The ViKTORIES trial: A randomized, double‐blind, placebo‐controlled trial of vitamin K supplementation to improve vascular health in kidney transplant recipients

Premature cardiovascular disease and death with a functioning graft are leading causes of death and graft loss, respectively, in kidney transplant recipients (KTRs). Vascular stiffness and calcification are markers of cardiovascular disease that are prevalent in KTR and associated with subclinical vitamin K deficiency. We performed a single‐center, phase II, parallel‐group, randomized, double‐blind, placebo‐controlled trial (ISRCTN22012044) to test whether vitamin K supplementation reduced vascular stiffness (MRI‐based aortic distensibility) or calcification (coronary artery calcium score on computed tomography) in KTR over 1 year of treatment. The primary outcome was between‐group difference in vascular stiffness (ascending aortic distensibility). KTRs were recruited between September 2017 and June 2018, and randomized 1:1 to vitamin K (menadiol diphosphate 5 mg; n = 45) or placebo (n = 45) thrice weekly. Baseline demographics, clinical history, and immunosuppression regimens were similar between groups. There was no impact of vitamin K on vascular stiffness (treatment effect −0.23 [95% CI −0.75 to 0.29] × 10−3 mmHg−1; p = .377), vascular calcification (treatment effect −141 [95% CI − 320 to 38] units; p = .124), nor any other outcome measure. In this heterogeneous cohort of prevalent KTR, vitamin K supplementation did not reduce vascular stiffness or calcification over 1 year. Improving vascular health in KTR is likely to require a multifaceted approach.


| INTRODUC TI ON
Cardiovascular disease is a major cause of morbidity in patients with chronic kidney disease (CKD) and is most pronounced in those with end-stage kidney disease (ESKD). The risk of cardiovascular disease is attenuated by kidney transplantation, but cardiovascular risk remains elevated compared with the general population, 1 progresses with declining transplant function 2 and is a major cause of death with a functioning graft. 3 In CKD, cardiovascular risk can be partly attributed to atherosclerotic risk factors seen in the general population, such as smoking, dyslipidemia, and hypertension. However, secondary prevention strategies for cardiovascular disease (including antiplatelet and statins) are inadequate to address the excess cardiovascular risk seen in kidney transplant recipients (KTRs). [4][5][6][7][8] People with advanced CKD including KTR exhibit markers of cardiovascular aging that are more CKD-specific, and disproportionate for age and gender, such as vascular stiffness and calcification. [9][10][11] These markers are independently and significantly associated with cardiovascular disease and mortality among KTRs, 12-14 and there are no specific treatments available to improve vascular stiffness and calcification.
Vascular calcification impairs the ability of a vessel to stretch and relax, resulting directly in vascular stiffness, increased afterload, hypertension, and left ventricular hypertrophy. 15 Vascular calcification in CKD results from a relative excess of calcification promoters (e.g., CKD mineral and bone disorder, dyslipidemia, and diabetes) compared with inhibitors (including vitamin K-dependent proteins and magnesium). 16 Vitamin K is an essential cofactor for posttranslational carboxylation (activation) of a number of calcification inhibitors, such as Matrix Gla protein, osteocalcin, and Gla-rich protein. 17 In subclinical vitamin K deficiency, these calcification inhibitors are incompletely activated, and vascular calcification can progress unopposed. 18 Activity of these calcification inhibitors, and thus vitamin K status, can be estimated by measuring the levels of undercarboxylated enzyme in the blood. 19,20 Of these, desphospho-undercarboxylated Matrix Gla Protein (dp-ucMGP) may be the most sensitive for detecting changes in vitamin K status. 21 Vitamin K deficiency is common among KTRs (at least in part due to suboptimal dietary intake 22 ) and is associated with cardiovascular disease and increased mortality in this population. 23 Vitamin K supplementation may provide an inexpensive and low-risk treatment to attenuate progression of vascular stiffness and calcification in KTR by optimizing the function of vitamin K-dependent calcification inhibitors.

Trials of vitamin K to improve vascular stiffness and calcification
show promising results in various populations. 24 We designed the ViKTORIES trial (Vitamin K in kidney Transplant Organ Recipients: Investigating vEssel Stiffness) to investigate the hypothesis that vitamin K supplementation improves vascular stiffness and calcification over 1 year in prevalent KTR.

| MATERIAL S AND ME THODS
The full ViKTORIES methods have been published previously. 25

| Trial design and participants
This was a single-center, phase II, parallel-group, randomized, doubleblind, placebo-controlled trial in prevalent KTR. Adult participants (18 years or over) who had a functioning kidney transplant for a year or more (eGFR >15 ml/min/1.73 m 2 by CKD-EPI 26 ) were included.
There was no minimum threshold for vascular stiffness or calcification. We excluded participants with the following: permanent or paroxysmal atrial fibrillation (known or identified at screening), warfarin use, taking vitamin K or indication for vitamin K, allergies to constituent ingredients of the study interventions (gelatin, lactose, or cellulose), breastfeeding or of childbearing potential, known glucose-6-phosphate dehydrogenase deficiency, life expectancy <12 months, standard contraindications to magnetic resonance imaging (MRI), 27 or inability to provide written informed consent in English.

| Interventions
Full details of the choice and preparation of interventions have been detailed previously. 25 In brief, participants were randomized to menadiol diphosphate 5 mg or a matching placebo, administered orally three times per week (Monday, Wednesday, Friday) for 12 months.
Menadiol diphosphate has not been used previously for this indication; however, it is a licensed preparation to correct vitamin K deficiency and has demonstrated similar biological activity to phytomenadione (vitamin K1) for correction of coagulopathy associated with liver disease. 28 This preparation was selected as it facilitated production of a matching placebo. Daily requirement of vitamin K is around 1 µg/kg. 29 The dose of 5 mg thrice weekly was, therefore, considered adequate to ensure saturation of vitamin K stores.

| Randomization sequence generation
A computer-generated code list was provided by Sealed Envelope (Sealed Envelope Ltd) in a password-protected file to the drug manufacturer (Tayside Pharmaceuticals) to produce identical, sequentially numbered bottles containing either vitamin K or placebo. The code list was organized in random permuted blocks to facilitate 1:1 unstratified allocation ratio.

| Allocation concealment mechanism
Randomization was conducted by study investigators using a custom-built, password-protected, web-based system, created and maintained by a third party not otherwise involved in the trial (Sealed Envelope Ltd).

| Implementation
Participants were recruited (J.S.L.) from routine outpatient transplant clinic appointments in the west of Scotland (NHS Greater Glasgow and Clyde with a Patient Identification Centre in NHS Lanarkshire [Scotland, UK]) and enrolled into the study, including obtaining informed consent and randomizing according to methods detailed above.

| Blinding
Local investigators, research nurses, pharmacy staff, and participants were blinded to treatment allocation by use of numbered but otherwise identical medication bottles. Investigations including laboratory analysis of blood and urine samples, pulse wave velocity measurements, and quality of life questionnaires were conducted by blinded study investigators or laboratory staff. After enrollment in the study, participants were given a five-digit study ID (two-digit site code "01" followed by three-digit sequential screening code).

| Data capture
A custom-designed electronic case report form was used, designed (by J.S.L.) using Castor Electronic Data Capture (www.casto redc. com; Amsterdam, the Netherlands).

| Outcomes
The primary outcome was between-group difference measured by MRI-based ascending aortic distensibility at 12 months. All secondary outcomes were assessed as between-group differences at 12 months and were tested hierarchically in the following order: coronary artery calcification score (CACS) by non-contrast computed tomography (CT), carotid-femoral pulse wave velocity and augmentation index (SphygmoCor XCEL PWA and PWV software, AtCor Medical Pty Ltd), MRI measures of cardiac structure and function (descending aortic distensibility, left/right ventricular mass, function and peak systolic strain, T1 and T2 relaxation times), office blood pressure, electrocardiogram (ECG), calcium metabolism and bone turnover markers (calcium, phosphate, parathyroid hormone, 25-hydroxyvitamin D), transplant function, proteinuria, and quality of life (EuroQol EQ-5D-5L instrument 30 ).

| Vitamin K status
dp-ucMGP was used as a marker of vitamin K status to confirm biological effect of supplementation. dp-ucMGP was measured in thawed plasma samples using an automated chemiluminescence sandwich immunoassay (InaKtifMGP) provided by ImmunoDiagnosticSystems (IDS PLC, Tyne and Wear) and processed on an IDS-iSYS instrument in the biochemistry laboratory at Glasgow Royal Infirmary, NHS Greater Glasgow and Clyde. During assay verification, there were concerns about linearity due to problems with variable recovery particularly at the lower end of the assay (<900 pmol/L). Reporting of absolute values of dp-ucMGP was possible above 900 pmol/L, but not below, and dp-ucMGP values <900 pmol/L were, therefore, considered to be equal to 900 pmol/L.

| Analysis and statistical considerations
ViKTORIES was analyzed in line with CONSORT guidelines, and the analysis plan was published in advance of locking the database (Data S1). The primary analysis was conducted by modified intention-to-treat (participants were included if they completed two scans irrespective of whether they stayed on study medication), except changes in dp-ucMGP, which were studied in participants who completed the study per-protocol. Outcomes were assessed as between-group difference at 12 months by two-way analysis of covariance (ANCOVA) adjusting for age, duration of ESKD, and the baseline value of the variable of interest. Plasma dp-ucMGP concentrations were log-transformed prior to analysis. Prespecified subgroup analysis was conducted in older (≥65 years) versus younger (<65 years) participants, testing for AD = Maximum aortic volume (ml) − Minimum aortic volume (ml) Minimum aortic volume (ml) × Pulse pressure (mmHg) multiplicative interaction effect between age and treatment group and heterogeneity of treatment effect. 35 Exploratory analyses by ANCOVA were conducted to assess the impact of baseline dp-ucMGP on vascular outcomes of interest, and ANCOVA repeated as for main outcomes in a subgroup of participants considered to be vitamin K deficient at baseline (dp-ucMGP >900 pmol/L). We conducted multiple imputation for missing data (using the average of five separately imputed data sets) for the main outcomes where the assumption of missing at random was met. 36 The number and characteristics of adverse events (AEs) were summarized as a whole and by study arm. Results from the ViKTORIES trial were combined with other published reports in an updated meta-analysis (using a random effects model) according to methods described previously. 24 Analyses were conducted using stats, subgroup, mice, and meta packages for R statistical software 37 (version 3.5.3 or higher).

| Sample size
The sample size was calculated to detect a 1.0 × 10 −3 (SD 1.3 × 10 −3 ) mmHg −1 improvement in ascending aortic distensibility in the intervention group relative to placebo at 12 months. This difference was shown to be the minimum meaningful difference associated with cardiovascular outcomes in a historical cohort of hemodialysis patients from our own center. 31 To achieve a power of 90% with alpha = 0.05 required 37 patients per group (74 in total). We anticipated 20% dropout over 12 months of follow-up and, therefore, recruited 90 participants.

| Follow-up and timetable
Randomized participants completed visits at baseline and 12 months for all relevant clinical, biochemical, and radiological data, with monitoring visits at 1 and 6 months. The study duration was selected by extrapolating the duration from other studies where a positive result was obtained (range 1-3 years), 24 but there was an influence of trial cost and feasibility on the duration selected. We considered that if no significant impact was demonstrated on surrogate markers at 1 year, the influence on hard cardiovascular outcomes over the longer term was likely to be minimal. Office. Participants with unresolved AEs at the last study visit were followed up until resolution or 30 days (whichever was sooner).

| Baseline data
Baseline data for all randomized participants are illustrated in Tables 1 and 2

TA B L E 2 (Continued)
and ESKD were similar across treatment groups, as were baseline levels of vascular stiffness, calcification, and dp-ucMGP.

| Secondary outcomes
Analyses were conducted hierarchically as described but displayed in full for completeness. There was no significant difference in any clinical, biochemical (Table 4), or imaging measure (Table 5).

| Vitamin K status and biological effect
Vitamin K status was similar in treatment groups at baseline. Vitamin K deficiency was evident in 15 (33.3%) versus 14 (31.1%) participants (p = .881) in vitamin K and placebo groups respectively. In n = 72 participants who completed the study per-protocol and had available dp-ucMGP measurements, vitamin K caused a reduction in dp-ucMGP (i.e., improved vitamin K availability and activity) compared

| Inter-and intra-observer consistency of agreement for magnetic resonance imaging outcome measures
On blinded repeat analysis of 20% of randomly selected data sets in random order, there was good or excellent inter-and intraconsistency of agreement on most outcome measures (Table S1).

| Adverse events
AEs were common in both treatment arms, although relatively less common in the group treated with vitamin K (57.8% vs. 80.0%):

TA B L E 3
Primary outcome by intention-to-treat analysis: aortic distensibility

| Prespecified subgroup analysis in older versus younger participants
Prespecified subgroup analysis was conducted in participants who were older (age ≥65 years; n = 20) or younger (age <65 years; n = 70) at randomization, based on a known positive association between vascular stiffness, calcification, and increasing age. There was no evidence of multiplicative interaction effect between treatment group and age for ascending aortic distensibility (p = .892) nor for CACS (p = .981).
Subgroup analysis in older versus younger participants shows no heterogeneity of treatment effect that cannot be explained by chance (p = .480). Those who were older at randomization displayed more progression in CACS when treated with vitamin K, although there was a significant interaction between age, duration of ESKD, and baseline CACS (p = .003) in these participants, which are all likely to be predictive of progressive calcification. There were no differences in any of the imaging outcomes by treatment group in younger participants.

| Sensitivity analysis: multiple imputation
Multiple imputation was conducted for missing outcome data for n = 88 participants (excluding n = 2 patients who died). The proportion of data missing at random for aortic distensibility (ascending and descending) was 5.7%; CACS 3.4%; left ventricular mass index 6.8%; pulse wave velocity 5.7%; and augmentation index 3.4%. Multiple imputation had no meaningful impact on the results of these selected endpoints (Tables 3 and 5).

| Update of meta-analysis
Among published reports across various populations, and including results from ViKTORIES, vitamin K supplementation does not TA B L E 5 Imaging secondary outcomes by intention-to-treat analysis  Figure 3B). Details of the trials included in this updated meta-analysis are available in Table S2.

| DISCUSS ION
The ViKTORIES trial showed no evidence that treatment with a synthetic form of vitamin K improves vascular stiffness or calcification in a heterogeneous group of KTRs. When these results were combined with other published reports in meta-analyses, there was no observed impact of vitamin K supplementation on vascular stiffness or calcification across varying populations.
In CKD, vitamin K deficiency is associated with reduced activity of calcification inhibitors and vascular calcification 38,39 and may be associated with cardiovascular disease and early mortality. 24 There are several reasons why supplementation may not have induced clinical benefit in this trial.
We assumed that vitamin K as menadiol diphosphate could function as a cofactor for carboxylation of vitamin K-dependent proteins.
Menadiol diphosphate is effective in the treatment of coagulation abnormalities, 28 proving activity as a cofactor for carboxylation of vitamin K-dependent clotting factors. We tested the ability of menadiol diphosphate to facilitate carboxylation of Matrix Gla protein.
Plasma dp-ucMGP testing is not routinely available in clinical laboratories; commercial companies (VitaK, the Netherlands) are no longer offering analysis. We used the only available dp-ucMGP assay, which demonstrated suboptimal performance at the lower end of the scale in our local laboratory, and we were unable to report accurate numerical values below 900 pmol/L. Griffin et al. suggested reference values for healthy Caucasian individuals from <300 to 532 pmol/ L 40 (although no assay validation was offered for values below 900 pmol/L). We, therefore, assumed that those with dp-ucMGP >900 pmol/L were "deficient" and those ≤900 pmol/L "sufficient" (or mildly deficient). Only 29 included participants had dp-ucMGP >900 pmol/L, and it is possible that a substantial proportion of participants were not vitamin K deficient at baseline. However, there was a clear improvement in the absolute values of dp-ucMGP in those who were adherent with study medication, confirming biological activity of menadiol diphosphate. Our results are consistent with other studies in CKD using different preparations of vitamin K; thus, it is less likely that the null result was due to the choice of preparation alone. It was not possible to confirm vitamin K status in advance of inclusion due to the lack of a reliable biomarker. Consideration could be given to testing the utility of vitamin K supplementation F I G U R E 2 Individual change in log dp-ucMGP (pmol/L) by treatment group from baseline to 12 months for n = 72 participants who completed the study per-protocol with quantifiable dp-ucMGP at baseline and 12 months. Values <900 pmol/L could not be measured reliably and were considered to be equal to 900 pmol/L. Data have been logtransformed for ease of viewing values at the lower end of the scale (either alone or in combination) in patients with confirmed vitamin K deficiency.
We used vascular stiffness (MR-based aortic distensibility) as the primary endpoint. Aortic distensibility is an accurate and highly reproducible measure of central vascular stiffness, that is strongly and inversely associated with cardiovascular risk. 31 Aortic distensibility has been shown to be modifiable in response to therapeutic interventions to reduce cardiovascular risk in CKD. 41 Vascular stiffness is modifiable over time, such as after kidney transplantation in patients with ESKD, 42 and after vitamin K supplementation in some, 43,44 but not all, 24 trials.
We detected no signal that vitamin K altered vascular stiffness; it is unlikely that the choice of endpoint explained the null result.
Vascular stiffness and calcification develop alongside exposure to multiple risk factors and an imbalance of calcification promoters and inhibitors. 16 CKD-specific risk factors-like CKD mineral and bone disorder-begin early in the disease process, and the effects persist after correction of kidney function with transplantation.
Vitamin K is primarily obtained from green vegetables, so vitamin K deficiency may serve as a marker of prolonged exposure to un- In conclusion, we did not find any sign that vitamin K supplemen- Note. Number of participants with adverse events by MedDRA System Order Class (SOC), number of deaths and participants who commenced renal replacement therapy. All participants who received at least one dose of study medication were included in safety analyses. All serious adverse events were characterized as expected according to the study protocol.

ACK N OWLED G M ENTS
We would like to thank the kidney research nurses, clinical tri- Centre.

D I SCLOS U R E
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author on reasonable request.