Does MTHFR C677T Cause Depression?

Oak
mthfr genomics biomarkers folate variant-interpretation
A dark feathered reptile specimen on a black plinth, studio-lit, with green pigment glowing unevenly through the translucent vessels of its neck.

No. Carrying one or two copies of MTHFR C677T does not cause depression, and knowing your MTHFR genotype in isolation tells you very little about your mood. The variant is extremely common (roughly 30 to 40 percent of chromosomes in European and East Asian populations carry the T allele, with TT homozygotes around 10 percent in Europe and higher in parts of Mexico and northern China), and the reported effect sizes on depression risk are odds ratios in the 1.1 to 1.4 range in meta-analyses of heterogeneous case-control studies. A risk factor that common with an effect that small cannot be a cause in any useful sense. What it can do is shift a biochemical measurement, and the measurement is what you should care about: homocysteine, red blood cell folate, and B12. If your MTHFR status matters to you at all, it matters because it changes your prior on those numbers, not because it explains how you feel.

This piece covers what C677T and A1298C do enzymatically, how to pull your own genotype out of a VCF file without paying a consumer testing company to interpret it, what to measure downstream, and why the genotype-first framing that dominates search results for this topic gets the causal arrow backwards.

What the variants do enzymatically

MTHFR encodes methylenetetrahydrofolate reductase, the enzyme that converts 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate (5-MTHF), the folate form that donates a methyl group to homocysteine to regenerate methionine. C677T (rs1801133, an alanine-to-valine substitution at codon 222) destabilizes the enzyme’s flavin adenine dinucleotide cofactor binding, reducing activity to roughly 65 percent of wild type in heterozygotes and 30 percent in TT homozygotes under low-riboflavin conditions. A1298C (rs1801131, glutamate to alanine at codon 429) sits in the regulatory domain and has a smaller effect on activity, with the compound heterozygous state (677CT/1298AC) producing a reduction somewhere between CT and TT.

The downstream consequence that is well replicated is mild hyperhomocysteinemia in TT homozygotes, and it is conditional on folate status. In populations with adequate folate intake, TT homozygotes often have homocysteine indistinguishable from CC. That conditionality is the whole story. The genotype sets how steep your response curve is to folate and riboflavin intake, and the phenotype only diverges when intake is low. This is the canonical example of a gene-diet interaction in nutritional genomics, and it is the reason position statements in the field have consistently argued that genotype should inform interpretation of measured nutritional status rather than substitute for it.12

The link to mood runs through two proposed mechanisms, neither of which is settled. The first is that 5-MTHF is the rate-limiting cofactor for regeneration of tetrahydrobiopterin (BH4), which is required by tryptophan hydroxylase and tyrosine hydroxylase for serotonin and catecholamine synthesis. The second is that elevated homocysteine and its precursor S-adenosylhomocysteine inhibit methyltransferase reactions broadly, including those involved in phospholipid and neurotransmitter methylation. Both are plausible. Neither has produced a reliable genotype-to-symptom mapping in an individual.

Finding your genotype in your own data

If you have whole-genome sequencing, both variants are straightforward to genotype yourself, and doing it directly is better than trusting a report because you get to see depth and quality. Both sit on chromosome 1 in the 11.79 to 11.80 Mb region on GRCh38 (rs1801133 at chr1:11,796,321, rs1801131 at chr1:11,794,419; on GRCh37 these are 11,856,378 and 11,854,476, so confirm which reference your VCF used by reading the ##reference and ##contig header lines).

bcftools view -r chr1:11794419,chr1:11796321 \
  -Ov sample.g.vcf.gz \
| bcftools query -f '%CHROM\t%POS\t%ID\t%REF\t%ALT\t%QUAL[\t%GT\t%DP\t%GQ]\n'

Two failure modes to watch for. First, if your VCF is a filtered variants-only file rather than a gVCF, a homozygous reference call at these positions appears as absence, which is indistinguishable from no coverage. Always check depth at the locus in the BAM or CRAM directly with samtools mpileup -r chr1:11796321-11796321 -f GRCh38.fa sample.cram before concluding you are wild type. Second, the strand and reference allele conventions for rs1801133 are a recurring source of error: dbSNP reports the variant on the minus strand relative to the gene, so “C677T” in the literature corresponds to a reference G to alternate A change in GRCh38 coordinates. If your answer disagrees with an array-based report, this is usually why. Cross-check against the rsID in the VCF rather than reasoning from the letters.

Array-based consumer genotyping calls both SNPs reliably, so if you already have a 23andMe or AncestryDNA raw file, the genotype there is almost certainly correct. What is not correct is the interpretation layered on top of it.

What to measure instead

The genotype is a fixed, low-information prior. The biochemistry is a measurable, modifiable state, and it is where you should spend your money and attention. Four assays give you the relevant picture.

Plasma total homocysteine is the direct functional readout of the remethylation pathway. It must be drawn fasting and the sample separated from cells within about 30 to 60 minutes, because red cells continue to export homocysteine and a tube sitting at room temperature will drift upward by several micromoles per liter per hour. If your homocysteine came back elevated and the lab processed it slowly, repeat it before believing it.

Red blood cell folate rather than serum folate, because serum folate reflects the last few meals while RBC folate integrates over the roughly 120-day erythrocyte lifespan. Serum B12 plus either methylmalonic acid or holotranscobalamin, since B12 deficiency is a far more common cause of elevated homocysteine than MTHFR genotype and is the thing you would want to find. And riboflavin status, because the C677T enzyme’s instability is specifically an FAD-binding defect, and riboflavin repletion partially rescues activity in TT homozygotes.

A retrospective cohort study of genotype-guided nutraceutical support in a mental health setting reported outcomes stratified by MTHFR genotype, which is the kind of design that generates hypotheses about who responds to what.3 It is not a randomized trial, and retrospective cohorts in this space carry selection effects that are hard to bound. Treat the literature here as suggestive of where to look rather than as a basis for action.

If your homocysteine is elevated, or you are managing depression, this is the point at which you involve a physician. Elevated homocysteine has differential diagnoses that include renal function, thyroid status, and B12 or B6 deficiency, and the workup is not something to do from a spreadsheet. We will not recommend a supplement, a dose, or a form of folate, and you should be skeptical of anyone who does so on the basis of a genotype alone.

Why the genotype-first framing fails

The pattern that dominates writing on this topic is: find a variant, assume a mechanism, prescribe a nutrient. The problem is that the variant is too common and the effect too small for individual prediction, and public health genomics has repeatedly made the point that population-level genomic findings require demonstrated clinical utility before they justify individual action.4 MTHFR genotyping has not cleared that bar for depression, which is why most professional bodies have moved away from recommending it as a standalone test.

There is a second reason to be cautious, which is that MTHFR is only one node in a network. One-carbon metabolism involves MTR, MTRR, CBS, BHMT, SHMT1, and the folate transporters, and variants in several of these have been examined in the context of homocysteine-related phenotypes including migraine, where MTHFR C677T appears alongside a set of other polymorphisms with comparably modest effects.5 Assigning a phenotype to a single enzyme in a redundant network is a category error. The useful framing treats diet, genotype, and measured metabolite levels as an interacting system whose state you sample over time.67

The practical version of that: genotype once, since it never changes, then measure homocysteine, RBC folate, B12, and a marker of riboflavin status two or three times a year and look at the trajectory. A single homocysteine value tells you less than four values over two years, because within-person variability in this analyte is substantial and the interesting signal is drift. If you change your diet, measure before and after with at least four months between draws so RBC folate has time to equilibrate.

Questions people also ask

What foods should someone with MTHFR avoid? There is no food that C677T or A1298C carriers need to avoid. The gene-diet interaction runs in the direction of adequacy rather than avoidance: the genotype matters when folate and riboflavin intake are low, so leafy greens, legumes, liver, eggs, and dairy are the relevant foods, and they are relevant to everyone. Claims that carriers must avoid folic acid, gluten, or specific food groups are not supported by the interaction literature.12

What is the best supplement for MTHFR? We will not name one, because that is a clinical decision and it depends on your measured folate, B12, and homocysteine, not your genotype. The broader point is that supplementing on the basis of a genotype without a measurement is guessing, and in the specific case of B12, masking a deficiency with high-dose folate is a recognized hazard that requires a clinician’s involvement.

Does caffeine affect the MTHFR gene? No. Caffeine does not interact with MTHFR enzyme activity or expression in any established way. Caffeine metabolism is largely a CYP1A2 story, which is a separate pharmacogenetic question.

What mental illness is associated with the MTHFR gene? C677T has been reported in association with schizophrenia, major depression, and bipolar disorder across many case-control studies, with pooled odds ratios generally below 1.5. Those are population-level associations with wide heterogeneity between studies, and they do not support inferring a diagnosis or an explanation for symptoms in one person.

How do you treat MTHFR naturally? MTHFR genotype is not a condition to treat. If a measurement downstream of it is abnormal, that measurement is what gets addressed, and how it gets addressed depends on which one is abnormal and why, which is a conversation with a physician.

Oak builds longitudinal molecular profiles of individuals: whole-genome sequencing, RNA sequencing, proteomics, blood biomarkers, and continuous glucose data, integrated into one model of you. Build your profile.

Footnotes

  1. Kathryn M. Camp, Elaine Trujillo. Position of the Academy of Nutrition and Dietetics: Nutritional Genomics. Journal of the Academy of Nutrition and Dietetics, 2014. https://doi.org/10.1016/j.jand.2013.12.001 ↩ ↩2

  2. Patrick J. Stover, Marie A. Caudill. Genetic and Epigenetic Contributions to Human Nutrition and Health: Managing Genome–Diet Interactions. Journal of the American Dietetic Association, 2008. https://doi.org/10.1016/j.jada.2008.06.430 ↩ ↩2

  3. Cristina Beer, Fiona Rae, Mikayla Watt, et al. Personalised Nutraceutical Treatment Guided by MTHFR Genotype in Mental Health: A Retrospective Cohort Study. Nutrients, 2026. https://doi.org/10.3390/nu18172791 ↩

  4. Muin J. Khoury, Michael S. Bowen, Wylie Burke, et al. Current Priorities for Public Health Practice in Addressing the Role of Human Genomics in Improving Population Health. American Journal of Preventive Medicine, 2011. https://doi.org/10.1016/j.amepre.2010.12.009 ↩

  5. Syed Sameer Aga, Mujeeb Zafar Banday, Saniya Nissar, et al. Role of Single Nucleotide Polymorphisms (SNPs) in Common Migraine. Genetic Polymorphism and Disease, 2022. https://doi.org/10.1201/9781003246244-14 ↩

  6. Peter J. Gillies. Nutrigenomics: the Rubicon of molecular nutrition. Journal of the American Dietetic Association, 2003. https://doi.org/10.1016/j.jada.2003.09.037 ↩

  7. Baoxuan Lin, Ran Cheng, Min Jia, et al. Nutrigenomics in exercise science: interactions between gene, diet, and exercise. Food Science and Human Wellness, 2025. https://doi.org/10.26599/fshw.2025.9250529 ↩