The ACM vs ferritin is U-shaped and below 30 the HR are higher.
The short summary is: For longevity, avoid <30 (HR 1.3–1.9); >200 (HR 1.1), >600 (HR 1.5)
The optimal band would be ~50–120
For those who want the details and associated evidence, here is my question to Fable 5.1 (High)
There is a discussion about ferritin in a longevity forum and most people there have a very low ferritin with some as low as 22. What is the best range to target and are they different for longevity vs performance?
I’ll start by pulling the project context, then the literature. A few targeted searches so the thresholds I give are anchored to primary sources rather than forum lore.Short answer: the forum has the direction half-right and the number wrong. The longevity evidence argues against high-normal iron, not for depletion; a ferritin of 22 in a non-anemic adult sits in the band with the clearest excess mortality in general-population cohorts. For you, the two goals converge on roughly ferritin 70–120 ng/mL with TSAT 25–35% (hs-CRP <1 so the ferritin is real), which is essentially the exit criterion already in the file. Nothing in this analysis changes the current plan.
What the “low ferritin” thesis actually rests on
FeAST (the only RCT). 1,277 PAD patients, phlebotomy targeting ferritin 25–60. Over a follow-up that averaged 3.5 years, all-cause mortality—the primary end point—was 20% for the 636 iron-reduction patients and 23% for the 641 controls — non-significant, as was the composite of death/MI/stroke. What the forum quotes is the secondary analyses: the main study demonstrated clinical benefits in reduction of death from primary and secondary end points for participants aged between 43 and 61 years, and the cancer substudy (reduced incidence of new cancer diagnoses, HR = 0.65; 95% CI = 0.43–0.97). Two things the forum usually omits: the treated arm only reached a mean achieved ferritin level of 79.7 ng/mL, down from a mean of 121.8 ng/mL — so the trial is evidence about ~80, not 25 — and the subgroup findings are post-hoc in a 98% male, 67-year-old, smoking-heavy VA cohort. Grade B for “≈80 is not harmful”; grade C for anything lower.
Mendelian randomization. Daghlas & Gill 2021 is the strongest causal evidence and it does point downward: the association of a 1-SD increase in genetically predicted iron status biomarker with lifespan years was −0.70 for iron, −1.64 for ferritin, −0.54 for transferrin saturation. But the authors are explicit that this MR approach only considers the linear associations of small changes in genetically predicted iron status around the population mean, and cannot be extrapolated to infer the effect of changes in iron status outside of this normal range, and that lifelong genetic exposure cannot be extrapolated to predict the effect of a discrete clinical intervention that modifies iron status. The instruments were three independent missense variants in genes implicated in iron homeostasis (rs1800562 – HFE, rs1799945 – HFE, rs855791 – TMPRSS6) — i.e. H63D and the TMPRSS6 SNP you carry. The correct reading: a lifelong lower iron set-point is mildly favorable; it says nothing about phlebotomizing a 63-year-old to 22.
Observational upper bound. Copenhagen City Heart Study: HR for total overall mortality for ferritin ≥600 vs <200 μg/L was 1.5, and only 1.1 for ≥200 vs <200; the pooled metaanalysis random effects odds ratio for total mortality for ferritin upper vs reference quartile or tertile was 1.0. The harm signal lives at ≥200–400 and above; below 200 there is essentially none.
The floor the forum is ignoring
Gutenberg (n=5,000, non-anemic general population, adjusted for Hb and CV risk factors): the hazard ratio of the association of iron deficiency with mortality was 1.3 (1.0-1.6) for the functional definition, and 1.9 (1.3-2.8) for absolute iron deficiency, where absolute = ferritin <30 and functional = ferritin levels below 100 µg/l, or ferritin levels between 100 and 299 µg/l and transferrin saturation below 20%. Three-cohort replication (n=12,164): severe AID was associated with all-cause mortality (HR 1.28, 95% CI 1.12-1.46), and absolute ID with incident CHD (HR 1.20). NHANES (18.8-year follow-up) is the cleanest picture of the two axes: a significant linear association between serum ferritin and all-cause mortality was observed among the overall population, and J-shaped associations between transferrin saturation (TSAT) and all-cause and CVD mortality were observed, with the lowest TSAT quartile (<17.1%) faring worst. Grade B (observational; reverse causation is real — in a man over 50, unexplained ferritin of 22 is a work-up for occult GI loss, not a biomarker to be proud of).
So the population-level trough is ferritin ~50–150 with TSAT ~20–35%. A forum member at 22 is on the wrong side of it.
Performance
The athlete literature sets a higher floor. Meta-analysis of IDNA endurance athletes (Burden 2015) shows VO2max gains from repletion; the 2025 female-athlete review (ferritin <40 as the ID criterion) found iron deficiency negatively affects endurance performance by 3%-4%. However, endurance performance improved by 2%-20% when ID athletes were treated. The ceiling for benefit is also visible: in Garvican-Lewis’s trained runners, haemoglobin mass increased in the iron deficiency group with intravenous treatment accompanied by an increase in VO2max and run time to exhaustion, whereas the group with ferritin <65 mcg/l did not show any changes, and the Swiss consensus concludes that a long-term daily oral iron intake or i.v. supplementation in the presence of normal or even high ferritin values does not make sense and may be harmful. Pushing to 100 in young athletes yielded no differences in performance between the groups. Grade A for treating <30–35; grade B for a 40–50 floor; no evidence of performance gain above ~65.
|
Ferritin (hs-CRP <1) |
TSAT |
Basis |
| Longevity, avoid |
<30 (HR 1.3–1.9); >200 (HR 1.1), >600 (HR 1.5) |
<17–20% |
Gutenberg, Schrage 2021, NHANES, Ellervik |
| Longevity, tolerated |
~40–150 |
20–35% |
trough of the cohorts; FeAST ≈80 |
| Performance floor |
≥40–50 |
≥20% |
Burden 2015, Garvican-Lewis, Swiss consensus |
| Performance ceiling |
no gain >~65–100 |
— |
Garvican-Lewis, basketball RCT |
| Intersection |
~50–120 |
25–35% |
|
Where you sit
Your series (all Labcorp): ferritin 200–254 in 2021–23 with hs-CRP 1.7–3.1 (so ~170–210 corrected), then 147 → 131 → 86 → 77 → 80 → 59 → 53; TSAT 35% → 30% → ~32% → ~29% → 21% → 18%; Hb 15.4 → 13.6. You are now in Gutenberg’s functional-ID band with TSAT <20%, and unlike the forum members you have documented functional cost — the FT3/FT4 ratio decline, ~0.1–0.3 of HbA1c inflation, a falling Hb despite an erythropoietic push — plus a CAD history, where the HF/CHD trials all treat ferritin <100 (or <300 with TSAT <20%) as the target population. The existing exit criterion (TSAT >30%, ferritin >100) is correctly placed, and the 2021–23 plateau was, if anything, slightly high by Ellervik’s data.
Three genotype-specific points. First, H63D is one of the iron-raising MR alleles, so once repleted your absorption may run warm — the exit criterion means “stop supplementing and surveil”, not “maintain”. Second, TMPRSS6 rs855791 was corrected to the iron-sparing direction, which removes the one genetic reason to expect chronic under-absorption. Third, ferritin is a poor primary marker in you specifically (empagliflozin lowers hepcidin and drives ferritin down independently of stores), so the December readout should be read on TSAT and Ret-He, with ferritin as a secondary — exactly as the handover already specifies.
Extrapolation flag: no trial has tested iron repletion on hard outcomes in a non-anemic endurance athlete with CAD; the HF IV-iron trials (AFFIRM-AHF, IRONMAN, HEART-FID) are the nearest population, and they show reduced HF hospitalization, not mortality. That is the honest limit of the evidence behind “50–120”.
