The a1298c research journal

In 1988, Soo-Sang Kang’s group at Rush Medical College in Chicago did something simple: they heat-treated lymphocyte extracts from 21 young coronary-artery-disease patients at 46 °C for five minutes. Controls kept 37.6% ± 5.6% of their MTHFR activity; fifteen patients matched controls — but six retained only 13.6% ± 5.1%. A heat-sensitive, “thermolabile” enzyme, running in families and, apparently, in arteries. Follow-up work mapped the inheritance and extended the coronary association, but for seven years thermolabile MTHFR remained a biochemical ghost: measurable, heritable, unmapped.

In May 1995 a Leiden–Nijmegen–Montreal–Michigan collaboration pinned the ghost to a single base: 677C→T, an alanine-to-valine substitution, present on about 38% of unselected chromosomes — one of the most common “disease” variants ever described, hiding in plain sight in a third of humanity. Three years later the same Nijmegen group, hunting through neural-tube-defect families, found the gene’s second common mutation — 1298A→C, allele frequency 0.33 — and discovered that the two variants compound each other’s effects, while Rozen’s Montreal lab measured the new variant at about 60% of control enzyme activity and noted that nobody alive seems to be homozygous for both.

Then came the plot twist that defines this site’s editorial stance. When human MTHFR carrying the 1298 substitution was purified to homogeneity, the protein was indistinguishable from the wild-type enzyme — the cellular activity reduction measured in lymphocyte extracts did not survive purification. A genetic association, a cellular phenotype and a molecular mechanism turned out to be three different claims requiring three different proofs. The weaker variant of the pair became the 25-year reminder that association is not destiny — and that is exactly the standard a1298c.com holds every A1298C claim to.

In 1998 the United States began mandatory folic-acid fortification — an accidental, population-scale methylation experiment. Fortification carries hard-outcome evidence behind it for neural-tube-defect prevention, and it changed the population’s folate baseline so thoroughly that it quietly redefined every later trial. In the Framingham Offspring cohort, median plasma unmetabolised folic acid doubled in non-supplement users after fortification (0.25 to 0.50 nmol/L, P<0.001), and the prevalence of high circulating UMFA rose from 9.4% to 19.1%. The authors’ own sentence carries the honest state of play: the biochemical and physiologic consequences of this are unknown.

Against that fortified background, the large homocysteine-lowering trials reported null after null: VISP in stroke patients, HOPE-2 in vascular disease and diabetes (primary composite RR 0.95, 95% CI 0.84–1.07), NORVIT after myocardial infarction (RR 1.08, 0.93–1.25, with a trend toward harm on triple therapy), SEARCH in 12,064 MI survivors, and VITATOPS in recent stroke (composite RR 0.91, 0.82–1.00). The trial-level synthesis — 8 RCTs, 37,485 participants (Clarke et al., Arch Intern Med 2010, PMID 20937919) — found no significant effect of folic acid on major vascular events (rate ratio 1.01, 95% CI 0.97–1.05) despite about 25% homocysteine reduction, and the 2017 Cochrane review agreed (Martí-Carvajal et al., PMID 28816346), with only a small stroke effect favouring lowering.

Then the exception that teaches the rule: CSPPT randomised 20,702 hypertensive Chinese adults — a largely non-fortified, folate-insufficient population the 1998 experiment never reached — within MTHFR C677T strata, and enalapril plus folic acid cut first stroke from 3.4% to 2.7% (HR 0.79, 95% CI 0.68–0.93). Inside CSPPT, genotype and baseline folate modified the homocysteine response, and the degree of homocysteine decline tracked stroke risk. Read together, the record says one coherent thing: the folate cycle matters most where folate is scarce. CSPPT-2 is now recruiting to test the genotype-guided version of that idea — stratified, again, by C677T.

Start with the discovery hope. The 1998 van der Put paper floated A1298C as an additional risk factor for neural-tube defects, with a compound-heterozygosity odds ratio of 2.04 whose 95% CI of 0.9–4.7 crossed 1. Two decades later the largest synthesis — 22 case-control studies, 3,224 NTD fetuses against 3,295 controls — found no evidence for a significant association between fetal MTHFR A1298C and NTD risk, overall or in any stratum. The early hope did not survive contact with sample size.

The autism arc is the cleanest refutation in the file. A 2013 meta-analysis suggested MTHFR associations with autism spectrum disorder; the larger 2020 synthesis separated the variants — C677T yes, A1298C no — and a 2026 ethnicity-stratified meta-analysis confirmed it again: a statistical association between ASD and C677T, but none between ASD and A1298C. Meanwhile the American College of Medical Genetics and Genomics states that MTHFR polymorphism testing, for either variant, has minimal clinical utility and should not be part of routine thrombophilia evaluation. The pattern repeats across domains: null in childhood leukaemia, null in preeclampsia at meta level, null for methotrexate response in one major synthesis, contradictory in recurrent pregnancy loss.

What remains genuinely open is more interesting than what was retired. Whether A1298C matters clinically on its own at all — its homocysteine effect is small and folate/B12-status-dependent. Whether the recurrent-pregnancy-loss signal, positive in one large meta-analysis and null in the umbrella review, is real or an artifact of ethnicity-specific linkage with C677T. Whether the colorectal-cancer protective direction, seen twice in Asian datasets, survives confirmatory work. And why a regulatory-domain variant whose purified protein behaves normally would track with cervical and lymphoma risk up but colorectal risk down in the same populations. These are the questions this site watches — with the nulls on the same shelf as the positives, where they belong.