Vitamin K Antagonists and Osteoporotic Fractures: Insights From Comparisons With the NOACs

Raffaele De Caterina; Santa Mundi; Maria Fusaro


Eur Heart J. 2020;41(10):1109-1111. 

Vitamin K antagonists (VKAs; warfarin mostly, but also acenocoumarol, phenprocoumon, phenindione, and fluindione) have been a cornerstone in medicine as the first and the only available oral anticoagulants (OACs) for clinical use for over half a century.[1] They have allowed incredible progress in cardiology and cardiac surgery in areas such as atrial fibrillation (AF), venous thrombo-embolism, prosthetic heart valves, and the artificial heart.[1] Indeed, after the first descriptions of vitamin K in the literature dating back to the 1930s,[2,3] discovery of the role of vitamin K in coagulation and thrombosis and the appraisal of its chemical nature allowed Henrik Carl Peter Dam and Edward Adelbert Doisy to win the Nobel Prize in Physiology or Medicine in 1943. The European Food Safety Authority recommends for all adults, including pregnant and lactating women, 70 μg/day of vitamin K1 (phylloquinone),[4] found in plants, which are its main dietary source. A second source of vitamin K is vitamin K2 (menaquinone), mainly synthesized by bacteria in the intestinal lumen and also found in cheese, meat, and fermented soya derivatives. Low amounts of vitamin K can be stored, so that the vitamin needs to be continually replenished through foods.

VKAs inhibit two key enzymes of the vitamin K cycle, mainly vitamin K epoxide reductase (VKOR) and, probably, to a lesser extent, vitamin K quinone reductase, leading to a dose-dependent deficit of vitamin K (Take home figure). This leads to inhibition of the vitamin K-dependent carboxylation of glutamic acid (Glu) into gamma-carboxyglutamic acid (Gla) residues in a variety of proteins, some of which have enzyme activity, known as vitamin K-dependent proteins (VKDPs) (Take home figure). These include coagulation factors, namely prothrombin (factor II), factor (F) VII, IX, X; and the anticoagulant proteins C, S, and Z. Inhibition of FII, VII, IX and X accounts for the anticoagulant action of VKAs.

Take home figure.

The human osteocalcin structure and the vitamin K cycle. Vitamin K hydroquinone (KH2) is a coenzyme of gamma-glutamyl carboxylase (GGCX) (1). Gamma-carboxylation of glutamic acid (Glu) residues leads to gamma-carboxyglutamic acid (Gla) residues in vitamin K-dependent proteins (VKDPs), such as human osteocalcin, here represented: this is characterized by three Gla residues at positions 17, 21, and 24, and one disulfide bond. Warfarin and other vitamin K structural antagonists block two important enzymes in the vitamin K cycle: vitamin K epoxide reductase (VKOR) (2) and—possibly—vitamin K quinone reductase (2*), resulting in a dose-dependent functional vitamin K deficiency and depletion of active vitamin K-dependent proteins. NADPH-dependent quinone reductase (3) inhibits warfarin, and is an escape route for the conversion of vitamin K quinone into the active vitamin K hydroquinone. Abbreviations: uc-VKDPs, undercarboxylated vitamin K-dependent proteins; c-VKDPs, carboxylated vitamin K-dependent proteins.

In addition to being the cofactor of coagulation factors, vitamin K is also, however, the cofactor of other proteins discovered much later, such as bone Gla protein (BGP, or osteocalcin), matrix Gla protein (MGP), growth arrest-specific 6 protein (Gas6), Gla-rich protein (GRP), and periostin,[5] mostly affecting bone, but also vascular health.[6] BGP, in addition to being a relevant player for correct bone mineralization (Take home figure), such that BGP knock-out mice develop hyperostosis,[7] may indeed have an intriguing protective role against vascular calcifications, as highlighted in human studies.[6]

Because of this evidence, the iatrogenic vitamin K deficiency induced by VKAs has been hypothesized to be associated with bone fractures and vascular calcifications similar to what occurs in states of nutritional vitamin K deficiency. Osteoporotic fractures are a major health problem in elderly people, with significant morbidity and mortality, and a considerable socioeconomic burden.[8] As OACs are usually prescribed to older patients, who are vulnerable at the same time to AF, venous thrombo-embolism, and osteoporotic fractures, there are clinical concerns about the possible effects of VKAs on fracture risks. VKA use had indeed been suggested to be associated with a higher risk for osteoporotic fractures in AF patients in a few older reports, but evidence has remained controversial, mostly due to inadequate design of such demonstrations. Systematic reviews on the topic have indeed come to opposite conclusions, one—comparing patients treated with VKAs with healthy controls or with medically ill patients—denying the association when eliminating confounding factors;[9] while two other, more recent, and based on the comparison of VKAs with the recently introduced non-vitamin K antagonist oral anticoagulants (NOACs), favouring it.[10,11] The NOACs which have become available over the past 10 years work by directly inhibiting either thrombin (dabigatran) or FXa (rivaroxaban, apixaban, edoxaban, and betrixaban), and are now largely replacing VKAs in preventing stroke and systemic embolism in AF and in the prevention and treatment of venous thrombo-embolism.[12] Because of their lack of interference with the vitamin K cycle, they now would offer an opportunity to assess the real extent of the effects of VKAs on bone health. Indeed, Lutsey et al., in a population of 162 275 patients with AF from administrative claims databases, reported a lower fracture risk in NOAC-treated patients [hazard ratio (HR) for hip fractures: 0.67, 95% confidence interval (CI) 0.45–0.98].[10] Formal randomized comparisons of the effects of these new drugs and VKAs on fracture risks are limited by variable, non-systematic, and non-uniform reporting of the fracture risk, and also by the limited follow-up of the randomized clinical trials that have allowed the approval of NOACs. Yet, evidence deriving from 12 such randomized controlled trials in the meta-analysis by Gu et al., involving 89 549 patients, among which 44 816 (50%) were receiving NOACs and 44 733 (50%) were receiving warfarin, has also pointed to a significantly lower risk of fracture in NOAC-treated compared with warfarin-treated patients [relative risk (RR) 0.82, 95% CI 0.73–0.93, P = 0.001].[11]

In this issue of the European Heart Journal, Huang et al. report a comparison of the fracture risks associated with the NOACs and warfarin among AF patients involved in a real-world nationwide retrospective cohort study using Taiwan's National Health Insurance Research Database.[13] Here all adult patients in Taiwan newly diagnosed with AF between 2012 and 2016 who received an NOAC or warfarin were enrolled and followed-up until 2017. Patients treated with an NOAC were subgrouped according to the NOAC used (rivaroxaban, dabigatran, and apixaban). Propensity score matching was performed for each head-to-head comparison. Cox regression analysis was used to calculate the adjusted HRs for hip, vertebral, and humerus/forearm/wrist fractures. After matching, 19 414 patients were included (9707 in each of the NOAC and warfarin groups). The median follow-up time was 2.4 years. Compared with warfarin, NOACs were also here associated with a statistically significant 16% lower fracture risk (adjusted HR 0.84; 95% CI 0.77–0.93; P < 0.001). Subanalyses revealed that all the NOACs had a similar fracture risk, always lower compared with warfarin.

This is therefore important additional evidence deriving from one single nationwide and comprehensive source, apparently confirming the results of most recent previous analyses. Precautions were taken, with propensity score matching, to limit the possibility of systematic bias, which—admittedly—can only be eliminated by randomized comparisons of the two types of OACs, now difficult to perform. Yet, the report by Huang et al. is a good example of the virtuous use of registry data (the so called 'real-world evidence') in areas where randomized controlled trials fall short of providing the full necessary evidence. The many other prospective registries being currently run with all the NOACs[14] could provide further additional confirmation on this important topic in other ethnicities. In parallel, further mechanistic studies on the effects on bone health of VKAs, which still have several viable indications,[1] are warranted.

For the time being, NOACs appear to present a better safety profile compared with VKAs also on bone health. This adds to the many other advantages of the NOACs, including convenience, the lower risk of intracranial haemorrhage, fewer drug and food interactions than the VKAs, and—probably as the result of some of these pluses—reduced mortality.[15]