Published on 25 September 2026
Circulating lipids and Alzheimer’s disease biomarkers: what does the ADNI cohort show?
The essentials
This ADNI cohort study measured 749 plasma lipid species (46 classes) by liquid chromatography-mass spectrometry in 1,395 participants, then analysed their associations with amyloid, tau and neurodegeneration biomarkers (the ATN system), at baseline (cross-sectional analyses) and during follow-up (longitudinal analyses). At baseline, LPC(O) species are associated with more severe amyloid and neurodegeneration biomarkers, ether PE species with more favourable biomarkers (amyloid, tau and neurodegeneration at species level), and a GM3 ganglioside with more marked neurodegeneration, including during follow-up. Of 57 species associated with less severe biomarkers at baseline, 20 contain docosahexaenoic acid (DHA). No lipid species is associated with the course of amyloid or tau: the longitudinal associations concern neurodegeneration only. These are observational associations, without independent replication, and without demonstration of causality.
Context
Alzheimer’s disease is characterised by changes in amyloid, tau and neurodegeneration biomarkers, grouped under the acronym ATN. The authors note that the involvement of lipids in Alzheimer’s disease has been suggested by many earlier studies, notably on the blood lipid profile. DHA is an omega-3 fatty acid for which they cite protective effects reported in the literature.
This study does not test a treatment. In the ADNI cohort, it looks for which circulating lipid species are associated with ATN biomarkers, at baseline and in their course. According to the authors, earlier studies of ATN biomarkers used techniques of more limited resolution and dealt mostly with cross-sectional associations.
The mechanism
The study rests on a hypothesis: plasma lipids are linked to brain processes of the disease. Lipids were measured once, at baseline, then related to the biomarkers at baseline and to their change during follow-up. The study did not measure the effect of DHA intake, nor establish the direction of the relationship between lipids and biomarkers.
The mechanisms suggested by the authors (phospholipase A2 activation for LPC(O), antioxidant role of plasmalogens, peroxisome function) belong to the discussion and rely on the literature; they were not tested in this study.
The study at a glance
| Item | |
|---|---|
| Population | |
| ADNI cohort participants (ADNI 1, GO, 2 and 3) with lipidomic and ATN biomarker data at baseline: 1,395 participants (mean age 73.6 years, 55.1% men, 47.0% carriers of the APOE ε4 allele). The breakdown by diagnostic stage is not reported in the text consulted. | |
| Exposure | |
| Targeted plasma lipid profile at baseline, 749 species from 46 classes, liquid chromatography-mass spectrometry; analyses at the level of species, classes and 46 modules of correlated lipids | |
| Comparator | |
| None: association study | |
| Outcomes | |
| Amyloid (amyloid PET), tau (cerebrospinal fluid p-tau), neurodegeneration (hippocampal volume on MRI and temporal uptake on FDG PET), at baseline and over time. Numbers per biomarker at baseline: 743 (amyloid PET), 1,013 (CSF), 1,387 (MRI), 1,060 (FDG PET) | |
| Design | |
| Observational study in the ADNI cohort; linear regression (cross-sectional analyses) and linear mixed-effects models (longitudinal analyses); false discovery rate correction (Benjamini-Hochberg); no independent replication | |
Quality check
| Item | Verdict |
|---|---|
| Sample size and cohort | Large sample |
| Finding1,395 participants, but numbers vary by biomarker (from 743 to 1,387 at baseline) and longitudinal follow-up is smaller for amyloid and CSF tau. | |
| Lipid coverage | Broad |
| Finding749 species from 46 classes, using a targeted assay method. | |
| Multiplicity | Corrected |
| FindingFalse discovery rate correction (Benjamini-Hochberg) stated for all p values; the p values reported for the omega-3 composites are presented without explicit indication of correction. | |
| Representativeness | Not documented |
| FindingThe ethnic or geographic characteristics of the participants are not reported in the text consulted; generalisability is not discussed by the authors. | |
| Funding and conflicts of interest | Worth reading |
| FindingFunding: ADNI (NIH U01 AG024904, US Department of Defense, contributions from many pharmaceutical and diagnostic companies via the FNIH), ADMC Consortium, the Korean KDCA agency, NIA grants. Declared conflicts of interest: one author (Kaddurah-Daouk) is an inventor of patents on metabolomics for the diagnosis and treatment of central nervous system diseases and holds shares in Metabolon, Chymia and PsyProtix. | |
Results
| Item | Value |
|---|---|
| LPC(O) species | Amyloid and neurodegeneration, higher levels = more severe |
| FindingAt baseline, associated with A and N biomarkers at species, class and module levels: higher levels go with more amyloid and lower glucose metabolism. During follow-up, associated with change in neurodegeneration (more LPC(O), faster progression). | |
| Ether PE species | Amyloid, tau, neurodegeneration at baseline, mostly more favourable |
| FindingAt species level, associated with all three biomarkers A, T and N at baseline, most with less severe values; at class and module levels, associated with N only. No significant association with the course of A, T or N at class or module levels. | |
| GM3 ganglioside | Neurodegeneration, higher = more severe |
| FindingAssociated with the N biomarker at baseline and during follow-up, at species and class levels: higher levels go with faster neurodegeneration. | |
| DHA-containing species | 20 of 57 |
| FindingAmong the 57 species associated with less severe biomarkers at baseline (cross-sectional analysis). At the level of the DHA-richest module (M45), the associations with p-tau and hippocampal volume are marginal (adjusted p 0.053 and 0.054), above the 0.05 threshold. | |
| Omega-3 surrogate index (plasma) | Lower tau at baseline |
| FindingAssociated with lower p-tau at baseline (β = -0.120, 95% CI -0.183 to -0.057, p < 0.001), with no association with amyloid, hippocampal volume or FDG PET, and none with the course (all p > 0.1). This index is calculated from plasma phospholipids, not from erythrocytes. | |
| Omega-3 plasmalogens (composite score) | Lower amyloid and tau, larger hippocampus at baseline |
| FindingAmyloid β = -0.081 (p = 0.04), tau β = -0.113 (p = 0.001), hippocampal volume β = 0.086 (p = 0.001) at baseline; no association with the course (all p > 0.1). Supplementary analysis, with no indication of correction for multiplicity. | |
Critical appraisal
| Item | Verdict |
|---|---|
| Sample size and longitudinal data | Solid for N only |
| FindingLarge sample and ATN biomarkers followed over time, but no significant longitudinal association for amyloid and tau: only neurodegeneration varies with lipids. | |
| Broad lipid coverage | Solid |
| Finding749 species, false discovery rate correction. | |
| Associative nature | Needs qualifying |
| FindingThe authors describe their results as observational. An association between a lipid and a biomarker does not say which comes first. | |
| Replication | Absent |
| FindingThe authors point out the lack of large independent data with which to replicate the results. | |
| Clinical applicability | Not assessed |
| FindingThe authors propose lipids as potential blood biomarkers, diagnostic and prognostic, but no diagnostic or prognostic performance is assessed. Assay of 749 species is not available in routine clinical practice. | |
Level of evidence
The level of evidence is that of an observational cohort study (Oxford CEBM level 2b). Confidence is good for sample size and breadth of lipid assay, but no independent replication was carried out.
It is low for any causal or therapeutic reading: these are observational associations with lipids measured once, without replication and without a tested intervention.
The colleague test
What an experienced colleague would say if you presented this study in two minutes, between two consultations.
“Nothing to change on Monday morning: this 749-species assay is not available in routine care, and these are associations. But the lead on ether lipids and DHA deserves to be followed.”
Translation for practice: note that a lipid signal exists, and prescribe nothing on that basis.
What you can do with this
- What you can understand: plasma lipids are associated with the course of Alzheimer’s disease biomarkers, without the direction of the relationship being established.
- What you can say to a patient asking about omega-3: this study observes associations between DHA-containing lipids and biomarkers; it does not test taking a supplement and does not allow a conclusion of benefit.
- What you can watch for: trials testing DHA intake or ether lipids on clinical outcomes, and replication of these results in other cohorts.
- What you can teach: the difference between a biomarker associated with a disease and a biomarker useful for making decisions.
Frequently asked questions
Do omega-3 fatty acids protect against Alzheimer’s disease?
This study cannot say. It notes that 20 of the 57 lipid species associated with less severe biomarkers at baseline contain DHA, but these are cross-sectional associations, not a supplementation trial. The authors see support for the beneficial effects of DHA already reported, while describing their results as observational.
Can these lipids be measured in practice?
The study does not assess this use. Assay of 749 species by chromatography coupled with mass spectrometry is not available in routine care.
What does the acronym ATN stand for?
It groups three families of biomarkers: amyloid (A), tau (T) and neurodegeneration (N).
Have these results been confirmed in another cohort?
No. The authors point out the lack of large independent data for replication and call for it to be reproduced in other cohorts with lipidomics and longitudinal biomarkers.
Annotated bibliography
Source study. Kim JP, Nho K, Wang T, Huynh K, Arnold M, Risacher SL, Bice PJ, Han X, Kristal BS, Blach C, Baillie R, Kastenmüller G, Meikle PJ, Saykin AJ, Kaddurah-Daouk R, for the Alzheimer’s Disease Neuroimaging Initiative and the Alzheimer’s Disease Metabolomics Consortium (ADMC). Circulating lipids are related to longitudinal changes of ATN biomarkers for Alzheimer’s disease. Molecular Psychiatry. 2026;31(9):5395-5408. Published online 16 June 2026. DOI: https://doi.org/10.1038/s41380-026-03626-z. PMID: 42304066. Funding: ADNI (NIH U01 AG024904; DOD ADNI W81XWH-12-2-0012; contributions from private companies via the FNIH), Alzheimer’s Disease Metabolomics Consortium (NIA and FNIH grants), Korea National Institute of Health (KDCA, 2023-ER1001-03), other NIA grants. Declared conflicts of interest: R. Kaddurah-Daouk is an inventor of patents on the use of metabolomics for the diagnosis and treatment of central nervous system diseases and holds shares in Metabolon Inc., Chymia LLC and PsyProtix; no other conflict is declared in the text consulted. Supplementary material: supplementary tables 1 to 10 and the code were consulted as text; the figures were not examined, and no data that appears only there is used here.
Context. Huynh K, Lim WLF, Giles C, et al. Concordant peripheral lipidome signatures in two large clinical studies of Alzheimer’s disease. Nat Commun. 2020;11:5698. Reference 16 of the source publication: a study of two large cohorts (AIBL and ADNI) that associated 218 lipid species with prevalent or incident Alzheimer’s disease, including ether lipids that the source publication sets against its own results. DOI not given in the source’s reference list.
Editorial collections
Tags
Verified on 25 September 2026 against the full text of the publication and its supplementary material where available. This analysis underwent an independent double reading. The English version was checked for conformity on 25 September 2026, against the figures of the French version and against the source. How we verify what we publish
This analysis is intended for healthcare professionals. It does not constitute a prescribing recommendation and does not replace individual clinical judgment.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigor.