Homocysteine Blood Test: What It Measures and How to Interpret It

A homocysteine blood test measures total homocysteine (tHcy) in plasma or serum. It is typically reported in µmol/L, with most labs using a reference interval of roughly 5–15 µmol/L. It is a sensitive downstream readout of one-carbon metabolism. When vitamin B12, folate, or vitamin B6 are short, or when the kidneys clear less, homocysteine accumulates. Treat it as a metabolic sensor. The number is not a diagnosis. A single elevated value with no context tells you almost nothing, because sample handling alone can move the number more than a real biological change.

What the assay quantifies

Homocysteine in plasma exists in four pools: free reduced homocysteine (~1%), homocystine (the disulfide dimer), mixed disulfides with cysteine, and protein-bound homocysteine (~70–80%, mostly bound to albumin cysteine-34). “Total homocysteine” means all four after a reducing step, usually dithiothreitol or tris(2-carboxyethyl)phosphine, which liberates the bound fraction. Every clinical assay does this. If a report says “homocysteine” without qualification, it is tHcy.

Two method families dominate:

• Enzymatic cycling / immunoassay (Abbott Architect, Roche Cobas, Siemens). Cheap, run on the main chemistry line, CV around 3–5%. This is what most consumer panels use.
• LC-MS/MS, often multiplexed with methylmalonic acid (MMA), cystathionine, and methionine. Better specificity, and you get the companion analytes from one draw. Slower turnaround, higher cost.

If you can choose, pick the LC-MS/MS panel that reports tHcy plus MMA plus methionine. The joint pattern is what makes the number interpretable, and running them from the same aliquot removes cross-draw variance.

Preanalytics decide half your result

This is where most homocysteine numbers go wrong. Red blood cells continue synthesizing and exporting homocysteine after the draw. In whole blood left at room temperature, tHcy climbs on the order of 10% per hour and keeps climbing for many hours. A tube that sat on a phlebotomy counter for three hours before centrifugation can turn an 8 into an 11 with no change in your biology.

What to insist on:

• EDTA plasma, not serum, unless the lab’s method is validated for serum and centrifuges fast.
• Tube on wet ice immediately after the draw.
• Centrifuge and separate plasma within 30 minutes, 60 at the outside.
• Fasting, 10–12 hours. A methionine-rich meal raises tHcy within hours.
• Same time of day across repeat draws. There is a modest diurnal pattern, and morning fasting draws are the cleanest baseline.

For at-home collection: dried blood spot and most fingerstick kits are poor for homocysteine specifically because the cells stay in contact with the sample. If you use a home kit, use one that ships a separated-plasma device (a microtube with a plasma separation membrane) and gets it cold, and expect a wider interval than a phlebotomy draw. If the kit says “let the card dry at room temperature for four hours,” the homocysteine value from it is not comparable to a lab draw.

Reference intervals, age, and what “high” means

Population distributions shifted downward after folic acid fortification of grain in the US and elsewhere, so older reference ranges run higher than current ones. Practical anchors:

• Below ~10 µmol/L: unremarkable in a fortified population.
• 10–15 µmol/L: common, and in most large cohorts this band already carries some risk signal.
• 15–30 µmol/L: moderate elevation. Nutrient status, renal function, thyroid, and drugs explain the great majority.
• 30–100 µmol/L: intermediate. Worth a clinician’s workup for B12 deficiency, renal disease, or a heterozygous inherited defect.
• Above 100 µmol/L: suggests classical homocystinuria (cystathionine β-synthase deficiency) or severe B12 deficiency. This is a see-a-clinician-now number. A value this high is not a supplement question.

tHcy rises with age and is 1–2 µmol/L higher in men than in women at the same age, driven by muscle mass, creatinine, and estrogen effects. In population data, the biochemical correlates of elevation are consistent and unsurprising: lower serum B12 and folate, higher MMA, higher creatinine, and hematological indices consistent with macrocytosis.¹ Homocysteine also tracks with inflammatory markers and lipid fractions in cohort data, which is one reason isolating its independent contribution to cardiovascular risk has been difficult.²

The panel to run alongside it

Homocysteine on its own is an alarm with no address. Order these together:

Analyte	What it separates
Serum B12 + methylmalonic acid	MMA is elevated in B12 deficiency, normal in folate deficiency. High tHcy + high MMA points to B12; high tHcy + normal MMA points to folate
Holotranscobalamin	The metabolically available B12 fraction. Better than total B12, which includes haptocorrin-bound
RBC folate	Reflects 3–4 months of status, unlike serum folate which tracks the last meal
Creatinine + cystatin C, eGFR	Renal clearance is one of the strongest determinants of tHcy. Correct for it before concluding anything nutritional
TSH, free T4	Hypothyroidism raises tHcy
Albumin	Most homocysteine is albumin-bound; low albumin lowers measured tHcy

Drugs matter too: metformin, methotrexate, some anticonvulsants, nitrous oxide exposure, and proton pump inhibitors all push the number up through B12 or folate pathways. Bring your medication list to whoever interprets this.

Acute illness confounds it as well. In surgical cohorts, serum homocysteine rises with ischemic tissue injury and has been evaluated as a predictor of intestinal necrosis in bowel obstruction.³ Do not baseline yourself during or shortly after an acute illness.

Genotype: what MTHFR tells you and what it does not

If you have a whole genome, pull the three variants that show up in every discussion of this pathway:

# GRCh38 coordinates
bcftools view -r chr1:11796321 sample.vcf.gz | \
  bcftools query -f '%CHROM\t%POS\t%ID\t%REF\t%ALT\t[%GT]\n'   # rs1801133, MTHFR C677T
# MTHFR is on the minus strand: 677C = G on the reference strand, 677T = A

Also worth extracting: rs1801131 (MTHFR A1298C), rs1805087 (MTR A2756G), rs1801394 (MTRR A66G) and rs1801198 (TCN2 C776G).

MTHFR 677TT homozygotes are roughly 10% of people of European ancestry and produce a thermolabile enzyme with about 30% of wild-type activity. The genotype effect on tHcy is real but modest and conditional: it shows up mainly when folate status is low, and largely disappears at good folate status. A TT genotype with an 8 µmol/L tHcy and a normal RBC folate is a genotype with no phenotype. The genotype is a modifier of your response. A genotype alone is not a result. In one intervention in older women, short-term folic acid lowered homocysteine but also shifted glucose and lipid measures, with C677T genotype modifying the response. That is a good illustration of why “supplement to the genotype” is the wrong reflex.⁴

Mendelian randomization and the large B-vitamin trials have carried the interpretive weight here. Lowering homocysteine pharmacologically produced smaller cardiovascular benefit than the observational associations predicted, though signals persist for stroke and for cognitive endpoints in people with adequate B12 status.⁵ Homocysteine has also been studied as a thrombosis marker, where the same gap between association and causal effect appears.⁶

Using it longitudinally

The value of homocysteine to someone tracking their own data is as a repeated measure. Biological within-person variation is on the order of 7–10% CV when preanalytics are controlled, so a change of less than about 15–20% between two draws is noise. Fix the protocol and a series of four draws over a year will tell you something a single draw never can. The protocol has these parts.

• same lab
• same method
• same fasting state
• EDTA plasma on ice
• spun in 30 minutes

If you are tracking a nutritional change, the response is fast. Folate and B12 repletion move tHcy within 2–6 weeks, and it plateaus by about 8–12 weeks. Recheck at 8 weeks. Do not recheck at 2. Anything that does not move by 12 weeks under adequate B-vitamin status is pointing at renal function, thyroid, drugs, or a genuine enzymatic defect, and that is a conversation with a physician.

One caution about the downstream biology: low central folate can exist with unremarkable peripheral labs, as in cerebral folate deficiency, and researchers are working on non-blood markers for exactly that gap.⁷ Plasma tHcy is a systemic average. The value is not a tissue readout.

Questions people also ask

How do I know if my homocysteine is high? Only by measuring it. Mild to moderate elevation produces no symptoms at all, which is why it is measured and not felt. Severe elevation from untreated B12 deficiency produces symptoms of the deficiency (neuropathy, anemia, cognitive change). The homocysteine itself does not produce them.

What is the root cause of high homocysteine? In order of frequency in healthy adults: low folate or B12 status, reduced renal clearance, age, hypothyroidism, and medications. Genetic variants like MTHFR 677TT amplify the nutritional cases rather than causing elevation on their own. Rare inherited CBS deficiency produces the extreme values.

Does magnesium lower homocysteine? There is no reliable evidence that magnesium supplementation lowers tHcy. The B vitamins in the pathway (folate, B12, B6 and riboflavin as the MTHFR cofactor) are the ones with measured effects, and riboflavin’s effect is largely confined to 677TT genotypes. Talk to a clinician before starting anything, particularly because folic acid can mask B12 deficiency while nerve damage continues.

What are normal homocysteine levels by age? Most labs apply one adult interval, usually 5–15 µmol/L. The underlying distribution shifts upward with age and is higher in men, so a 10 µmol/L result means something different at 30 than at 75. Ask your lab for its own age-partitioned intervals if it publishes them.

How long does it take to lower homocysteine? When the cause is nutritional, most of the drop occurs in the first 4–6 weeks and plateaus by 8–12. Retest at 8 weeks with identical preanalytics.

Can homocysteine damage be reversed? The number itself is readily lowered in nutritional cases. Whether lowering it reverses vascular or neurological changes is a separate question, and the trial evidence is mixed, with the clearest signals for stroke and cognitive endpoints rather than coronary events.⁵ This is a question for your physician in the context of your full history.

Footnotes

1. May A. Beydoun, Hind A. Beydoun, Peter H. MacIver, et al. Biochemical and Hematological Correlates of Elevated Homocysteine in National Surveys and a Longitudinal Study of Urban Adults. Nutrients, 2020. https://doi.org/10.3390/nu12040950 ↩

2. Iris Shai, Meir J. Stampfer, Jing Ma, et al. Homocysteine as a risk factor for coronary heart diseases and its association with inflammatory biomarkers, lipids and dietary factors. Atherosclerosis, 2004. https://doi.org/10.1016/j.atherosclerosis.2004.07.020 ↩

3. Guojun Wang, Su Jing, Youlong Zhu, et al. Serum homocysteine as a novel biomarker for predicting intestinal necrosis in patients with bowel obstruction: a prospective observational study. Scientific Reports, 2026. https://doi.org/10.1038/s41598-026-64566-w ↩

4. Agata Chmurzyńska, Anna M. Malinowska, Jolanta Twardowska-Rajewska, et al. Elderly women: Homocysteine reduction by short-term folic acid supplementation resulting in increased glucose concentrations and affecting lipid metabolism (C677T MTHFR polymorphism). Nutrition, 2013. https://doi.org/10.1016/j.nut.2012.09.015 ↩

5. A. David Smith, Helga Refsum. Homocysteine – from disease biomarker to disease prevention. Journal of Internal Medicine, 2021. https://doi.org/10.1111/joim.13279 ↩ ↩²

6. Bushra Moiz. Homocysteine as a Biomarker for Thrombosis: What is the Learning Curve?. eCommons - AKU (Aga Khan University), 2017. https://ecommons.aku.edu/pakistan_fhs_mc_pathol_microbiol/869 ↩

7. George Ayoub, Craig Brown. Retinal Biomarkers of Folate and Vitamin B12 Metabolic Dysfunction: A Framework for Machine Learning-Assisted Detection of Cerebral Folate Deficiency. Cells, 2026. https://doi.org/10.3390/cells15161453 ↩