Vitamin A

What Vitamin A actually measures — the analyte, and the assay

The analyte is serum retinol — all-trans retinol, not ‘vitamin A’ in the dietary sense, and not the retinyl esters that carry it in food or store it in the liver. It circulates in a 1:1 complex with retinol-binding protein 4, which in turn circulates bound to transthyretin. That three-protein arrangement is not decoration: it is why anything that lowers hepatic protein synthesis lowers your retinol without your vitamin A stores changing.

The molecular identities are worth carrying, because they explain every interference below. Retinol is a 286 Da isoprenoid alcohol. RBP4 is a 21 kDa lipocalin. Transthyretin circulates as a 55 kDa tetramer, and the assembled complex is large enough to escape glomerular filtration — which is why renal clearance of retinol is negligible in health, and why RBP4 accumulates in kidney disease.

Retinol is also not the active molecule. It is oxidized by retinol dehydrogenase to retinaldehyde, then irreversibly by retinaldehyde dehydrogenase to retinoic acid — the ligand for the RAR and RXR nuclear receptors that drive transcription. The assay measures a precursor two enzymatic steps upstream of the molecule that does the work, and roughly 90% of your body's vitamin A is not in the sample at all: it sits in hepatic stellate cells as retinyl palmitate.

Dietary retinyl esters are hydrolyzed by pancreatic lipase and an intestinal retinyl ester hydrolase before absorption, and cellular uptake of the retinol–RBP4 complex is receptor-mediated through STRA6.

Measurement is by HPLC or LC-MS/MS. Retinol is light- and heat-labile, so tube handling matters more than for most analytes. Some programs measure RBP instead, because it is cheaper and can be run on an immunoassay platform — but the substitution is not fully calibrated: a controlled human norovirus challenge study concluded that the World Health Organization's recommendations on an RBP cut-off need updating Victor 2023.

The governing fact about this test, though, is physiological rather than analytical. The liver holds the great majority of body vitamin A as retinyl esters and releases retinol to hold serum roughly constant across a wide range of stores. Serum retinol is therefore homeostatically defended: it falls only when liver stores are substantially depleted, and it rises only at frank excess. Across the whole middle of the range — which is where nearly every reader sits — it is deliberately insensitive to status Zheng 2023.

Vitamin A: what changes the blood, and what only changes the reading

What changes the vitamin in your body:

1. Liver stores, but only at the extremes, for the reason above.
2. Fat malabsorption — cholestasis, pancreatic insufficiency, cystic fibrosis, bariatric surgery. Vitamin A is fat-soluble and follows the fat.
3. Zinc status. Zinc is required for hepatic RBP synthesis, so a genuine zinc deficiency can produce a low retinol with adequate vitamin A stores. This is one of the most useful cross-links on the site and it runs in the direction people do not expect.
4. Protein-energy malnutrition, which cuts the RBP and transthyretin available to carry it.
5. Over-supplementation and isotretinoin, which is itself a retinoid and changes the whole picture.

What changes only the reading — and the first term here is larger than most of the diet:

1. Inflammation. Retinol and RBP are negative acute-phase reactants: they fall during an inflammatory response regardless of stores. This has been measured under controlled conditions rather than inferred — in a human norovirus challenge, α1-acid glycoprotein was associated with RBP and with the retinol:RBP ratio controlling for CRP, and CRP was associated with both retinol and RBP controlling for AGP, across a month of follow-up Victor 2023. Healthy volunteers, a deliberate infection, and the vitamin A numbers moved with the inflammatory markers.
2. The correction for it — which is not free. The BRINDA method exists to adjust micronutrient biomarkers for CRP and AGP, and there is a published practical guide to doing it Luo 2023. But adjusting trades one error for another: against the modified relative dose-response test, inflammation adjustment of retinol and RBP improved specificity and reduced sensitivity Suri 2021. Anyone who tells you the adjustment simply fixes the problem has not read the validation.
3. Recent intake and non-fasting state, which add circulating retinyl esters on top of the retinol pool.
4. Light and temperature exposure of the sample.

For scale. The WHO deficiency cut-off is 0.70 µmol/L and the marginal-status cut-off 1.05 µmol/L — a gap of 0.35 µmol/L. That is well inside the shift an acute-phase response can produce, and it is the arithmetic behind the rule that a retinol drawn during illness cannot be classified at all Luo 2023.

One negative result worth carrying: when retinol was measured alongside α-tocopherol and CoQ10 in children on statin therapy, there was no significant difference in retinol between groups Motkowski 2022 — a useful counterweight to the assumption that any lipid-lowering therapy depletes every fat-soluble vitamin.

Reference interval or decision threshold — which kind of number Vitamin A is

A deficiency floor, defined for populations rather than for you. The widely used WHO survey cut-offs — deficiency below 0.70 µmol/L, marginal status below about 1.05 µmol/L — were built to estimate the prevalence of vitamin A deficiency in a population, which is a different job from classifying one person. Between those floors and the top of the range there is no outcome-anchored optimum, and this site will not invent one; its verdict for this marker is deficiency floor, not an optimum.

Two conversions, because US laboratories report µg/dL while the WHO cut-offs are quoted in µmol/L. A retinol of 20 µg/dL (200 ng/mL) is 0.70 µmol/L, the deficiency threshold. A retinol of 30 µg/dL (300 ng/mL) is 1.05 µmol/L, the marginal one.

Those population cut-offs are also under active revision, which is the honest state of play rather than a criticism: the challenge work above ends with a call for updated WHO guidance on the RBP cut-off Victor 2023, and the adjustment methodology has its own validation literature with real trade-offs Luo 2023 Suri 2021.

At the top end the meaningful threshold is toxicity, and serum retinol is a poor sentinel for it. In hypervitaminosis A it is fasting retinyl esters — expressed as a fraction of total circulating vitamin A — that rise first, while retinol itself is still being held in range by the same homeostasis that makes it a bad deficiency test. A normal retinol is not evidence that a high-dose supplement is safe.

The intake arithmetic belongs beside the serum number. 1 µg of retinol activity equivalent is 1 µg of retinol but 12 µg of dietary beta-carotene, because conversion is inefficient — beta-carotene 15,15′-dioxygenase performs it and common variants in the gene reduce it substantially. The adult tolerable upper intake level for preformed vitamin A is 3 mg (3,000 µg RAE) per day, and it is set on hepatotoxicity and teratogenicity rather than on any blood concentration. Isotretinoin, which some readers of this page will be taking, is itself a retinoid with a plasma half-life of roughly 10–20 hours.

How you would know your Vitamin A was wrong — and when to redraw

Three months, and the reason is the size of the store. The hepatic vitamin A pool turns over on a scale of months, and serum retinol is defended against changes in it. A retest sooner than about twelve weeks after an intake change is measuring the draw, not the store. This is the opposite of a marker like zinc, which has no reserve and moves in weeks.

Three timescales are in play and they differ by orders of magnitude. Plasma retinol has a half-life measured in hours. An ingested retinyl ester load clears from circulation over days. The hepatic pool the number is defended from turns over across months. A retest is only interpretable when it is timed against the third of those, which is why 12 weeks is the floor and a 6-week redraw is a wasted tube.

What would have to change for a second value to mean anything. hs-CRP on the same sample, and ideally AGP, both unremarkable — because a retinol that has risen while inflammation has settled has told you about the inflammation Victor 2023 Luo 2023. Then zinc plasma or zinc RBC, and albumin: a low retinol with a low zinc or a low albumin is a transport problem, and correcting the vitamin will not fix it.

And if the question is the other direction — too much rather than too little — the informative test is not another retinol. It is a fasting retinyl ester fraction with liver enzymes, because that is the pair that moves first in hypervitaminosis A while retinol is still normal.

What Vitamin A cannot tell you

It cannot measure your liver stores across the normal range. That is not a defect in the assay; it is what homeostasis does. A mid-range retinol is compatible with substantially depleted stores and with comfortably adequate ones.

It cannot be interpreted during any infection without an inflammatory marker — and even with one, the adjustment costs sensitivity to buy specificity Suri 2021.

It is not an early warning for toxicity, which is the reason many people order it. Retinyl esters, not retinol, are the early signal.

And a low value does not establish deficiency, while a normal one does not exclude it. The wrong inference here is unusually expensive: a mildly low retinol drawn during a viral illness, read as deficiency, treated with high-dose vitamin A — the one supplement on this site where over-correction is both hepatotoxic and, in pregnancy, teratogenic. If a low retinol appears while you are unwell, the correct next action is to get well and redraw, not to supplement.

Sources read for these sections

• Victor CP, et al. Vitamin A biomarkers were associated with AGP and CRP over the course of a human norovirus challenge infection. British Journal of Nutrition 2023 · PMID 37694547
• Suri DJ, et al. Inflammation Adjustments to Serum Retinol and Retinol-Binding Protein Improve Specificity but Reduce Sensitivity when Estimating Vitamin A Deficiency Compared with the Modified Relative Dose-Response Test in Ghanaian Children. Current Developments in Nutrition 2021 · PMID 34386690
• Luo H, et al. A Practical Guide to Adjust Micronutrient Biomarkers for Inflammation Using the BRINDA Method. Journal of Nutrition 2023 · PMID 36792034
• Motkowski R, et al. Effect of Statin Therapy on the Plasma Concentrations of Retinol, Alpha-Tocopherol and Coenzyme Q10 in Children with Familial Hypercholesterolemia. Cardiovascular Drugs and Therapy 2022 · PMID 33052507
• Zheng J, et al. Biomarkers of Micronutrients and Phytonutrients and Their Application in Epidemiological Studies. Nutrients 2023 · PMID 36839326