Crunching the numbers. Responsibly.
Alterations in plasma sphingomyelin (SM) levels have been reported in Alzheimer's disease (AD), pointing to disturbances in lipid metabolism that may contribute to disease pathogenesis. Neuronal damage in early AD triggers tau release into central and peripheral systems. Despite influence from peripheral contributions, alterations in plasma total-tau (T-tau) remain valuable in indicating AD-related neurodegeneration. Investigating relationships between SM metabolism and tau release during preclinical AD may uncover important biochemical processes and support advancing early non-invasive detection and treatment approaches. This cross-sectional study investigated cognitively unimpaired (CU) older adults from the KARVIAH cohort, grouped by cortical amyloid-β (Aβ) status through positron emission tomography (PET) imaging (CU Aβ- and CU Aβ+) and utilised a Biocrates-targeted metabolomic platform and Single-molecule array (Simoa) technology to quantify plasma levels of SMs and T-tau, respectively. Associations between circulating SMs and T-tau were examined within each group, with T-tau-associated SMs further evaluated for their association with cognitive performance and cortical Aβ burden and their potential to discriminate CU Aβ+ from CU Aβ- individuals. Significant positive correlations were observed between SMs and T-tau levels exclusively in CU Aβ+ individuals, suggesting connections between SM-mediated biochemical pathways and tau release from early neurodegeneration in preclinical AD. Lower SM levels were associated with weaker working memory and executive function, as well as poorer global cognition, indicating their potential predictive value for weaker cognitive performance. Moreover, SMs were also inversely associated with cortical Aβ load in CU Aβ+ individuals, possibly reflecting early SM-mediated neuroprotective responses against AD pathogenesis. Receiver operating characteristic analysis further revealed the significant potential of the SM panel in distinguishing cortical PET-Aβ status and enhancing the predictive performance of plasma T-tau in CU individuals. Therefore, circulating T-tau-associated SMs may serve as promising early biomarkers of lipid-mediated processes in CU older adults with cortical amyloid pathology and tau-related neurodegeneration.
Significant positive correlations were observed between SMs and T-tau levels exclusively in CU Aβ+ individuals, suggesting connections between SM-mediated biochemical pathways and tau release from early neurodegeneration in preclinical AD.
cross-sectional stratified associations without a reported interaction test to establish that the relationship differs by amyloid status
Lower SM levels were associated with weaker working memory and executive function, as well as poorer global cognition, indicating their potential predictive value for weaker cognitive performance.
cross-sectional cognitive associations with conflicting Methods and Results descriptions of the statistical test
SMs were also inversely associated with cortical Aβ load in CU Aβ+ individuals, possibly reflecting early SM-mediated neuroprotective responses against AD pathogenesis.
cross-sectional correlations support association but cannot identify a neuroprotective response or its temporal direction
Receiver operating characteristic analysis further revealed the significant potential of the SM panel in distinguishing cortical PET-Aβ status and enhancing the predictive performance of plasma T-tau in CU individuals.
same-sample model development lacks validation and shows no significant improvement over T-tau plus the base model
Derived from the full evaluation — not a separate score.
Strengths
Cortical amyloid status was determined by 18F-florbetaben PET using a stated SUVR threshold, allowing associations to be examined separately in CU Aβ− and CU Aβ+ participants.
↳ Methods, Brain Imaging Assessments
T-tau was measured in duplicate with reported LLOQ and CV, while SM analyses used blinded randomisation, plate QC normalisation, covariate adjustment, and Benjamini–Hochberg correction.
↳ Methods, Quantification of plasma T-tau; Quantification of plasma sphingomyelins; Statistical Analyses
The paper identifies its cross-sectional design, modest sample, residual metabolic confounding, and diversity limitations, then calls for independent longitudinal validation, cross-validation, and calibration.
↳ Discussion, limitations paragraph; Conclusion
Limitations
Methods specify Kruskal–Wallis testing for cognitive tertiles, whereas Results state that ordinary one-way ANOVA was used. The discrepancy leaves the inferential basis of the cognitive associations uncertain.
↳ Methods, Statistical Analyses; Results, Cross-sectional relationships across cognitive performance
The SM panel and stepwise logistic model were developed and evaluated in the same N=100 cohort without cross-validation. The added AUC over T-tau plus the base model was not statistically significant at p=0.05638.
↳ Results, Table 4 and Figure 4
The Discussion characterises the associations as likely reflecting SM-mediated protective mechanisms, although the cross-sectional analyses cannot establish protection, compensation, or temporal direction.
↳ Discussion, mechanistic interpretation paragraph
The PET-stratified analysis and adjusted, FDR-corrected models support an exploratory association between selected sphingomyelins and plasma T-tau. Methodological confidence is reduced by the inconsistent Kruskal–Wallis and ANOVA descriptions, absent interaction testing for the Aβ-stratum contrast, and internally evaluated ROC models. The Discussion appropriately identifies major limitations and requests independent validation, but its protective-mechanism and predictive-marker language remains stronger than the cross-sectional evidence. The paper therefore supports hypothesis generation rather than a validated biomarker panel or SM-mediated neuroprotection.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.0
Confidence mediumThe study provides an identifiable but incremental extension by examining SM–T-tau associations within PET-defined amyloid strata and relating selected SMs to cognition, amyloid burden, and classification. Its contribution remains exploratory because the cohort is small and several claims go beyond cross-sectional association.
“Significant positive correlations were observed between SMs and T-tau levels exclusively in CU Aβ+ individuals”
PET stratification, Simoa measurement, targeted metabolomics, covariate adjustment, and FDR correction support the exploratory analyses. Cross-sectional mechanistic inference, unaddressed cohort selection, absent interaction testing, same-sample ROC modelling, and the Kruskal–Wallis/ANOVA discrepancy materially limit execution.
“the current findings most likely reflect circulating SM-mediated protective mechanisms”
The progression from stratified associations through cognition, amyloid burden, and ROC analysis is readily followed. References to prognostic value, predictive markers, and improved diagnostic performance are stronger than the cross-sectional, internally evaluated evidence supports.
“indicating their viability as early predictive markers”
The Discussion engages relevant SM, cognition, tau, and amyloid literature and states specific limitations concerning design, sample size, confounding, and diversity. The explicit validation caveat is substantive, although extended protective-mechanism interpretation is not fully reconciled with those limitations.
“further validation in independent cross-sectional and longitudinal cohorts”
Caveats4 of 4 checks
The statistical test reported for the cognitive-tertile analysis conflicts with the test specified in Methods, and two central interpretations extend beyond the analyses reported. These issues affect the cognitive and classification claims but do not invalidate the primary adjusted SM–T-tau associations.
Ethics approval, consent, lack of preregistration, restricted data access, and relevant industry relationships are disclosed. Consultancy involving the manufacturer of the T-tau assay is a transparent potential conflict rather than an undisclosed conduct problem.
Flags: 3 declared / 5 total
64 references in manuscript 64 of 64 checkable references found in an index
No retraction notice found in Retraction Watch.
Sources: Retraction Watch ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
The work is at an exploratory discovery stage rather than operational validation. ROC models were developed and assessed in the same N=100 cohort, with cross-validation and calibration deferred to future work.
“rigorous cross-validation, calibration and evaluation of multicollinearity, is necessary”
AI-generated, human-governed. Something look off? Contact us to request a review.