA) exhibited high sensitivity (93%) for diagnosing AD and frontotemporal dementia, but low specificity (23%) [30]. However, a study of the latest criteria for AD (the IWG-2 criteria), which include criteria relating to CSF biomarkers and amyloid positron emission tomography (PET), reported that the use of this combination of biomarkers resulted in a sensitivity value of 90–95% and a specificity value of about 90% for diagnosing AD, and the agreement between florbetapir amyloid PET images and post-mortem results reached 96% [31]. Therefore, the updating of diagnostic criteria to account for new methodologies could have contributed to the heterogeneity observed in this study. Other possible reasons for the heterogeneity include variations in the disease duration, stage, or activity of AD, and differences among the subtypes of AD [32]. The meta-regression analysis of AD conducted in this study showed that age contributed to the heterogeneity in the CSF NSE levels of the AD patients. Many previous studies did not stratify their data according to disease duration or stage. In addition, another study did not detect a clear difference between the CSF NSE levels of early-onset and late-onset AD patients [11], but further studies with larger samples are needed to examine this issue. Moreover, technical factors, such as sampling procedures or assay methods, should be evaluated to clarify whether they can explain the heterogeneity of the results.
2. CSF NSE levels in PD, PDD, and DLB
This study revealed significantly elevated CSF NSE levels in PDD/DLB, but not in PD.
The detection of significant changes in CSF NSE levels in both AD and DLB reminds us that AD and DLB share common amyloid β- and tau-related pathologies [6], but examining the effects of these pathologies on CSF NSE levels would require further studies, such as studies involving amyloid or tau PET. The meta-regression analysis of PDD/DLB did not show significant effects of age or sex on CSF NSE levels, but these results were inconclusive because of the small number of studies included.
3. NSE levels in MSA
This study did not reveal any significant changes in the CSF NSE level in MSA, and significant evidence of publication bias relating to this topic was detected. Recent studies have identified other potentially useful biomarkers of MSA, such as α-synuclein, neurofilament light chain, DJ-1, 8-hydroxyguanosine (8OHG), Flt3 (Fms-related tyrosine kinase) ligand, YKL-40 (CHI3L1), and ubiquitin carboxy-terminal hydrolase L1 (UCHL-1) [33]. Further studies are needed to identify the optimal molecular biomarkers of MSA.
This study had some limitations. First, the CSF NSE level data exhibited high heterogeneity, and there were quite large overlaps between the disease groups and controls. Therefore, the application of this biomarker to clinical practice should be performed cautiously. Second, elevated CSF NSE levels reflect neuronal damage, but are not disease-specific. Several of the studies included in this study adopted panels of biomarkers (amyloid β, total tau, phosphorylated tau, α-synuclein, and neurofilament light chain, etc.) to detect combinations of molecular pathological changes.
Nevertheless, measuring CSF NSE levels could be useful because NSE assays are available in many laboratories.
In conclusion, this meta-analysis revealed significantly elevated CSF NSE levels in AD and PDD/DLB, but not in MSA. This study will aid our understanding of the pathological mechanisms underlying these diseases and support further investigations, more accurate diagnosis, and evaluations of therapeutic interventions.