Immune "Aging Waves" at ~40 and ~60: Is Vitamin D Metabolism Part of the Story?

Claude AI - Nov 2026

Possibly in a minor, contributing way, mostly at the ~60 wave, but no study has tested it, and the available data argue against vitamin D being a main driver of the ~40 wave. The Telegraph piece reflects a 2026 single-cell study showing transcriptional "peaks" in blood T cells around 40 and after 60.[1] Age-related changes in vitamin D metabolism are real. But they are mostly gradual: lower skin synthesis, reduced renal 1α-hydroxylation as kidney function declines, rising PTH/FGF23, and tissue VDR loss. They do not show a step at 40, and in US data mean serum 25(OH)D actually rises with age because older people take more supplements.[2][3] The best-supported causal link is narrower: correcting true deficiency (<25–50 nmol/L) lowers inflammation (Mendelian randomization) and modestly slows some markers of biological aging (VITAL telomeres).[4] The link from vitamin D to the specific 40/60 waves is still an untested hypothesis.

TL;DR

What the "40 and 60" story is: Park et al., Nature Communications (published 21 August 2026), pooled four public single-cell datasets: 3.8 million PBMCs from 1,828 healthy people aged 19–97. Gene-expression change peaked around 40 (mainly CD4 T cells) and after 60 (mainly CD8 T cells). Alongside the peaks came falling ribosome/translation machinery and rising inflammatory polarization. The design is cross-sectional, and vitamin D was never measured.
Vitamin D metabolism changes, but not in steps at 40/60:
- Skin: cutaneous previtamin D3 capacity was >2-fold lower at 77–82 than at 8–18 years (MacLaughlin & Holick 1985), though a 2024 study found no 7-DHC difference between younger and older adults.
- Kidney: Gallagher's 2013 review states that renal production of 1,25(OH)₂D "is reduced by 50% as a result of an age-related decline in renal function," with serum levels partly maintained by secondary hyperparathyroidism.
- Hormones and receptors: PTH rises, and tissue VDR falls.
- Population levels: NHANES 2011–2014 mean 25(OH)D climbs from 62 nmol/L (age 20–39) to 85 nmol/L (age ≥70).
- Where the fit is best: the second wave (~60), which in the Stanford multi-omics study included kidney-function and immune-regulation molecules.
Causal evidence is modest and mostly about deficiency:
- Non-linear MR shows vitamin D deficiency raises CRP, with the effect levelling off near 50 nmol/L.[5]
- In the VITAL trial, 2,000 IU/day cut new autoimmune disease by 22%. In the VITAL telomere sub-study (Zhu et al., AJCN 2025), it reduced leukocyte telomere attrition by 0.14 kb over 4 years versus placebo, about 0.035 kb per year.
- In DO-HEALTH, vitamin D alone did not slow epigenetic clocks.[6]
- The respiratory-infection benefit (meta-analysis OR 0.94, 0.88–1.00) is now borderline.[7]

Key Findings

1. What the Telegraph article is based on

The Telegraph page is paywalled and could not be retrieved directly. Its framing ("why your immune system ages at 40 and 60") matches the Duke-NUS press release of 1 October 2026 and the underlying paper:[8]

Study: Park H, Le Bert N, Bertoletti A, Tolwinski N, Gruber J, Behmoaras J, et al. "Sex-specific trajectories of nonlinear immune aging at single-cell level." Nat Commun 2026;17:10008.[1]
Design:
- Integrated re-analysis of four published PBMC scRNA-seq datasets, giving ~3.8 million cells from 1,021 women and 807 men aged 19–97 (34.8% Asian).[1]
- Cross-sectional: different people at different ages, not people followed over time.[9]
- Peripheral blood only.[9]
Method: DE-SWAN, a differential-expression sliding-window analysis borrowed from the plasma-proteome "undulating aging" work of Lehallier et al. 2019. It counts how many genes change significantly between adjacent age windows.[1]
Findings:
- A primary peak of differential expression at ~40 years, dominated by CD4 T cells.[1]
- A secondary peak after 60, dominated by CD8 T cells.[1]
- Canonical immunosenescence was confirmed: loss of CD8 naïve T cells and MAIT cells, and more NK cells and CD16 monocytes.[1]
- Coordinated downregulation of rRNA processing, translation and mRNA processing across cell types, alongside upregulated activation/inflammatory pathways ("inflammaging").[1]
- Women had higher peak amplitudes. They also had a female-specific "late increase" cluster with rising ferritin transcripts (FTL, FTH1), which the authors link to menopause-related iron accumulation (NHANES ferritin rises around age ~50).[1]
- Men showed early fluctuation in CD4 T-cell immunometabolism linked to SSH3 hypomethylation at 11q13.[1]
- Sex-specific deep-learning age clocks outperformed combined-sex clocks.[1]
Important caveats from the paper itself:
- No single dataset showed both peaks. Each captured either the first or the second, and both only appeared once datasets were pooled. The authors read this as a sample-size/age-coverage issue, and the peaks did disappear under age permutation.[1] But pooling cohorts that differ in age range, ethnicity and processing is exactly where batch-by-age confounding can create apparent "waves."
- The authors' own explanation for the timing is hormonal: the fourth decade "coincides with declining reproductive capacity and early hormonal changes," and the sixth with perimenopause/menopause and androgen decline.[1] Vitamin D is not discussed.

Related prior work: Shen X, … Snyder MP, Nature Aging 2024;4:1619–1634.[10]

Longitudinal multi-omics in 108 Californians aged 25–75 found that most age-associated molecules changed non-linearly, with crests at ~44 and ~60.[11][12]
The ~44 wave involved lipid, alcohol and caffeine metabolism and cardiovascular markers.[11]
The ~60 wave involved immune regulation, carbohydrate metabolism and kidney function.[13]
Critics note the small sample, with few participants under 40. Secondary press reports say later methodological work questioned whether the exact peak ages partly reflect the analysis method; I could not verify that primary source.[13]

Interpretation: two independent approaches (plasma multi-omics and PBMC transcriptomics) both point to mid-life and early-60s transitions. That convergence is real but still soft evidence. Treat "40" and "60" as broad windows, not birthdays.[13]

2. Age-related changes in vitamin D metabolism

Component	Direction with age	Timing	Key evidence
Skin previtamin D3 synthesis	Down	Gradual; clearest at ≥70	MacLaughlin & Holick 1985: >2-fold lower at 77–82 y vs 8–18 y
Skin 7-DHC (healthy adults)	Possibly unchanged	—	Borecka 2024: 0.22 vs 0.25 µg/mg, ages 18–40 vs 65–89[14]
Serum D3 rise after UVR	Somewhat blunted	≥65 y	Borecka 2024: +107% young vs +67% old (n=21)
Renal CYP27B1 / calcitriol output	Down	Tracks falling eGFR, accelerates after ~60	Gallagher 2013: 50% lower production, ~50% blunted PTH response
CYP24A1 catabolism	Up (renal)	Older age	Christakos et al., Bone Research 2016
PTH	Up	Clear by >70	Vieth 2003: elderly need 25(OH)D >100 nmol/L to match young PTH at ~70[15]
FGF23	Up with CKD	Early CKD onward	Suppresses renal and monocyte CYP27B1
VDR in gut, bone, kidney, parathyroid	Down	Older age	Jiang 2020: parathyroid VDR r=−0.89 with age; PTH r=0.60
VDR / CYP27B1 in PBMCs	No clear change	50–69 vs 70+	Coleman 2016, Oncotarget: n=40 vitamin D-replete adults
DBP; free 25(OH)D	DBP down; free/bioavailable down	Older age	ESTHER cohort; Pop et al.[16][17]
Population 25(OH)D (US)	Up	Rises each decade	NHANES 2011–14: 62→68→72→78→85 nmol/L

Skin:

MacLaughlin and Holick (J Clin Invest 1985) used surgical skin from people aged 8–92. They found an age-dependent fall in epidermal 7-dehydrocholesterol, and older skin (77 and 82 years) made more than two-fold less previtamin D3 than young skin (8 and 18 years).[18]
A modern, controlled comparison (Borecka, Rhodes et al., Nutrients 2024; 11 adults aged 18–40 vs 10 aged 65–89, buttock skin, 1.3 SED UVR to ~35% of body surface) found no difference in skin 7-DHC (0.22 vs 0.25 µg/mg). Serum vitamin D3 rose 107% in younger and 67% in older adults, but age was not a significant determinant.[19]
Net: older skin may lose some synthetic capacity, mainly late in life. There is no evidence of a cliff at 40. The larger real-world factor is behaviour: less outdoor time, more clothing, sunscreen, and institutionalization in the very old.[20]

Kidney and endocrine axis:

In Gallagher's review ("Vitamin D and Aging," Endocrinology and Metabolism Clinics of North America 2013), renal production of 1,25(OH)₂D is reduced by 50% as a result of age-related decline in renal function. When stimulated by PTH, the rise in 1,25(OH)₂D was ~50% lower in older subjects. Serum calcitriol is partly maintained by secondary hyperparathyroidism.
Christakos et al. (Bone Research 2016) describe a combined defect: less renal 1α-hydroxylation, more CYP24A1 catabolism of 1,25(OH)₂D, and lower renal and intestinal VDR.[21]
Vieth et al. (JCEM 2003) found that adults over 70 needed 25(OH)D above 100 nmol/L to reach the PTH levels younger adults had at ~70 nmol/L, and serum 1,25(OH)₂D fell with rising creatinine in every age group.[15]
Jiang et al. (Int J Endocrinol 2020) studied parathyroid glands from 70 thyroidectomy patients aged 10–70. Age correlated inversely with VDR (r = −0.89), CYP27B1 (r = −0.74) and CYP24A1 (r = −0.71), and positively with PTH expression (r = 0.60; p < 0.0001).
FGF23 rises as kidney function falls. It suppresses renal CYP27B1, and it also inhibits extra-renal 1,25(OH)₂D synthesis in human monocytes (Bacchetta et al., JBMR 2013).[22][23] Uremia likewise impairs immune-signal-induced monocyte CYP27B1.[24]
This is the most plausible mechanistic bridge to immune aging, because the antimicrobial vitamin D pathway is intracrine: TLR → CYP27B1 → cathelicidin/LL-37. It also lines up with the kidney-function signal in the Stanford ~60 wave.

Immune-cell VDR and local activation: Coleman et al. (CDC; Oncotarget 2016) studied 40 healthy, vitamin D-replete adults aged 50–69 vs 70+ (mean 25(OH)D 29.7 vs 30.2 ng/mL). Baseline and induced expression of VDR, 1α-hydroxylase and cathelicidin in PBMCs did not differ by age. So far there is no good human evidence for intrinsic immune-cell "vitamin D resistance" with age. The resistance that is documented is intestinal, renal and parathyroid.[25] Monocyte CYP27B1 suppression by FGF23 or uremia is relevant mainly to people with CKD.

DBP and free 25(OH)D: the German ESTHER cohort (Saarland, ~5,900 adults aged 50–75; Nutrients 2021) had a mean total 25(OH)D of 49.8 nmol/L. Bioavailable and free 25(OH)D fell with increasing age and BMI, and VDBP itself showed "quite pronounced inverse associations" with age and BMI. Calculated free-25(OH)D equations are imprecise in older people, so these findings are hypothesis-level.[26]

Sex hormones:

Estrogen up-regulates VDR. In ovariectomized rats, estrogen doubled duodenal VDR mRNA and increased VDR bioresponse. Menopause is associated with an apparent intestinal resistance to vitamin D that estrogen replacement reverses.[27]
In mouse EAE, estrogen was required for vitamin D3's protective effect, acting through vitamin D metabolism and VDR expression.[28][29]
Reviews describe two-way crosstalk: estrogen boosts VDR expression, and 1,25(OH)₂D down-regulates immune-cell aromatase.[28]
This is the clearest mechanistic route by which a vitamin D change could be female-specific and stepwise around the menopausal transition (late 40s to early 50s). That timing sits between the two Park peaks and overlaps their female-specific late-increase cluster. It is entirely untested in human immune cells.

Population 25(OH)D by age — no step down:

In NHANES (LC-MS/MS, 2011–2014), mean 25(OH)D was 62.2 nmol/L at 20–39, 67.6 at 40–49, 72.2 at 50–59, 78.4 at 60–69 and 85.1 at ≥70. In 2007–2010 the gradient was much flatter (64.6 → 71.4).[3] The steepening was driven by supplement use in older adults.[2]
Herrick et al. (AJCN 2019) found that the risk of deficiency (<30 nmol/L) fell with age: 7.6% at 20–39, 5.7% at 40–59 and 2.9% at ≥60.[30]
So in a supplementing population, the people who show the immune "waves" are, on average, not more vitamin D-deficient than younger adults. European cohorts without heavy supplement use (e.g., ESTHER) look worse. That makes the vitamin D contribution population-dependent, which is itself a testable prediction.

3. Vitamin D and markers of immunosenescence / inflammaging

Outcome	Study	Design and size	Finding
Thymic aging	Cyp27b1-KO mice, Science Advances 2024	Animal	Premature thymic involution, reduced Aire/TRA, skewed mTEC differentiation[31][32]
Thymic output	22q11.2 deletion cohort	Observational	25(OH)D correlated with recent thymic emigrants[33]
CRP (inflammaging)	Zhou & Hyppönen, IJE 2023	Non-linear MR, n=294,970	L-shaped: CRP falls steeply <25 nmol/L, levels off ~50 nmol/L; no reverse causation[5][34]
Leukocyte telomeres	Zhu et al., AJCN 2025 (VITAL)	RCT, n≈1,031, 2,000 IU/d, 4 y	0.14 kb less attrition (~0.035 kb/y); ~3 years of aging
Epigenetic age	Chen et al., J Gerontol A 2019	RCT, n=51, mean age 26, 16 wk	Horvath age slowed (~1.85 y at 4,000 IU); Hannum inconclusive
Epigenetic age	Bischoff-Ferrari, Nat Aging 2025 (DO-HEALTH)	RCT, n=777, ≥70 y, 3 y	Vitamin D alone null; omega-3 slowed 3 clocks; additive on PhenoAge[35][36]
Autoimmune disease	Hahn et al., BMJ 2022 (VITAL)	RCT, n=25,871, 5.3 y	HR 0.78 (0.61–0.99); ~39% lower after excluding first 2 years[37]
Respiratory infection	Jolliffe et al. 2021	Meta-analysis, 43 RCTs, n=48,488	OR 0.92 (0.86–0.99); best with daily 400–1,000 IU[38][39]
Respiratory infection	Jolliffe et al. 2025	Meta-analysis, 46 RCTs, n=64,086	OR 0.94 (0.88–1.00), p=0.057; no effect modification by age[39][40]
Influenza vaccine	Goncalves-Mendes, Front Immunol 2019	RCT, deficient >65 y	No antibody gain; higher TGF-β, more tolerogenic profile[41]
Skin VZV recall	Chambers et al., Immunother Adv 2021	Uncontrolled, n=18, 6,400 IU/d, 14 wk	Stronger antigen-specific response; less monocyte inflammation

How to read this:

Thymus: the mechanistic case is strongest here. Vitamin D signaling supports thymic epithelial differentiation and delays involution in mice.[42] But human thymic output is concentrated in the first two decades, and no human trial has measured TRECs or recent thymic emigrants after supplementation in mid-life.[31] One small 2013 observational report even linked higher 25(OH)D to fewer naïve CD8 T cells in early aging, a reminder that the human data are thin and inconsistent.[43]
Inflammaging: MR supports causation only within deficiency. That fits a threshold model: vitamin D matters for immune aging in people below ~50 nmol/L, not as a general lever.[44] Non-linear MR methods have since been scrutinized, with a 2025 methods paper and a doubly-ranked reanalysis, so even this is not bulletproof.[45][46]
Biological-age markers:
- VITAL telomeres is the strongest trial signal: large, randomized, in people over 50, though measured by qPCR on leukocytes.[47]
- Chen 2019 is small, short and done in young adults.[48][49]
- DO-HEALTH, in replete Swiss adults over 70, found no effect of vitamin D alone on second-generation clocks.[6] This fits the theme that benefit is concentrated in people who are deficient.
Clinical immunity: VITAL's autoimmune result (women ≥55, men ≥50) is the most relevant RCT outcome for the "60" wave,[50] especially given that women predominate in the Park study's late-life signatures.[37] The infection benefit has shrunk to borderline.[39] Vaccine antibody responses in older adults do not improve, although the uncontrolled VZV study hints that vitamin D can damp the monocyte-driven non-specific inflammation that blunts recall responses in old skin.[41][51][52][53]

Assessment: cause, contributor, or coincidence?

Question	Verdict	Reasoning
Does vitamin D status drop at ~40?	No	NHANES 25(OH)D rises across 20–39 → 40–49; deficiency prevalence falls
Could vitamin D drive the CD4 wave at ~40?	Unlikely as main driver	No step change in substrate or immune-cell VDR; hormonal/metabolic shifts fit better
Could vitamin D contribute to the ~60 wave?	Plausible, partial	eGFR decline, PTH/FGF23 rise, lower calcitriol and VDR; Stanford ~60 wave includes kidney function
Female-specific midlife changes?	Plausible hypothesis	Estrogen loss reduces VDR expression/response in animal and gut data
Is there causal human evidence?	Limited	MR for CRP in deficiency; RCTs for telomeres and autoimmunity; nothing on the waves themselves
Has anyone tested vitamin D alongside the waves?	No	Neither Park 2026 nor Shen 2024 analyzed 25(OH)D or vitamin D-pathway genes as drivers

Confounding cuts both ways:

Inflammation and acute illness lower 25(OH)D (DBP and albumin behave as negative acute-phase reactants), so observational links can be reverse-causal. The Zhou–Hyppönen MR, however, found no effect of CRP on 25(OH)D.[5]
Obesity, which rises in midlife, both lowers circulating 25(OH)D (sequestration and volumetric dilution) and independently drives inflammaging.
Supplement users are healthier.
Medications that become common after 40–60 affect vitamin D metabolism: glucocorticoids, anticonvulsants, some antiretrovirals, and loop or thiazide diuretics via calcium handling.

The bottom line on mechanism: vitamin D sits downstream of, and interacts with, the same systemic drivers the authors propose (sex hormones, metabolic and renal decline). It is better modeled as a modifier of immune aging trajectories in deficient people than as the clock that sets them.

Recommendations

For a VitaminDWiki reader:

Keep 25(OH)D out of the deficiency zone, especially after ~60 and in women after menopause. The MR threshold for inflammation (~50 nmol/L), and the PTH data showing older adults need higher 25(OH)D to suppress PTH, support targeting at least 50 and plausibly 75 nmol/L or more after 60.[5][15]
Prefer daily dosing. Bolus dosing performed worse in the respiratory-infection meta-analyses, and VITAL and DO-HEALTH both used 2,000 IU/day.[39][54][55]
Check eGFR (and PTH if eGFR is reduced) after 60. That is where the age-related vitamin D defect is concentrated. In CKD, 25(OH)D repletion alone may not normalize calcitriol or monocyte CYP27B1 activity.[56]
Combine with omega-3. In DO-HEALTH, omega-3 was the active agent on epigenetic clocks and combined additively with vitamin D on PhenoAge.[57]
Brief co-factor notes:
- Magnesium is required for the 25- and 1α-hydroxylases and for DBP function, and is commonly low in older adults.
- Zinc supports thymic function; high-dose zinc regenerated thymic output after stem-cell transplant, which is unproven in normal aging.[32]

For research (a clear gap worth flagging on VitaminDWiki):

Several of the datasets Park et al. pooled come from cohorts with stored serum. Re-analyzing CD4/CD8 trajectories stratified by 25(OH)D (and eGFR, menopause status and BMI) would directly test whether vitamin D status modifies the ~40 and ~60 peaks.
VITAL and DO-HEALTH biobanks could test whether randomized vitamin D shifts single-cell immune age clocks.

Caveats

The Telegraph text itself could not be accessed; its content is inferred from the Duke-NUS release, the paper, and secondary coverage, which all say the same thing.
Both "wave" studies are hypothesis-generating:
- Park 2026 is cross-sectional and pooled, and no single dataset showed both peaks.[1]
- Shen 2024 is small (n = 108) and thin under age 40.[13]
Most mechanistic vitamin D–immune-aging data are animal-based (Cyp27b1-KO thymus, ovariectomized rats, EAE mice) or come from small human studies.
Human PBMC data (n = 40) do not show age-related VDR loss.[58]
Secondary claims about MacLaughlin & Holick, such as "70-year-olds make 50% or 75% less," vary between sources.[59][60] The primary abstract supports only ">2-fold lower previtamin D3 at 77–82 vs 8–18 years."[18]
US NHANES gradients reflect supplement use. Unsupplemented European or Asian populations may show falling 25(OH)D with age, so conclusions about "no step at 40/60" are population-specific.
Trial effects (telomeres, autoimmunity) are modest. The autoimmune result had p = 0.05,[61] and the infection benefit is now statistically borderline.