Safety of Vitamin K in mechanical heart valve patients with supratherapeutic INR: A systematic review and meta-analysis

Background:

Patients who had mechanical heart valves and an international normalized ratio (INR) of >5.0 should be managed by temporary cessation of vitamin K antagonist. This study aimed to investigate the safety of low-dose vitamin K1 in patients with mechanical heart valves who have supratherapeutic INR.

Methods:

CINAHL, Cochran Library, Clinical trial.gov, OpenGrey, PubMed, ScienceDirect, and Scopus were systematically searched from the inception up to October 2021 without language restriction. Studies comparing the safety of low-dose vitamin K1 treatment in patients with placebo or other anticoagulant reversal agents were included. We used a random-effect model for the meta-analysis. Publication bias was determined by a funnel plot with subsequent Begg’s test and Egger’s test.

Results:

From 7529 retrieved studies, 3 randomized control trials were included in the meta-analysis. Pooled data demonstrated that low-dose vitamin K was not associated with thromboembolism rate (risk ratio [RR] = 0.94; 95% CI: 0.19–4.55) major bleeding rate (RR = 0.58; 95% CI: 0.07–4.82), and minor bleeding rate (RR = 0.60; 95% CI: 0.07–5.09). Subgroup and sensitivity analysis demonstrated the nonsignificant effect of low-dose vitamin K on the risk of thromboembolism. Publication bias was not apparent, according to Begg’s test and Egger’s test (P = .090 and 0.134, respectively).

Conclusion:

The current evidence does not support the role of low-dose vitamin K as a trigger of thromboembolism in supratherapeutic INR patients with mechanical heart valves. Nevertheless, more well-designed studies with larger sample sizes are required to justify this research question.

1. Introduction

Over 4 million people worldwide have received prosthetic heart valves. An estimated 300,000 valves are implanted every year, mainly to improve the quality of life and survival of patients with severe valvular heart disease.^[1] According to the guidelines, all patients who are diagnosed with valvular heart disease, particularly those who have mechanical heart valves, should use oral vitamin K antagonists (VKAs) to prevent thromboembolism in either artery or vein.^[2] VKAs inhibit vitamin K-dependent g-carboxylation of several coagulation factors, for example, II, VII, IX, and X^[3], preventing blood coagulation. International normalized ratio (INR) is used as a surrogate for the monitoring of VKA therapy. In patients with mechanical heart valves, the target of INR should be ranging between 2.5 and 3.5.^[4]

Supratherapeutic INR during VKA treatment can be caused by multiple reasons, for example, interactions with other drugs or food, inappropriate VKA dosing, or poor adherence to the VKA regimen.^[5] Patients with prosthetic heart valve implantation experiencing an increased INR of >4.0 to 4.5 are at risk of bleeding, especially when anticoagulant treatment is not ceased.^[6] Bleeding can be inevitable if the INR elevation is not properly managed.^[7] The 2020 American College of Cardiology (ACC)/American Heart Association (AHA) Guideline^[2] suggested that for patients with mechanical valves and uncontrollable bleeding, 4-factor prothrombin complex or fresh frozen plasma (FFP) is preferred to high-dose vitamin K for the management of bleeding. Vitamin K administration with temporary cessation of vitamin K antagonist has been suggested in individuals with mechanical heart valves and INR of >5.0 who are not actively bleeding. However, the evidence on whether vitamin K1 should be used for managing supratherapeutic INR, which is >5.0, in patients with prosthetic heart valves is still conflicting. Evidence shows that using vitamin K1 as an antagonist in such patients increases the risk of prosthetic valve thrombosis.^[8] On the contrary, some randomized controlled trials have proved that using low doses of vitamin K to reverse anticoagulation in patients tends to reduce bleeding and does not link to a higher risk of having thrombosis.^[9–11]

Although this topic is not new since vitamin K has been used for warfarin overdose for a long time, there are still misunderstandings that low dose vitamin K can cause thromboembolism in patients with mechanical heart valves with warfarin overdose. The information supporting the use of warfarin in such patients derives from small RCTs. Therefore, conducting this systematic review and meta-analysis increases the power of the analysis on this topic and provides valuable data for healthcare providers treating patients with mechanical heart valves.

2. Methods

2.1. Search strategy

This study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020.^[12] The study protocol was registered with PROSPERO (CRD42022289966) in December 2020. Systematic searches were conducted in 7 databases including CINAHL, Cochran Library, Clinical trial.gov, OpenGrey, PubMed, ScienceDirect, and Scopus. The searches were conducted without language and study design restrictions and from their inception to October 30, 2021. The following search strategy was used: “anticoagulant OR warfarin OR ‘vitamin K’ antagonist AND ‘heart valve prosthes’ OR ‘mechanical heart valve’ AND ‘vitamin K’ OR phytonadione” (Table 1, Supplemental Digital Content 1, http://links.lww.com/MD/H179).

2.2. Study eligibility criteria

This study included randomized controlled trials, cohort studies, and case-control studies in which patients received vitamin K (phytomenadione). In those studies, controls were no treatment, placebo, or other anticoagulant reversal agents. The included studies had to ostensibly illustrate statistics necessitate for the meta-analysis, for example, risk ratio (RR), odds ratio (OR), hazard ratio with a 95% confidence interval (95% CI), or the number of thromboembolism and bleeding with the total number of the patients. The definition of thromboembolism and bleeding outcomes are detailed in (Table S2, Supplemental Digital Content 2, http://links.lww.com/MD/H180).^[11] We excluded nonresearch articles, nonhuman studies, case series or case reports, cross-sectional studies, and retrospective studies. Studies whose effect estimates could not be extracted were also excluded.

2.3. Data extraction

Two investigators independently screened titles, abstracts, and full texts of the retrieved articles independently. Disagreements were resolved by consulting the third author. Also, three investigators independently extracted the author’s name, country, study design, sample size, effect estimates, participant characteristics, comorbidities, treatment regimens, controls, and outcomes. Data that were not available were retrieved by contacting the corresponding authors of selected articles. If the corresponding authors did not respond in a reasonable time, the articles were excluded.

2.4. Quality assessment

Three investigators independently assessed the risk of bias or quality of the included studies. For RCTs, the Cochrane Risk-of-bias tool 2.0 (RoB 2.0)^[13] and Grading of Recommendations, Assessment, Development and Evaluations (GRADE) for randomized control trials were used for evaluating the risk of bias and the quality of evidence, respectively. The Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I tool) were used to evaluate the risk of bias in nonrandomized trials because it was accessible and easy to use.^[14] The Newcastle-Ottawa Scale^[15] were used to assess the quality of the observational studies because of its reliability, validity, and ease of use.^[16]

2.5. Statistical analysis

Pooled risk ratio and 95% CI were calculated using the DerSimonian-Laird random-effect models^[15] (STATA, version 16.0, StataCorp LLC, USA) to explain the risk of thromboembolism from using low-dose vitamin K in the patients. Heterogeneity was assessed using Cochrane Q statistic and I² values. The P value of Cochrane Q of <.10 or I² of >75% indicated high heterogeneity while the P value of >.01 or I² of lower than 25% indicated low heterogeneity.^[16] The funnel plot was used to observe for the publication bias. Further tests were conducted using (RevMan 5.3, The Nordic Cochrane Center, Copenhagen, Denmark). Begg’s test and Egger’s test were used to further detect^[17,18] and the trim-and-fill method was performed to further justify the publication bias.^[19]. A value of 0.5 was used to impute the data points that were zero.^[20]

2.6. Subgroup analysis and meta-regression

The influence of baseline characteristics, which can be the cause of heterogeneity, was determined by subgroup and sensitivity analyses, and meta-regression. The following strata were planned priori: age ≥ 65 years or <65 years, sex, and types of valvular. Meta-regression was calculated using OpenMetaAnalyst for Windows 8^[21] posthoc.

3. Results

3.1. Studies retrieved and characteristics

We retrieved 7529 studies from the databases. After the removal of duplicate articles (n = 341) and impertinent articles (n = 6863), 3 articles, all of which were randomized controlled trials studies (RCT),^[9–11] were included for data synthesis. The PRISMA diagram for the study screening is shown in Figure 1. The essential characteristics of included studies are exhibited in Table 1. In brief, the PICO of the included studies was described as the following. The total number of participants from the 3 included studies was 241. The majority of them were East Asian with the mean age of 59 years (range: 50.08 years–66.00 years). More than 50% of participants were female. All studies measured the efficacy and safety of oral or intravenous vitamin K and used FFP^[10], no treatment^[9] or placebo^[11] as comparators. The outcomes measured were thromboembolism, major bleeding, and minor bleeding in all studies.^[9,11]

Figure 1.

The PRISMA flow chart of the study selection process. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Table 1.

Characteristics of studies included in the meta-analysis.

Characteristic		Author (year)
		Ageno et al[9]	Yiu et al[10]	Zhang et al[11]
Region		Italy, Mexico	Hong Kong, China	China
Study design		Randomized, controlled trial	Randomized, controlled trial	Randomized, double-blind, placebo-controlled trial
Sample size		59	102	80
No. of participants	Interventions	29	57	40
	Comparators	30	45	40
Duration of study		4 ± 1 weeks	6 hours and 1 week later	3 months
Age (year)		64.50*	61.0 ± 1.93*	50.24 ± 7.69*
Male (%)		27	35	33
Characteristics of participants		-Patients with mechanical heart valves who were receiving warfarin therapy and who presented -INR values between 6.0 and 12.0	-Patients with mechanical heart valves receiving long-term warfarin -international normalized ratios (INRs) from 4 to 7	-Adult patients aged ≥ 18 years with bileaflet mechanical valve prostheses who were undergoing warfarin sodium treatment -INR values from 4.0 to 10.0
Comorbidity of participants		- Atrial fibrillation	N/A	- Hypertension - Prior stroke - Diabetes - Peripheral vascular disease - Coronary artery disease
Valvular type		- Aortic prosthetic valve(50.85%) - Aortic prosthetic valve(44.07%) - Mitral and aortic position(5.08%)	- Aortic prosthetic valve(24%) - Mitral prosthetic valve (60%) - Mitral and aortic position(16%)	- Aortic prosthetic valve (13%) - Mitral prosthetic valve(53%) - Mitral and aortic position(35%)
Treatment regimen		Oral administration of 1-mg vitamin K	Intravenous administration of 1-mg Vitamin K	Oral administration of 2.5 mg vitamin K
Comparison group		No treatment	Fresh frozen plasma (FFP) 1 U	Placebo
Effect size (95% CI)		1.03 (0.07–15.79)	0.80 (0.05–12.40)	1.00 (0.06–15.47)
Outcome measurement		- Mean INR day 0, 1 - Number of patients with INR 2.3 to 4.5 on day 1 - Number of patients with INR < 1.8 on day 1 - Major/minor bleeding - Thrombotic events	- Major bleeding - Thromboembolic stroke, other systemic thromboembolic events - Adverse reaction to treatment (anaphylaxis, fever, rash). - The INR at 6 hours and 1 week after treatment	Primary outcome - Percentage of patients that achieved an INR value of 1.5 to 2.5 on the day following treatment - The time necessary to achieve an INR < 2.5 Secondary outcome - The number of patients having INR values < 1.5 or > 2.5 on any day following drug administration - The variation in INR values after drug administration - The incidence of resistance to warfarin, as described by the mean of the final 2 INR values determined during the study period being < 1.5 - Adverse event incidence

^* Mean ± SD.

N/A = not available.

3.2. Quality assessment

Since all included studies were RCTs, RoB 2.0, and GRADE were used to analyze the risk of bias. The risk of bias was high in 1 study, concerned in 1 study, and low in 1 study, according to RoB 2.0 (Figure 1A and 1B, Supplemental Digital Content 3, http://links.lww.com/MD/H181). The quality of included studies according to GRADE criteria was moderate because of the serious imprecision in the pooled estimate (Table 3, Supplemental Digital Content 4, http://links.lww.com/MD/H182).

3.3. Thromboembolism

The forest plot for the risk of thromboembolism in 241 mechanical heart valve patients with supratherapeutic INR who were treated with low-dose vitamin K and comparators is shown in Figure 2. Using random-effect model, low-dose vitamin K was not associated with thromboembolism rate (RR = 0.94; 95% CI: 0.19–4.55; P = .94; I² = 0.0%). Heterogeneity among RCT trials was low.

Figure 2.

Forest plot showing risk ratio of thromboembolism, major bleeding, and minor bleeding in patients with mechanical heart valves, who have supratherapeutic INR receiving Vitamin K or placebo—use using random-effect model. INR = international normalized ratio.

3.4. Bleeding

The forest plot for the risk of major bleeding and minor bleeding in 241 and 139 patients who had mechanical heart valves, had supratherapeutic INR, and were treated with low-dose vitamin K is shown in Figure 2. Using random-effect model, low-dose vitamin K also was not associated with major bleeding rate (RR = 0.58; 95% CI: 0.07–4.82; P = .62; I² = 0.0%) and minor bleeding rate (RR = 0.60; 95% CI: 0.07–5.09; P = .64; I² = 45.0%). Heterogeneity among RCT trials was low.

3.5. Sensitivity and subgroup analysis

The results of the subgroup analysis are shown in Table 2. The data were stratified by age, sex, and types of valvular. The risk of thromboembolism in patients with mechanical heart valves who had supratherapeutic INR and were treated with vitamin K was consistent across all subgroups. In addition, sensitivity analysis by influence plot is shown in (Figure 2, Supplemental Digital Content 5, http://links.lww.com/MD/H183). The result of a meta-regression supported no association between selected variables (sample size, age, percentage of male participant, aortic prostatic heart valve, route of administration, and study duration) and the outcomes (Table 4, Supplemental Digital Content 6, http://links.lww.com/MD/H184).

Table 2.

Subgroup and sensitivity analysis.

Subgroup	All studies
	Pooled risk ratio (95% CI)	Heterogeneity
		I2 value (%)	P
Age
≥65 years	N/A	N/A	N/A
<65 years	0.94 (0.19, 4.55)	0.0	0.990
Sex
Male	N/A	N/A	N/A
Female	0.94 (0.19, 4.55)	0.0	0.990
Types of valvular
AVR	1.03 (0.07, 15.79)	N/A	N/A
MVR	0.89 (0.13, 6.21)	0.0	0.908

AVR = aortic valve replacement, CI = confidence interval, MVR = mitral valve replacement, N/A = not available.

3.6. Publication bias

The funnel plot was asymmetrical, suggesting the existence of the publication bias (Figure 3, Supplemental Digital Content 7, http://links.lww.com/MD/H185). However, the results from Begg’s test and Egger’s test were not statistically significant, suggesting that publication bias may have not affected this analysis (Figure 4, Supplemental Digital Content 8, http://links.lww.com/MD/H186, P = .090 and .134, respectively).

4. Discussion

According to the AHA/ACC guideline for the management of patients with valvular heart disease 2020,^[2] vitamin K administration with temporary cessation of vitamin K antagonist has been suggested in individuals with mechanical heart valves and INR of >5.0 who are not actively bleeding. This study did not find the association between the use of low dose vitamin K and the risk of thromboembolism thus, supporting the use of vitamin K in such patients. This finding agrees with a previous systematic review and meta-analysis that compared the rate of major bleeding in nonvalvular patients with supratherapeutic INR using oral anticoagulants.^[22] Moreover, the onset of action for vitamin K in warfarin reversal is approximately 8 to 40 hours,^[23] which is the time required for the synthesis of vitamin K-dependent coagulation factors in the liver. On the contrary, FFP immediately replaced coagulation factors. The rapid replacement of FFP can lead to the sudden reduction of INR which can lead to thrombosis.^[24] This might explain the nonassociation between the use of vitamin K and thrombosis. In fact, vitamin K can be beneficial over the use of FFP in some scenarios. For example, FFP can cause transfusion reactions.^[25] In addition, vitamin K can be an interesting alternative in patients with cardiac dysfunction since FFP can increase the risk of pulmonary edema.^[26]

Although vitamin K is an interesting treatment option, there is no consensus in doses use for treating patients with the overdosage of vitamin K antagonists in patients with mechanical heart valves. The study by Ageno et al^[9] and Yiu et al^[10] are published early and investigated the use of 1-mg vitamin K. This is different from the dose used by the newer study by Zhang et al^[11] which used 2.5 mg vitamin K. Moreover, in the study by Yiu et al,^[10] vitamin K was administered intravenously which is different from other studies which used oral vitamin K. In addition, Zhang et al^[11] provide more details on patients’ comorbidities including prior stroke, and coronary artery disease. The discrepancy in the dose used in among studies may have resulted in differences in the major/minor bleeding outcome.

There are some limitations worth mentioning in this study. First of all, the majority of the participants were Asian (182/241) which was highly homogeneous. This might limit the generalization of the result to other populations. Although there is no direct support for the racial difference in vitamin K response, plenty of evidence shows the difference in the baseline vitamin K among ethnicities.^[27,28] Therefore, more studies in other populations are warranted before the role of races in vitamin K response can be ignored. Second, there is an insufficient number of studies that answer this research question, reflected by the low number of included studies in our meta-analysis. In addition, the included studies were of moderate quality since most of the articles did not specify the methods for randomization, allocation, and blinding. These 2 factors can directly affect the quality of our meta-analysis. Third, the included studies fail to mention how they control the factors that affect INR including diarrhea, fever, and food/drug-VKA interactions.^[9–11] However, this study still has several implications. Our findings reinforce the use of vitamin K as an INR reversal agent in patients with mechanical heart valves with VKA overdose. Our study also suggests the scarcity of research on the risk of thromboembolism in vitamin K users. RCTs with a larger sample size that clearly specifies the process of randomization, allocation, and blinding or well-designed large cohort studies that observe the use of vitamin K or other anticoagulant reversal agents should be conducted to further ensure the safety of vitamin K as an antidote for supratherapeutic INR occurred in patients with mechanical heart valves. In addition, participants with diverse ethnicities should also be included.

In conclusion, using low-dose vitamin K in patients with mechanical heart valves and supratherapeutic was not associated with the risk of thromboembolism. This study also shows no association between low-dose vitamin K and bleeding. Further studies are required to confirm the benefits over the risk for the use of low-dose vitamin K to reduce the INR in patients with mechanical heart valves and supratherapeutic INR.

Author contributions

B.S., S.K., C.S., P.S., O.L., N.L., and S.S. contributed to the research idea and design. B.S. and S.K. created the search strategy. B.S., C.S., and P.S. screened titles, abstracts, and full texts. B.S., C.S., and O.L. contributed to data extraction. B.S., S.K., and N.L. contributed to quality assessment. B.S., S.K., and N.L. contributed to statistical analysis and interpretation of data. B.S. and N.L. wrote the first draft of the article. B.S., S.K., N.L., and S.S. edited the draft of the article. All authors contributed to the critical revision of the article for important intellectual content, approved and reviewed the final article.