Vitamin K2: What is the Optimal Dose?

Another way to ask this question is as follows: what is the minimum effective dose to achieve the maximal desired effect? While there is no established toxicity for high doses, there are good reasons to be cautious before taking far more than we need (see below), hence the term “minimum effective dose.” At the same time, we don’t want to reap just some of the health benefit. We want to reap as much of the health benefit as we can in a safe and effective manner, hence the term “maximal desired effect.”

The only rigorous way to approach this is to look at dose-finding studies, which are studies where different doses were directly compared with one another. Ideally, the studies are randomized, controlled, long enough in duration to believe the dose was able to achieve its full effect, and conducted within a context where we would expect to see a benefit.

Pharmacological Doses of MK-4

A Japanese dose-finding study compared 15, 45, 90, and 135 milligrams per day (mg/d) of MK-4 to reduce fracture risk in postmenopausal women with osteoporosis and found 45 mg/d to be the minimal effective dose (Iwamoto, 2013). This is a pharmacological dose that is hundreds of times greater than what can be obtained from food. It probably works through mechanisms that are independent of the those seen for nutritional doses of vitamin K, such as overriding the body’s natural regulation of bone resorption. Thus, we should view MK-4 at these doses with the same type of cost-benefit analysis we would use for other osteoporosis drugs, like Fosamax, and we should not use these studies to determine the optimal nutritional dose of MK-4.

Nutritional Doses of MK-4

Unfortunately, there is a dearth of dose-response studies for nutritional doses of MK-4. Nakamura (2014) compared the effect of 0, 300, 600, 900, and 1500 micrograms per day (μg/d) on osteocalcin carboxylation, a marker of vitamin K status in bone. They fed everyone the same doses in the same order, increasing the dose from 0 one week at a time. The carboxylation status did not change with 300 μg/d, but improved with 600 μg/d. However, it is not at all clear that 300 μg/d would not have provided the same benefit if given for longer than one week. I do not consider this study to offer any clear insight about the optimal dose of MK-4.

MK-7 in Healthy Populations

Dalmeijer (2012) compared 180 and 360 μg/d MK-7 to a placebo given to healthy, non-obese men and postmenopausal women aged 40-65 years over the course of twelve weeks. The mean K₂ intake from food was 25 μg/d, so these treatments effectively compared total K₂ intakes of 25, 200, and 380 μg/d. Both treatment doses lowered desphospho-uncarboxylated MGP (dp-ucMGP), a marker of vitamin K deficiency in blood vessels, and improved the carboxylation status of osteocalcin. While 360 μg seemed to cause a slightly larger effect than 180 μg, the lion’s share of benefit came from 180 μg and the difference between the two doses was not statistically significant. Thus, the study hints at a possible benefit of doses higher than 200 μg that would have to be confirmed in future studies with greater statistical power, but provides rigorous evidence only that 200 μg is better than 25 μg.

Knapen (2012) reported a more extensive array of doses given to healthy men and premenopausal women aged 25-45 over the course of twelve weeks. The doses included 0, 10, 20, 45, 90, 180, and 360 μg/d MK-7 and the primary endpoint of interest reported was the carboxylation status of osteocalcin. Unfortunately, the sample size (n=42) was small for having so many groups, precluding a rigorous statistical analysis of the endpoints between each group. Additionally, while carboxylated osteocalcin levels were similar across groups at baseline, undercarboxylated osteocalcin levels were highly variable. The changes in undercarboxylated osteocalcin between baseline and the study’s end within any given group were generally about the same size as the difference in baseline values between groups. All of this makes it extremely difficult to know whether the the difference between groups for changes in undercarboxylated osteocalcin or its ratio to total osteocalcin are true biological differences or simply random variation resulting from noisy data.

Doses that were 90 μg/d or greater caused statistically significant decreases in undercarboxylated osteocalcin, but only the 180 μg and 360 μg doses increased the levels of carboxylated osteocalcin or improved the ratio. From among these measurements, the increase in carboxylated osteocalcin seen with the two higher doses is most convincing because the variation in baseline values for that measurement was so low. The ending values for this measurement were higher in the 180 and 360 μg groups than in any of the the others, but they were nearly identical between groups. K₂ intake from food was not reported, but presumably would have added at least 20 μg/d to the doses. I therefore consider this study to offer limited support to 200 μg/d as the optimum dose for improving vitamin K status at bone.

Inaba (2015) compared 0, 50, 100, and 200μg/d MK-7 in postmenopausal women aged 50 to 69 years over the course of four weeks. The primary endpoint of interest was the carboxylation status of osteocalcin, reported as the ratio of the carboxylated to the undercarboxylated form. The study was conducted in Hokkaido, Japan, where natto is popular. The subjects were required to avoid all MK-7-rich foods and to consume prepared meals that provided 65 μg/d of total vitamin K as a combination of K₁ and MK-4 in unspecified proportions. Whether intentional or not, this is effectively a study of how much MK-7 you need to preserve the carboxylation status of your osteocalcin when you stop eating natto. Indeed, the largest effect across all groups was for carboxylation status to significantly worsen in the 0 μg/d group. Carboxylation status was significantly different from that group in the 100 and 200 μg/d groups, but not in the 50 μg/d group. The authors did not report a statistical analysis for the difference between 100 and 200 μgd, but 200 μg/d was the only group in which carboxylation status actually improved over the course of the study. I therefore consider this study to offer limited support to 200 μg/d as the optimum dose for improving vitamin K status at bone.

In further support of this conclusion, Ikeda (2006) found that postmenopasual women who reported consuming enough natto to provide 200 μg/d K₂ or more (mostly as mostly MK-7) suffered less bone loss over the course of three years than women who consumed less. Since all lower intakes of natto were grouped together for the statistical analysis, it is not clear exactly where the line of maximal benefit lies, and it may be less than 200 μg/d. As an observational study, we should also be more cautious about inferring cause and effect. Nevertheless, the fact that it measured an actual health endpoint (bone loss) instead of just a surrogate marker (osteocalcin carboxylation), and the fact that it was three years long instead of four to twelve weeks, makes it very worthy of consideration.

MK-7 in Hemodialysis Patients

Westenfeld (2012) and Caluwé (2014) both conducted dose-finding studies in hemodialysis patients. Patients with kidney disease have high levels of vascular calcification, which is a major contributor to mortality in this population. Since MGP protects blood vessels from calcification, dp-ucMGP was the major endpoint in both studies. Lower dp-ucMGP suggests better vitamin K status in blood vessels and a better defense against pathological calcification.

Westenfeld compared 45, 135, and 360 μg/d MK-7 over six weeks. MK-7 dose-dependently decreased dc-ucMGP, with the effect almost doubling in size for each increase in the dose from 18% to 37% to 61%. However, change from baseline analysis is vulnerable to regression to the mean and it is more rigorous to compare the absolute levels of dc-ucMGP after treatment. When looked at this way, 135 and 360 μg/d had equal benefit over 45 μg/d.

Nevertheless, Caluwé later tested even higher doses and provided evidence of benefit for more than 300 μg/d. They fed the patients 360, 720, or 1080 μg MK-7 three times per week for eight weeks, which equates to average daily doses of 154, 309, and 463 μg. MK-7 dose-dependently decreased dp-ucMGP by 17%, 33%, and 46%. The ending dp-ucMGP values were lower in the 463 μg/d group than in the 309 μg/d group and were lower than Westenfeld found after feeding 360 μg/d, but they were still about four times higher than that found in healthy controls. The average K₂ intakes in the Caluwé paper were 16 μg/d, making the doses compared effectively 325 and and 479 μg/d. Future studies may clarify whether even higher doses can bring dp-ucMGP levels even closer to those found in healthy controls. Thus, there is strong evidence that the K₂ requirement for kidney patients is higher than 325 μg, possibly as high as 480 μg, and may well be much higher than that.

Conclusions

For healthy populations, there is no smoking gun, but several studies converge towards the conclusion that 200 μg/d is the optimal dose. Most of the benefit probably comes from the first 100 μg, and the evidence for the superiority of 200 μg is limited. There may be benefits to higher doses, but there is no convincing evidence of that at this time. Thus, there is a high likelihood that I will revise my recommendation as new data comes in, but I currently recommend at least 100 μg/d and preferably 200 μg/d.

For kidney disease patients, there is good reason to see 480 μg/d as better than lower doses. Since 480 μg/d almost slashes dp-ucMGP in half yet leaves it four times higher than in healthy controls, the optimal dose may well be much higher than this. I suspect it is at least 1 mg/d. Nevertheless, K₂ should only be used to augment treatment for kidney disease under medical supervision.