Higher midlife vitamin D linked to less tau in the brain 16 years later

Higher midlife vitamin D linked to less tau in the brain 16 years later – April 2026

A Framingham cohort found that people with higher vitamin D in their late 30s had lower tau-PET burden in midlife — but not lower amyloid. Association only, single measurement, vitamin-D-replete cohort. Promising, not proven.

Quick summary

In ~800 dementia-free adults, higher serum 25(OH)D measured around age 39 was associated with lower tau-PET burden ~16 years later, both globally and in the brain regions where Alzheimer's pathology starts.
No association was seen with amyloid-beta.
It is an observational association — it does not prove vitamin D lowers tau or prevents dementia.
Vitamin D was measured only once, and the cohort was relatively vitamin-D-replete (mean 38 ng/mL), which shapes how far the finding can be pushed.
The authors frame midlife as the modifiable window and call for trials of supplementation in younger adults — a framing VitaminDWiki has made for years.

The study at a glance


Title	Association of Circulating Vitamin D in Midlife With ~~Increased~~ Tau-PET Burden in Dementia-Free Adults (see note on title below)
Authors	Mulligan MD, Scott MR, Yang Q, Wang R, Ghosh S, Johnson KA, Beiser AS, Seshadri S, McGrath ER
Journal	Neurology: Open Access (American Academy of Neurology)
Published	April 1, 2026
DOI	10.1212/WN9.0000000000000057
Cohort	Framingham Heart Study, Generation 3
Design	Prospective community-based cohort; single baseline 25(OH)D, later brain-PET
Institutions	University of Galway + Boston University + UT Health San Antonio
Funding	NIA, NINDS, Irish Research Council, Health Research Board of Ireland (Framingham core by NHLBI)

Who was in it

793 dementia-free participants; mean age 39.2; 53.8% female.
Serum 25(OH)D measured 2002–2005 (in a subset of 435).
Brain-PET 2016–2019, an average of 16 years later.
Tau imaging available for 369; amyloid imaging for 424.
Baseline mean 25(OH)D 38 ± 15 ng/mL; about 34% were below 30 ng/mL; only 5% reported taking vitamin D.

How they measured the brain

Tau via ¹⁸F-flortaucipir (FTP) PET.
Amyloid via ¹¹C-PiB PET.
Tau was assessed both as a global cortical measure and in early-AD-vulnerable regions: entorhinal cortex, parahippocampal gyrus, fusiform gyrus, amygdala, and temporal cortices.
Models adjusted for age, sex, and depressive symptoms.

What it found

Higher midlife 25(OH)D was associated with lower tau burden later — globally and regionally. The relationship held when vitamin D was treated as a continuous variable, not just split into high/low. There was no association with amyloid-beta.

The tau-but-not-amyloid pattern — why it may matter

This is the most interesting wrinkle and the part worth highlighting on the page. The authors argue the tau-only signal in a young cohort is consistent with the temporal sequence of Alzheimer's pathology: medial-temporal (entorhinal) tau begins accumulating before widespread cortical amyloid. In a cohort imaged in their mid-50s, tau changes in these earliest regions are simply the thing you can still see moving — amyloid divergence may come later, or require an older sample to detect. In other words, the null on amyloid is at least as easily explained by timing as by absence of effect.

This dovetails with VitaminDWiki's longstanding interest in preclinical, modifiable windows rather than late-stage intervention.

What this does NOT show (read this before citing it as proof)

Listing the limits openly is what makes the page trustworthy. This study does not establish that:

Vitamin D lowers tau. It is an association. Reverse causation and residual confounding are not excluded.
Vitamin D prevents dementia. Tau-PET is a preclinical biomarker, not a diagnosis. No one in this analysis was followed to a dementia endpoint.
Supplementation helps. Only 5% supplemented; the signal is from circulating 25(OH)D, which tracks sun, diet, adiposity, and lifestyle — not a pill.

Evidence gaps to flag: single measurement of vitamin D replete group

Single measurement. One 25(OH)D draw at baseline — no trajectory, no accounting for how levels changed over the 16 years. The authors name this as the headline limitation.
A vitamin-D-replete cohort. Mean 38 ng/mL is high; only about a third were under 30. That limits inference at the low end where the deficiency burden actually sits — though it arguably makes the signal more striking, since it appears even across a fairly sufficient range.
Generalizability. Framingham Gen 3 is overwhelmingly white and of European descent. Directly relevant to VitaminDWiki's outreach priorities — this tells us little about populations with higher deficiency prevalence (e.g., the communities targeted through IHS/FDPIR work).
No dose-response threshold reported in the lay coverage. Worth pulling the actual betas/CIs from the full text before quoting an effect size on the page.
Amyloid null is ambiguous. Could be a true absence of effect, a power issue, or a timing issue (see above). Don't let a reader conclude "vitamin D doesn't touch amyloid."

Why this matters

The practical argument the authors make is the one VitaminDWiki has made for years, now with a hard imaging biomarker behind it: the leverage is in midlife, not in the clinic at age 75. A modifiable exposure measured around age 39 tracking with a core Alzheimer's protein at age 55 is exactly the kind of "intervene early and cheaply" story that fits a prevention-economics framing. The honest version is that it raises the priority of a randomized supplementation trial in younger adults — which is precisely what senior author McGrath calls for — rather than settling anything.

Reference

Mulligan MD, Scott MR, Yang Q, Wang R, Ghosh S, Johnson KA, Beiser AS, Seshadri S, McGrath ER. Association of Circulating Vitamin D in Midlife With Increased Tau-PET Burden in Dementia-Free Adults. Neurology: Open Access. April 1, 2026. doi:10.1212/WN9.0000000000000057 https://www.neurology.org/doi/10.1212/WN9.0000000000000057

PDF

By Claude AI May 2026

YouTube Review of the study

13 minutes