Vitamin K from greens

Vitamin K is the one nutrient in this section with a cartilage-specific job description. Cartilage stays un-mineralised because matrix Gla protein inhibits calcification, and matrix Gla protein is inert until vitamin K carboxylates it — a switch, not a modifier. A second vitamin-K-dependent protein, Gla-rich protein, is the most carboxylated protein known and is abundant in cartilage, so the pathway has more than one arm.

The appraisal of the supplement — MK-7, its dose, its trials, and the evidence that cuts against it — lives in the vitamin K2 entry. This one is about the food.

K1 and K2, and why the split is softer than it sounds

Green vegetables carry phylloquinone, vitamin K1. Fermented foods and animal products carry menaquinones, vitamin K2, of which the long-chain MK-7 from natto is the form that reaches the circulation best at capsule doses. The usual telling of this makes K1 the inferior version.

It is not that clean. Tissues convert dietary phylloquinone into menaquinone-4 themselves, through a described enzymatic route — so eating greens is not simply a way of missing out on K2. How much MK-4 a given human tissue makes from a given dietary dose is not settled, which is a real gap rather than a rhetorical one, but the dichotomy the supplement market runs on is a simplification.

Absorption from a vegetable, measured

The step this section usually has to assume was done properly here. Broccoli was grown with deuterium so its phylloquinone carried a traceable label, fed to people, and the label was found in their serum. Phylloquinone from a vegetable is absorbed, and it is identifiable as having come from that vegetable.

A controlled feeding study adds the dose-response: vitamin K status tracked both how much phylloquinone was eaten and which food it came from, and younger and older adults responded differently to the same intake. The practical consequence is that "eat more greens" is not a single dose — the amount that moves status depends on the source, the fat eaten with it, and who is eating.

What the joint evidence says

The cohorts are consistent and the appraisal of them belongs to the supplement entry; the food-relevant readings are these.

Low vitamin K status has been associated with hand and knee osteoarthritis, and in the MOST cohort with incident knee OA rather than prevalent disease. Dietary vitamin K intake — intake, not blood concentration — has been set against symptomatic and structural change in knee osteoarthritis, which is the most directly food-relevant analysis in the set. And people taking vitamin K antagonist anticoagulants, the drugs that switch the pathway off, show more osteoarthritis incidence and progression: a natural experiment nobody would have been allowed to design.

Two analyses put vitamin K next to vitamin D. In the Health ABC knee OA substudy, people sufficient in both circulating vitamin K (≥ 1.0 nmol/L) and 25(OH)D (≥ 50 nmol/L) had better Short Physical Performance Battery scores and faster gait over follow-up (p ≤ 0.002); in the Osteoarthritis Initiative, combined sufficiency measured as dietary intake tracked with faster gait and quicker chair stands (p ≤ 0.029). Low vitamin K status also tracks with mobility limitation and disability in older adults more generally.

And the one randomized trial with a joint endpoint — K1 in hand osteoarthritis — was null overall, positive only in a post-hoc subgroup of people who had been insufficient to begin with. That subgroup is the reason the observational data have not been dismissed, and the reason the next trial is obvious.

Practical notes

Phylloquinone is fat-soluble and absorbed poorly from raw leaves eaten alone; absorption improves with fat in the meal and with cooking. Anyone taking a vitamin K antagonist should keep intake consistent rather than high or low — which is a matter for their prescriber, and made more interesting rather than less by the anticoagulant cohort finding above.

What would change this entry

The trial Shea 2018 asks for, in as many words: vitamin K and vitamin D together, in people insufficient in both. That is the group in which the null K1 trial's post-hoc subgroup moved, and it is the one testable version of the hypothesis this literature has been circling for twenty years.