Factor V Leiden in your DNA file: rs6025, clotting risk and the numbers behind it

The brake that Leiden weakens

A clot forms through a chain of clotting factors, each switching on the next, and the chain ends with thrombin, the enzyme that turns fibrinogen into the fibre mesh of a clot. Factor V circulates inactive until an injury starts clotting. Activated, as factor Va, it pairs with factor Xa, and the pair converts prothrombin into thrombin^[4]. The same protein is also part of the brake: activated protein C cuts factor Va at specific sites, which shuts it down and keeps clots from growing too large^[4].

In 1993 Björn Dahlbäck and colleagues in Malmö described families with repeated venous clots whose plasma responded poorly to activated protein C in a clotting test. They called it a poor anticoagulant response to activated protein C, now usually shortened to APC resistance, and could not yet say what caused it^[1]. A year later the Leiden group found a cause: a G to A change in F5 that replaces arginine with glutamine at position 506, one of the sites activated protein C cuts. Factor Va carrying the change was not properly inactivated by activated protein C^[2].

The variant goes by several names. R506Q and Arg506Gln count amino acids in the mature protein, as the 1994 paper did; current nomenclature counts from the start of the precursor protein and calls the same change p.Arg534Gln^[5]. In a raw DNA file it is the row or record with the reference number rs6025, at position 169,549,811 of chromosome 1 on the GRCh38 assembly and 169,519,049 on the older GRCh37^[6].

# rsid    chromosome  position   genotype
rs6025    1           169519049  CT

F5 is read from the minus strand of chromosome 1, so the same change is a G to A in the gene's own sequence and a C to T on the chromosome's reference strand. Unlike some variants, this one cannot be misread from the wrong strand. The two letter pairs, C and T against G and A, share no letter, so a row reading CT or GA is one copy whichever strand the chip reported, a row reading TT or AA is two copies, and a row reading CC or GG is none. The HFE variant H63D is the counter-example, and that article explains why.

#CHROM  POS        ID      REF  ALT  GT
chr1    169549811  rs6025  C    T    0/1

One copy, two copies, three studies

Three Dutch and Danish studies carry most of the weight on venous thromboembolism, the umbrella term for deep-vein thrombosis and pulmonary embolism. They agree on one copy and disagree on two, because of how they were designed.

The Leiden Thrombophilia Study genotyped 471 consecutive patients under 70 with a first confirmed deep-vein thrombosis and 474 healthy controls. Heterozygotes had about 7 times the risk of non-carriers. Seven patients were homozygous and no control was, so the authors estimated the homozygous risk from the number of homozygous controls they would have expected, about 0.1, and arrived at roughly 80 times^[7]. The homozygous patients had their first clot at a mean age of 31, against 44 for heterozygous patients^[7].

The Copenhagen studies started from the general population and waited. The Copenhagen City Heart Study followed 9,253 randomly selected Danish adults for up to 23 years and found hazard ratios of 2.7 for one copy and 18 for two^[8]. A 2013 analysis of that cohort together with the larger Copenhagen General Population Study, 66,001 white Danes in all, found 2.2 and 7.0^[9].

A hazard ratio says how much faster events happen in one group than another. It says nothing about how often they happen, and venous thrombosis is uncommon in most of the people who carry Factor V Leiden. The Copenhagen City Heart Study turned its figures into absolute 10-year risks for groups defined by age, weight and smoking. For one copy the 10-year risk was 0.7 percent in non-smokers under 40 with a body mass index under 25, and 10 percent in the oldest, heaviest smokers. For two copies the same groups were at 3 and 51 percent, from very few people and with wide intervals^[8].

The same 2013 Copenhagen analysis looked for an effect on arteries and found none: the hazard ratios for myocardial infarction did not differ from 1 for either variant. It also found that ABO blood type accounted for more venous thromboembolism in the population than Factor V Leiden did, about 20 percent against 10^[9].

Oestrogen and pregnancy, as the studies and guidelines describe them

The first study of the variant and the contraceptive pill together was published in 1994, months after the variant itself. A Dutch case-control study of 155 premenopausal women with deep-vein thrombosis and 169 controls found that oral contraceptive use raised the risk about 3.8 times and carrying Factor V Leiden about 7.9 times. Women with both had 34.7 times the risk of women with neither, with a confidence interval from 7.8 to 154^[10]. The interval is wide because the combined group was small.

Pregnancy and the weeks after birth weigh more. A Dutch family study followed 798 female relatives of people with a clot and a known clotting variant. Compared with time spent neither pregnant nor taking a combined pill, the hazard ratio for venous thromboembolism was 16.0 during pregnancy and the postpartum period and 2.2 during combined pill use. Across all the time observed, women with one clotting defect had 0.35 events per 100 years of follow-up, against 0.13 for relatives with none^[11].

Menopausal hormone therapy has its own measurement. In the Women's Health Initiative trial of oestrogen plus progestin, venous thrombosis occurred at 3.5 per 1,000 women a year on the hormones and 1.7 on placebo, a hazard ratio of 2.06. Women who carried Factor V Leiden and took the hormones had 6.69 times the risk of non-carriers on placebo^[12].

Clinical guidelines turn these measurements into categories. The CDC's 2024 U.S. Medical Eligibility Criteria for Contraceptive Use list known thrombogenic mutations, naming Factor V Leiden among them, and rate each method from 1, a condition with no restriction on the method's use, to 4, a condition that represents an unacceptable health risk with it^[13].

The same guideline says routine screening of the general population for these mutations before starting contraception is not recommended^[13], and the American Society of Hematology's 2023 guideline on thrombophilia testing made the same point. Of its 23 recommendations it made one strong one: against testing the general population before combined oral contraceptives. Nearly all the rest were conditional, on very low certainty in the evidence^[14].

For pregnancy, the American Society of Hematology's 2018 guideline separates one copy from two. For women with one copy of Factor V Leiden and no previous clot, it suggests against preventive anticoagulation during pregnancy, with or without a family history of clots. For women with two copies and a family history it suggests it^[15]. Both are conditional recommendations addressed to clinicians, and the one for a single copy rests on very low certainty evidence^[15].

Prothrombin G20210A, the sibling variant

The second common inherited clotting variant was found by the same Leiden group two years later. Prothrombin G20210A, rs1799963, is a G to A change in a part of the F2 gene that is not translated into protein, and it raises the amount of prothrombin in the blood. In the first report it was carried by 18 percent of 28 patients from thrombosis-prone families and 1 percent of 100 healthy controls^[16]. In the Copenhagen cohorts it carried a hazard ratio of 1.5 for one copy and 11 for two, the latter from very few people, with an interval from 2.8 to 44^[9].

Raw files handle it differently from Factor V Leiden. 23andMe ships it under an internal identifier instead of its rs number, and our reports missed it until September 2026; the Prothrombin row in 23andMe files tells that story. 23andMe's own Hereditary Thrombophilia report covers exactly these two variants and says it does not describe a person's overall risk of thrombosis^[17].

Finland carries Leiden at the European rate

Many variants that matter in Finland are either far more common there than elsewhere in Europe or nearly absent; the Finnish disease heritage is the first kind. Factor V Leiden is neither. In gnomAD's genomes it is on 2.3 percent of Finnish chromosomes and 2.6 percent of other European ones. It is rarer in South Asian, admixed American and African or African American groups, and gnomAD found no copy among 5,182 East Asian chromosomes^[18].

An allele frequency counts chromosomes, and each person has two. If people pair up without regard to F5, the share carrying at least one copy is one minus the square of the share of chromosomes without it: 1 − 0.9766 × 0.9766, about 4.6 percent of Finns, against 5.2 percent of other Europeans. Two copies is the frequency squared, 0.0234 × 0.0234, about 1 in 1,800 Finns. Those are our arithmetic on gnomAD's counts, and the direct count agrees: 4 of the 5,308 Finnish genomes have two copies, about 1 in 1,300 from a small number^[18]. MedlinePlus puts carriers at 3 to 8 percent of people with European ancestry^[3].

The sibling variant splits the other way. Prothrombin G20210A is on 0.52 percent of Finnish chromosomes against 1.36 percent of non-Finnish European ones, and none of the 5,308 Finnish genomes has two copies^[19]. A Finnish file is about as likely as any European one to show Factor V Leiden, and about a third as likely to show the prothrombin variant.

What our report says, file by file

Every kind of file can carry this one substitution. What differs is what the report can say when the variant is absent.

The ClinVar rule gates every clinical finding: the variant must be classified pathogenic or likely pathogenic with at least one review star, and a one-star record goes to the lower-confidence expander. The gVCF's "screened" is about the gene's coverage, not proof at this one position. Only the Deep Read note changes its wording with the number of copies.

Every tier also computes a polygenic score for venous thromboembolism, PGS000043, from 297 variants found in a 2019 genome-wide study^[21]. Factor V Leiden is not one of them, so the score and the card are separate readings. The score is shown only to people who consented to sensitive findings. The free file check says which kind of file you have before anything is paid for.