A systematic review of the pharmacokinetic and pharmacodynamic interactions of herbal medicine with warfarin
A systematic review of the pharmacokinetic and pharmacodynamic interactions of herbal medicine with warfarin
Songie Choi 0 1
Dal-Seok Oh 0 1
Ui Min Jerng 0
0 Editor: Francesca Borrelli, Universita degli Studi di Napoli Federico II , ITALY
1 The K-herb Research Center, Korea Institute of Oriental Medicine , Daejeon , South Korea , 2 Clinical Research Division, Korea Institute of Oriental Medicine , Daejeon , South Korea
It was difficult to determine whether ten herbal medicines had significant effects on the
PKPD parameters of warfarin. Low quality of evidence, different compounds within and
The aim of this study was to systematically review data regarding pharmacokinetic
(PK)pharmacodynamic (PD) parameters from randomized controlled trials relating to
interactions between herbal medicines and warfarin.
Three electronic databases were searched to identify relevant trials. Two reviewers
independently performed the study selection and data extraction. The risk of bias and reporting
quality were also assessed independently by two reviewers using the Cochrane risk of bias
tool and the consolidated standards of reporting trials (CONSORT). Outcomes were
measured for all reported PK-PD parameters and adverse events.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
Funding: This study was funded by the Research
and development of concomitant administration on
combinatory effect of herbal and Western medicine
(No. K17252, grant recipient DSO) and
development of Korean Medicine contents for
clinical practice (No.K17124, grant recipient UMJ)
of the Korea Institute of Oriental Medicine. The
funders had no role in study design, data collection
and analysis, preparation of the manuscript, or
decision to publish.
Competing interests: The authors have declared
that no competing interests exist.
different compositions of the herbs, and methodological limitations of the crossover study,
which is a clinical study in which subjects receive a sequence of different interventions,
made it difficult to form conclusions. Additional studies that remedy these vulnerabilities are
necessary to verify these results.
Warfarin is the most common oral anticoagulant used for treating or preventing
thromboembolic disorders. It has a narrow range between therapeutic and toxic doses, suggesting that
warfarin should be administered after calculating the optimal dose.
Patients taking warfarin should also be aware of its interaction with other drugs and foods,
including herbal medicines, because the concomitant use of these agents might alter the
metabolism and action of warfarin, necessitating an adjustment to the dose of warfarin for its safe
and effective administration [
]. Close monitoring of the anticoagulant effect of warfarin is
recommended through the international normalized ratio (INR) in clinical practice [1±3].
Herbal medicines are often used by patients receiving anticoagulants. Nearly 40% of
patients with cardiovascular disease have used complementary and alternative medicine,
including herbal medicine, concomitantly with their prescribed medications [
]. Those who
used herbal medicine for health management perceived herbal medicine to be helpful for their
cardiac condition [
]. In fact, some herbs, such as ginger, ginkgo, and garlic, have antiplatelet
and anticoagulant activity [6±11]. However, the mechanism of action of herbal medicines is
difficult to study in vitro and in vivo because these medicines comprise complex mixtures of
various compounds [
], which may simultaneously exhibit multiple physiological activities.
Therefore, patients taking herbal medicine with warfarin are more likely to be exposed to
potential herb-drug interactions [
Previous studies have revealed some of the mechanisms of interaction of warfarin with
herbal medicines via clinical reports [
]. Each herb has a different chemical composition,
hindering generalizations about herb-warfarin interactions. Although narrative reviews on
herb-warfarin interactions are available [
], no study has systematically reviewed them
based on changes in pharmacokinetic (PK)-pharmacodynamic (PD) parameters.
The aim of this article was to systematically review clinical data, including PK-PD
parameters, from randomized controlled trials (RCTs) and to discuss interactions between herbal
medicines and warfarin.
Sources of information and search strategies
Clinical trials were searched for and retrieved from core electronic databases, including
PubMed, EMBASE, and CINAHL. The last search of the databases was performed in
Search terms consisted of text terms and controlled vocabulary, such as medical subject
headings (MeSH). Three types of search terms were used: warfarin-related terms, herb-related
terms, and interaction-related terms. Article type or study design-related terms were not
included in the search terms. The search strategy for PubMed is stated below. The search
terms for the two other databases were similar.
#1 Warfarin [MeSH Terms]
2 / 17
#2 Warfarin [Title/Abstract]
Two reviewers (SIC and UMJ) reviewed the titles and abstracts of the studies retrieved from
the electronic searches to identify studies that met the inclusion criteria. Disagreements were
resolved by discussion between the two reviewers or consultation with a third reviewer (DSO).
No language restriction was applied. The inclusion criteria were as follows:
· Type of study. All relevant RCTs that reported interactions between herbal medicines and
warfarin were included.
· Type of participant. Studies that evaluated subjects who received herbal medicine
concomitantly with warfarin were included.
· Type of intervention. Trials using warfarin alone or warfarin with placebo drug versus
warfarin with herbal medicine were included. An herb was defined as a product or an extract
originating from a single botanical source. The definition of herb included raw or
manufactured single or complex medicinal plants, plant extracts, and dietary supplements. However,
single or synthesized substances from plant material were excluded.
· Type of outcome measures. Studies that measured more than one PK or PD parameter for
herb-warfarin interactions were included. Because the inhibition of the metabolism of
Swarfarin is clinically more important than the inhibition of the metabolism of R-warfarin,
3 / 17
the PK of R-warfarin were not investigated in this study [
]. The PK parameters
included time to maximum plasma concentration (Tmax), maximum plasma concentration
at steady state (Cmax), apparent volume of distribution after extra vascular administration
(V/F), fraction of total drug unbound in plasma (fu), terminal half-life (T1/2), apparent
plasma clearance of drug after extra vascular administration (CL/F), and area under the
plasma concentration-time curve from zero to infinity (AUCinf). The PD parameters
included all outcomes that reflected the biochemical and physiological effects of warfarin on
the human body.
Data extraction and quality assessment
Data were extracted from the titles and abstracts of the searched studies independently by two
reviewers. The study selection and data extraction used standard eligibility inclusion criteria as
determined by two reviewers. The quality of methodology in all included studies was
independently assessed according to the Cochrane Collaboration's seven criteria: 1) random sequence
generation, 2) allocation concealment, 3) blinding of participants and personnel, 4) blinding
of outcome assessment, 5) incomplete outcome data, 6) selective reporting, and 7) other bias
(defined as baseline data comparability). For each domain, the evaluation was denoted as low
risk, high risk, or unclear risk, according to the description of the methods used in each study.
We also assessed the reporting quality of all included studies based on the Consolidated
Standards of Reporting Trials (CONSORT) [
]. We used the CONSORT 2010 checklist and
the extension of the CONSORT statement simultaneously for trials of herbal medicinal
interventions. The Consort 2010 checklist is the latest version for assessing reporting quality. The
elaborated CONSORT statements for trials of herbal interventions enhance the checklist items
regarding the relevance to trials of herbal interventions [
Study design and herbal medicines were analyzed among the included studies. Results of PK
or PD parameters and type and proportion of adverse events in concurrent use of herbal
medicine and warfarin groups were compared with those in warfarin alone groups to identify
whether herbal medicine significantly affected the PK or PD parameters of warfarin.
Quantitative data synthesis was planned in a meta-analysis when the study design, type of herbal
medicine, and outcomes of the included studies were homogeneous; otherwise, we suggested results
in a narrative synthesis without meta-analysis [
Description of included studies
The search generated a total of 4437 potentially relevant studies; 295 duplicate and 4065
irrelevant studies were excluded by screening the titles and abstracts. Of the remaining studies, 77
full-text articles were reviewed and 9 studies [21±29] met our eligibility criteria. The PRISMA
diagram of the search process and study selection is presented in Fig 1.
Of these trials, three [23±25] were conducted in the United States, four [
21, 22, 26, 28
conducted in Australia, and one  was conducted in Korea. One trial [
] was conducted in
China and published in Chinese. Four studies [
21, 22, 26, 28
] had a three-way cross-over
randomized design and compared two different herbal medicine plus warfarin groups with a
warfarin alone group in healthy subjects. Three studies [
23, 25, 29
] used a placebo-controlled
parallel design and compared the concomitant administration of herbal medicine and warfarin
with the concomitant administration of placebo and warfarin. Two studies [
] used a
4 / 17
Fig 1. PRISMA flow diagram for selecting related articles.
5 / 17
double-blind crossover design. The study sample sizes ranged from 7 to 48 and a total of 160
subjects were involved in the nine trials. Thirty participants were reported to have dropped
out of the nine studies. The key data from the included RCTs are summarized in Table 1.
Nine herbal medicines were identified in the included studies: Panax ginseng, Panax
quinquefolius, Allium sativum, Gingko biloba, Vaccinium macrocarpon, Hypericum perforatum,
Echinacea angustifolia, Echinacea purpurea, and Zingiber officinale. Ginseng [
21, 23, 27
administered in three studies, but three ginsengs that have different scientific names were
used: Korean ginseng root (Panax ginseng) [
], American ginseng root (Panax quinquefolius)
], and Korea red ginseng (steamed Panax ginseng) [
]. Garlic (Allium sativum)was
administered in two studies, but each study used garlic manufactured with a different process [
]. One study  used an aged garlic product that was made by soaking raw garlic in
ethanol, whereas the other study [
] used an enteric-coated garlic tablet. Gingko (Gingko biloba)
] and cranberry (Vaccinium macrocarpon) [
] were used in two studies and St
John's wort (Hypericum perforatum) [
], echinacea (Mixture of Echinacea angustifolia and
Echinacea purpurea) [
], and ginger (Zingiber officinale) [
] were administered in one study.
In addition, policosanol [
] was mentioned in one study. However, policosanol is a complex
mixture of fatty alcohols derived from sugar cane wax and was not included in the inclusion
criteria. Extraction and formulation method, composition, and bioanalytical data regarding
the herbal preparations from the included RCTs are summarized Table 2.
Risk of bias assessment
No study had a low risk of bias in all seven domains. For random sequence generation, two
] of the nine studies (22%) used a random table, whereas the other studies (78%) did not
report a specific method of random sequence generation. For allocation concealment, one
] (11%) used an opaque envelope method, whereas the other studies (89%) did
not report any information about concealment. For blinding, four studies [23±25, 27] (44%)
used a double-blinding method by blinding participants and researchers and four studies [
22, 26, 28
] had an open-label design. One study [
] did not provide information about
blinding. For incomplete outcome data, seven trials [21±24, 26, 28, 29] reported detailed
information regarding attrition by describing the number and reasons for withdrawal. For selective
outcome reporting, only one study [
] presented the clinical trial identifier number, whereas
the other trials did not report registration information. Therefore, we could not compare the
protocols and trial reports. Information for other risks of bias was not reported in the studies,
except for one study [
] that was at high risk. The risk of bias assessment information is
presented in Figs 2 and 3.
Based on the two CONSORT statements, the reporting percentage for each of the articles
ranged from 36.5% to 61.5% and the mean percentage for all articles was 45.6%. All RCTs
described the eligibility criteria, participant flow, and interpretation of the results. No trials
provided information about the qualitative testing of interventions, periods of recruitment, or
follow up. Detailed results are presented in the S1 Table.
Because the two CONSORT statements were developed with the aim of evaluating and
reporting quality for parallel design and two treatment groups, they may be insufficient to
evaluate and reflect the characteristics of the crossover design. However, most of the items in the
CONSORT checklist apply to all trial designs because they reflect the characteristics of RCTs
6 / 17
m o )
ta tr (6
e n 2
r co r 1
y - e
a o v
w b o
- e s
o c s
w la ro
T p c
-ew voe )
re s 1
h s (
t o 2
l, rc 1
,] o ,
7 ic a d )
[2 lb re lin r (4
e u o -b e 5
e p K le vo 2
L e b s
R u s
26 lia trh tn )
l[u tras l,eb tem r (12
d u la ea voe 12
b A - r
A n t s
e y s
p a ro
O w c
,] d s r
[42 ite te -e ve )7
lb , o (
i n ta d ss 7
L U S u
o iln ro
D b c
,] te ,
3 a d
2 t n )
[ S li
-b - d (9
n d o lle 0
au ite le b ro 2
Y n ub ce t
U o la o
D p c
trh re )
l, v 1
e so (
b s 1
a f e ts
h n o e la c
C ,ro ry ig r h m )s je
s ) eb (sT f t b
.1 thu tnu ed TC um tec reo jceb fsu
leb ts1a oC tud ( n j
eh sbu u p
m s e
a e s
v m n
ca ce ie
d la a
r p p
ia t n
tr llia ite
A r a
, t,n tr , s
a n e
iss ide he ito lv
o c r la va
b ca la ilr tr )
itrvenhom l-rvscaua il,ssvvoau ili,tfreaab ittchehea ittspaen (56eanM
ep rbo b s s
e e om e p
D r r is r
ce th d o
the am fra
on ydn ew
ts o s
c c o
e a d
f m le
o h in
N p s
[2 rry ium rpa
i e in c
L b c ro
n c c
)n qu de r ito g
o li ra fo p in
e kgaed -odog lityau aunm snuom ckaag
raC (aV am DN icJu DN caP seY ·F q h c ·P
u G u
Y a q g e
irce xna iindn ne ne lspu
r o o a o
G N N C N
su .4 in iz w
p s 0 g rh po
f e r
g eg om de
G (Z ffo ND ND ND aC Y
/L s S ve /g
g id fo it
l c 7 l
305 sdo iton ,ea :.14 itcua irng
ing tce tra ien und ce tcu s
ign t-regm soue d itonan ftrxea conne lltycys opom rhaam ood fanua itrccae
ilc no uq lio se ·C o ·C a c ·P G M P
S L A S Y
y b s 2 le lf m b
D r D a e ·
N D N T Y
H g c l & lo
.2 troh ilra iiftc )e issen iltrae trxae ten iltaon e ircae tcu tsen trno
leb tsa1u teaM i(cenS anm rrceop faom feop lvoS ruom typ omm rdop ittsun ltyaC
a y F C o u
b 0 %
R .2 :
l d .6
a e 0
to de id :2 ;R :1 )
t i s b e %
f s % R 5
o o on ;R 1 ; 3
n e 6 .
% e s
% .0 2 :1
9 sn in 93 :
.1 i c .4 g
5 g (G .1 R 0 R
a og .4 nd
b k 2 a
ilo ign ,se se
b f d lid
o o i
g g o g
n m c k
iG .6 lyg ign lide
f 9 n f a
o ,f o o b
g a va g lio
n sa 1g
a t s R
n e e
a oo fo isd id
f g g o o
o n m ne n
g se 3 s s
8 / 17
Fig 2. Risk of bias summary. Review of authors' judgments about each risk of bias item for all nine included
studies. Plus (+) marked circle, Low risk of bias; Question (?) marked circle, Unclear risk of bias; Minus (-)
marked circle, High risk of bias.
9 / 17
Fig 3. Risk of bias graph. Review of authors' judgments about each risk of bias item presented as percentages across all included studies.
rather than those of parallel design. Therefore, the CONSORT statements were used to assess
the reporting quality of all the included trials.
There are many different methods of measuring outcome parameters. Because the aim of this
review was to assess herb-warfarin interactions, various types of PK-PD parameters were
evaluated (S2 Table).
Five trials [
21, 22, 26, 28, 29
] reported PK data based on the absorption, distribution,
metabolism, or elimination of warfarin when herbal medicine was co-administered with
warfarin. There were three herbs that affected the PK of warfarin in healthy subjects. St John's
] increased S-warfarin clearance and reduced R-warfarin clearance. Echinacea [
increased the apparent clearance of warfarin. One study [
] reported that ginkgo significantly
increased Cmax, AUCinf, and T1/2 and decreased the CL/F of warfarin, whereas another study
] reported that ginkgo did not markedly change the PK parameters of warfarin. There were
no significant changes in the PK parameters of warfarin when ginseng [
], ginger [
], and cranberry [
] were co-administered. Garlic and cranberry did not affect the
S-warfarin clearance in subjects with different genotypes of cytochrome P450 2C9 (CYP2C9) [
Whereas garlic increased the half maximal effective concentration (EC50) of S-warfarin in the
subjects with the CC genotype of vitamin K epoxide reductase complex (VKORC1), cranberry
decreased the EC50 of S-warfarin in the subjects with the CT or TT genotype of VKORC1 [
The PD parameters of warfarin were reported when ten herbal medicines, including
ginkgo, ginger, ginseng, St John's wort, echinacea, cranberry, Korean red ginseng, American
ginseng, garlic, and aged garlic were co-administered with warfarin. The AUC of INR for time
to treatment was used in six studies [21±23, 26, 28, 29], INR max [
] was evaluated in two
studies, and INR change was reviewed in two studies [
]. Five studies [
21, 22, 26, 28, 29
used area under concentration-time curves until the last concentration observation (AUCobs)
as an outcome measure. Platelet aggregation [
], peak INR change , vitamin K intake
], prothrombin time [
], and incidence of hemorrhage [
] were also included as PD
10 / 17
parameters. The results of two studies [
] indicated that co-administration of herbal
medicine altered the PD parameters of warfarin. The results of one study indicated that cranberry
significantly increased the area under the INR-time curve when administered with warfarin in
healthy subjects . However, the results of another study indicated that cranberry did not
markedly change INR values in patients with atrial fibrillation [
]. American ginseng also
reduced the anticoagulant effect of warfarin [
]. There were no significant changes in the PD
parameters of warfarin when St John's wort [
], ginseng [
], Korea red ginseng [
], ginger , garlic [
], aged garlic [
], and echinacea [
] were co-administered
All trials evaluated adverse events (AEs). As shown in Table 1, AEs were reported for three
studies and none of the events was major. Among these events, one study [
] reported that
one subject experienced gastrointestinal side effects, including constipation, during the first
two days of ginkgo pre-treatment and mild diarrhea during the first two days of ginger
pretreatment. One study [
] reported that three subjects experienced changes in sleeping habits
during St John's wort treatment and one study [
] reported rashes in two subjects using
cranberry-warfarin. In studies including garlic, one subject had evidence of nasal bleeding and one
subject reported lip dryness.
In this review, the interaction between ten herbs and warfarin, as indicated by changes in the
PK and PD parameters of warfarin, was analyzed based on published evidence. We assessed
the methodological quality of RCTs using the Cochrane risk of bias tool and CONSORT and
the results of quality assessments were reflected in the interpretation of this study. Two herbs
(St John's wort and echinacea) affected the PK parameters of warfarin, whereas four herbs
(ginseng, ginger, garlic, and cranberry) did not. There were conflicting results as to whether
ginkgo affected the PK parameters of warfarin. American ginseng changed the PD parameters
of warfarin, but eight herbal medicines (St John's wort, ginseng, Korea red ginseng, ginkgo,
ginger, garlic, aged garlic, and echinacea) did not. There were mixed results as to whether
cranberry changed the PD parameters of warfarin. There was a low risk of AEs after
co-administration of herbs and warfarin. However, most of the included studies had low reporting
quality and a crossover design that was unsuitable for meta-analysis. There were also inconsistent
results from several studies that used the same herbal medicine.
The use of herbal medicine is rapidly expanding and many reports have raised concerns
about possible herb-drug interactions. Herb-drug interactions may be categorized as either PK
or PD interactions. PK interactions include changes in absorption, distribution, metabolism,
and elimination [
]. PD interactions result from synergistic, additive, or antagonistic effects
of herbs when co-administered with drugs. Well-aligned PK-PD data provides information
regarding clinical efficacy and safety outcomes and guides the selection of doses and dosing
schedules for clinical trials [
All included studies investigated the PD interactions of warfarin with herbal medicines, but
different parameters were used in each study. Only four studies reported PK interaction
parameters between an herbal medicine and warfarin, whereas the five remaining studies did
not measure PK parameters. There were inconsistent results among those that used the same
herbal medicine. Two studies that evaluated the interaction between cranberry and warfarin
reported contradictory PD effects, possibly because one study [
] investigated healthy
subjects and the other [
] included patients with atrial fibrillation. The results of two studies that
11 / 17
assessed the PK-PD interactions between warfarin and ginkgo also differed. One study [
reported no significant differences in PK-PD parameters, whereas another [
] observed that
ginkgo had limited effects on PK parameters. These studies used different types of clinical
designs and subject conditions. Three studies [
21, 23, 27
] investigated the interaction between
ginseng and warfarin and the results differed according to the type of ginseng used (Korean
ginseng root (Panax ginseng), American ginseng root (Panax quinquefolius), or Korea red
ginseng (steamed Panax ginseng)). Korean ginseng and American ginseng have different
ginsenoside profiles [
] and Korea red ginseng contains converted ginsenosides transformed from
the ginsenosides in fresh ginseng [
]. The heterogeneous composition of compounds among
the three types of ginseng might have led to different results.
Several case reports have pointed to the risk associated with concomitant herb and warfarin
use. There were two case reports of an increased INR after co-administration of warfarin with
cranberry juice [
]. Interactions between ginseng and warfarin were also mentioned in
one case report . Two case reports suggested that a warfarin-St John's wort interaction was
associated with a change in INR [
]. These relevant case reports indicated a potential
herb-warfarin interaction, but it was difficult to identify a causal relationship. Suspected
herbwarfarin interactions are primarily limited to anecdotal case reports. In addition, these case
reports did not provide sufficient information about the patients' medical records and
compounding factors may have existed, such as administration of other medications, dietary
supplements, foods, or alcohol intake. Some studies pointed out such limitations, that is, that case
reports often result in misleading conclusions for multiple reasons [
]. There have been
previous experiments and clinical studies of platelet aggregation caused by herbs. Several
studies have suggested that herbal constituents may affect PK-PD and alter the anticoagulant and
platelet aggregation effects of warfarin [
Other studies have shown conflicting results as to whether herb-warfarin interactions were
associated with increased risks. Garlic and ginger are known potent inhibitors of platelet
]. One review article reported that spontaneous bleeding occurred during the
concurrent use of warfarin and these herbs. Conversely, some cases did not indicate a
significant inhibition of platelet function [
]. One in vitro study suggested that gingko contributed
to the altered platelet aggregation [
]. In contrast, a clinical study confirmed that gingko did
not change platelet aggregation. Furthermore, it was difficult to determine whether the
combined use of warfarin and herbs led to increased platelet aggregation [
Warfarin is predominantly metabolized via CYP2C9 and changes in CYP2C9 may
significantly alter the PK-PD parameters of warfarin [
]. However, an in vivo study indicated
that echinacea did not significantly affect the metabolism of drugs metabolized by CYP2C9
. Another in vivo study reported that gingko induced CYP enzyme activity in a
dosedependent manner, but did not cause hepatic damage [
]. Two clinical trials also evaluated
the effects of ginkgo in healthy volunteers and the results indicated that warfarin
concentrations did not significantly change with concomitant administration of ginkgo [
results implied that concerns regarding increased hemorrhagic complications resulting from
an herb-warfarin interaction were unfounded. This study also indicated that herbal medicine
might not lead to a clinically significant change in the PK-PD parameters of warfarin.
Furthermore, herbal medicine did not significantly alter the anticoagulant effects of warfarin and no
severe AEs were reported.
There were four limitations of the current review. First, the rationale for determining the
washout period in each study was lacking. Several studies adopted a crossover design, which
has been used by many researchers to investigate potential drug-drug interactions. Because a
crossover trial carries the risk of a carry-over effect, trials with a crossover design should use a
sufficient washout period. Otherwise, the effect of the first period treatment may persist into
12 / 17
the subsequent period [
]. The average washout period in seven studies was two weeks and
one study did not mention the washout period. To minimize the carry-over effect, the washout
period in a crossover study should be at least five times the half-life of the drug [
]. The mean
range of T1/2 was 29.2 to 38.7 hours for the control group and 27.2 to 76.6 hours for the
experimental group. The washout period was calculated to be 6±8 days for the control group and
6±16 days for the experimental group based on these results. These results included studies for
which it was difficult to determine if the washout period was sufficient.
Second, clinical data are lacking in order to provide synthesized evidence for herb-drug
interactions from crossover trials. We attempted to conduct a meta-analysis using five trials
that assessed the effects of herbs combined with warfarin versus warfarin alone or warfarin
combined with placebo. A meta-analysis for crossover trials can be conducted if one of the
following three measurements is available: 1) individual subject data, 2) the mean and standard
deviation (or standard error) of the subject-specific differences between the experimental
group and control group, 3) the mean difference and variables from a paired t-test (28).
However, no studies were reported these data. Therefore, it was impossible to synthesize the data
into a meta-analysis.
Third, the reporting quality of the included studies was poor based on CONSORT 2010 and
the extension of the CONSORT statement for trials of herbal medicine interventions. This
CONSORT extension enhances the checklist items regarding the relevance of herbal
interventions to trials [
]. Several studies did not describe some of the items in this elaborated
statement. For example, some studies autonomously prepared the intervention drug or herbal
medicine without a quality control, making it difficult to report the characteristics of the herbal
product, including the concentration of the extraction solvent, the method of the
authentication of raw material, fingerprinting, and standardization. The majority of studies also did not
discuss randomization, including sequence generation, allocation, and concealment. The four
crossover design studies did not report how the treatment group and control group were
crossed. Therefore, the response rate for each article was less than 50%.
Finally, all included RCTs focused on changes in the PK-PD parameters of warfarin and
not those of the herbal medicine. Each single herb has a variety of biochemical compounds
and the composition of components in the herb could vary upon cultivation, delivery, and
product manufacturing conditions. These uncertainties and complexities make it difficult to
determine standards for the PK-PD parameters of herbal medicines. Therefore, this study
could not include PK-PD evaluation of herbal medicines themselves and did not reflect the
purpose of the administration and intention for use of the herbal medicine.
Further studies evaluating interactions between warfarin and herbal medicines should avoid
the limitations mentioned in this study. The details of the information for a trial should be clearly
described and fully reported. The crossover design is a common study design, but is
inappropriate for obtaining valid evidence through meta-analysis. We recommend using high-quality RCTs
to confirm herb-warfarin interactions. In addition, to ensure the quality of the clinical trials,
randomization and allocation concealment procedures should be performed to minimize bias.
Although there are some limitations, to our knowledge, this study was the first systematic
review of the clinical outcomes and PK-PD effects of herb-warfarin interactions. We
deliberately selected related studies after searching a wide range of databases and this analysis was
based on available clinical trials that evaluated herb-warfarin interactions.
It was difficult to decide whether ten herbal medicines significantly affected the PK-PD
parameters of warfarin. Low quality of evidence, herbal uncertainties and complexities of different
13 / 17
compounds and their compositions, and methodological limitations of the crossover study
made it difficult to form conclusions. Further studies with an appropriate study design and
reporting quality are necessary to verify herb-warfarin interactions.
S1 Table. Reporting quality of RCTs based on the consolidated standards of reporting
S2 Table. Pharmacokinetic and pharmacodynamic parameters of the included studies
Conceptualization: Songie Choi, Dal-Seok Oh, Ui Min Jerng.
Data curation: Songie Choi, Ui Min Jerng.
Formal analysis: Songie Choi, Dal-Seok Oh, Ui Min Jerng.
Funding acquisition: Dal-Seok Oh, Ui Min Jerng.
Investigation: Songie Choi, Ui Min Jerng.
Methodology: Dal-Seok Oh, Ui Min Jerng.
Project administration: Dal-Seok Oh.
Supervision: Ui Min Jerng.
Validation: Dal-Seok Oh, Ui Min Jerng.
Visualization: Songie Choi. Writing ± original draft: Songie Choi, Dal-Seok Oh. Writing ± review & editing: Ui Min Jerng.
14 / 17
15 / 17
16 / 17
1. Ge B , Zhang Z , Zuo Z. Updates on the clinical evidenced herb-warfarin interactions . Evid Based Complement Alternat Med . 2014 ; 2014 : 957362 . Epub 2014/05/03. https://doi.org/10.1155/ 2014 /957362 PMID: 24790635 .
2. Nutescu EA , Shapiro NL , Ibrahim S , West P. Warfarin and its interactions with foods, herbs and other dietary supplements . Expert Opin Drug Saf . 2006 ; 5 ( 3 ): 433 ± 51 . Epub 2006/04/14. https://doi.org/10. 1517/14740322.214.171.1243 PMID: 16610971 .
3. Nutescu EA , Shapiro NL , Chevalier A , Amin AN . A pharmacologic overview of current and emerging anticoagulants . Cleve Clin J Med . 2005 ; 72 Suppl 1 : S2 ± 6 . PMID: 15853173 .
4. Yeh GY , Davis RB , Phillips RS . Use of complementary therapies in patients with cardiovascular disease . Am J Cardiol . 2006 ; 98 ( 5 ): 673 ± 80 . https://doi.org/10.1016/j.amjcard. 2006 . 03 .051 PMID: 16923460 .
5. Sanjai KS , Aiswarya P . Warfarin interactions with complementary medicines, herbs, and dietary supplements . Journal of Chemical and Pharmaceutical Research . 2015 ; 7 ( 6 ): 71 ± 5 .
6. Rui TQ , Zhang L , Qiao HZ , Huang P , Qian S , Li JS , et al. Preparation and Physicochemical and Pharmacokinetic Characterization of Ginkgo Lactone Nanosuspensions for Antiplatelet Aggregation . J Pharm Sci . 2016 ; 105 ( 1 ): 242 ±9. https://doi.org/10.1016/j.xphs. 2015 . 10 .002 PMID: 26852855 .
7. Ryu KH , Han HY , Lee SY , Jeon SD , Im GJ , Lee BY , et al. Ginkgo biloba extract enhances antiplatelet and antithrombotic effects of cilostazol without prolongation of bleeding time . Thromb Res . 2009 ; 124 ( 3 ): 328 ± 34 . Epub 2009/04/08. https://doi.org/10.1016/j.thromres. 2009 . 02 .010 PMID: 19349067 .
8. Nicoll R , Henein MY . Ginger (Zingiber officinale Roscoe): a hot remedy for cardiovascular disease? Int J Cardiol . 2009 ; 131 ( 3 ): 408 ± 9 . Epub 2007/11/27. https://doi.org/10.1016/j.ijcard. 2007 . 07 .107 PMID: 18037515 .
9. Rodriguez M , Ringstad L , Schafer P , Just S , Hofer HW , Malmsten M , et al. Reduction of atherosclerotic nanoplaque formation and size by Ginkgo biloba (EGb 761) in cardiovascular high-risk patients . Atherosclerosis . 2007 ; 192 ( 2 ): 438 ± 44 . https://doi.org/10.1016/j.atherosclerosis. 2007 . 02 .021 PMID: 17397850 .
10. Lau AJ , Toh DF , Chua TK , Pang YK , Woo SO , Koh HL . Antiplatelet and anticoagulant effects of Panax notoginseng: comparison of raw and steamed Panax notoginseng with Panax ginseng and Panax quinquefolium . J Ethnopharmacol . 2009 ; 125 ( 3 ): 380 ± 6 . Epub 2009/08/12. https://doi.org/10.1016/j.jep. 2009 . 07 .038 PMID: 19665534 .
11. Yun YP , Do JH , Ko SR , Ryu SY , Kim JH , Song HC , et al. Effects of Korean red ginseng and its mixed prescription on the high molecular weight dextran-induced blood stasis in rats and human platelet aggregation . J Ethnopharmacol . 2001 ; 77 ( 2 ±3): 259 ± 64 . Epub 2001/09/06. PMID: 11535373 .
12. Yang Y , Zhang Z , Li S , Ye X , Li X , He K. Synergy effects of herb extracts: pharmacokinetics and pharmacodynamic basis . Fitoterapia . 2014 ; 92 : 133 ± 47 . https://doi.org/10.1016/j.fitote. 2013 . 10 .010 PMID: 24177191 .
13. Tsai HH , Lin HW , Lu YH , Chen YL , Mahady GB . A review of potential harmful interactions between anticoagulant/antiplatelet agents and Chinese herbal medicines . PLoS One . 2013 ; 8 ( 5 ): e64255 . Epub 2013 /05/15. https://doi.org/10.1371/journal.pone.0064255 PMID: 23671711 .
14. Izzo AA , Ernst E . Interactions between herbal medicines and prescribed drugs: an updated systematic review . Drugs . 2009 ; 69 ( 13 ): 1777 ± 98 . https://doi.org/10.2165/ 11317010 -000000000-00000 PMID: 19719333 . Language : English. Entry Date: 20091106 . Revision Date: 20110520 . Publication Type: journal article .
15. Hirsh J . Current anticoagulant therapyÐUnmet clinical needs . Thrombosis Research . 2003 ; 109 (SUPPL .):S1± S8 .
16. Jiang X , Blair EY , McLachlan AJ . Investigation of the effects of herbal medicines on warfarin response in healthy subjects: a population pharmacokinetic-pharmacodynamic modeling approach . J Clin Pharmacol . 2006 ; 46 ( 11 ): 1370 ± 8 . Epub 2006/10/20. https://doi.org/10.1177/0091270006292124 PMID: 17050802 .
17. Moher D , Hopewell S , Schulz KF , Montori V , Gotzsche PC , Devereaux PJ , et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials . BMJ . 2010 ; 340 :c869. https://doi.org/10.1136/bmj.c869 PMID: 20332511 .
18. Gagnier JJ , Boon H , Rochon P , Moher D , Barnes J , Bombardier C , et al. Recommendations for reporting randomized controlled trials of herbal interventions: Explanation and elaboration . J Clin Epidemiol . 2006 ; 59 ( 11 ): 1134 ± 49 . https://doi.org/10.1016/j.jclinepi. 2005 . 12 .020 PMID: 17027423 .
19. Gagnier JJ , Boon H , Rochon P , Moher D , Barnes J , Bombardier C , et al. Reporting randomized, controlled trials of herbal interventions: an elaborated CONSORT statement . Ann Intern Med . 2006 ; 144 ( 5 ): 364 ± 7 . PMID: 16520478 .
20. Ferrer P , Ballarin E , Sabate M , Vidal X , Rottenkolber M , Amelio J , et al. Antiepileptic drugs and suicide: a systematic review of adverse effects . Neuroepidemiology . 2014 ; 42 ( 2 ): 107 ± 20 . https://doi.org/10. 1159/000356807 PMID: 24401764 .
21. Jiang X , Williams KM , Liauw WS , Ammit AJ , Roufogalis BD , Duke CC , et al. Effect of St John's wort and ginseng on the pharmacokinetics and pharmacodynamics of warfarin in healthy subjects . Br J Clin Pharmacol . 2004 ; 57 ( 5 ): 592 ± 9 . Epub 2004/04/20. https://doi.org/10.1111/j.1365- 2125 . 2003 . 02051 .x PMID: 15089812 .
22. Jiang X , Williams KM , Liauw WS , Ammit AJ , Roufogalis BD , Duke CC , et al. Effect of ginkgo and ginger on the pharmacokinetics and pharmacodynamics of warfarin in healthy subjects . Br J Clin Pharmacol . 2005 ; 59 ( 4 ): 425 ± 32 . Epub 2005/04/02. https://doi.org/10.1111/j.1365- 2125 . 2005 . 02322 . x PMID : 15801937 .
23. Yuan CS , Wei G , Dey L , Karrison T , Nahlik L , Maleckar S , et al. Brief communication: American ginseng reduces warfarin's effect in healthy patients: a randomized, controlled Trial . Ann Intern Med . 2004 ; 141 ( 1 ): 23 ± 7 . Epub 2004/07/09. PMID: 15238367 .
24. Li Z , Seeram NP , Carpenter CL , Thames G , Minutti C , Bowerman S . Cranberry does not affect prothrombin time in male subjects on warfarin . Journal of the American Dietetic Association . 2006 ; 106 ( 12 ): 2057 ±61. https://doi.org/10.1016/j.jada. 2006 . 09 .012 PMID: 17126638 . Language : English. Entry Date: 20070202 . Revision Date: 20120302 . Publication Type: journal article .
25. Macan H , Uykimpang R , Alconcel M , Takasu J , Razon R , Amagase H , et al. Aged garlic extract may be safe for patients on warfarin therapy . J Nutr . 2006 ; 136 ( 3 Suppl) : 793S±5S. Epub 2006 /02/18. PMID: 16484565 .
26. Mohammed Abdul MI , Jiang X , Williams KM , Day RO , Roufogalis BD , Liauw WS , et al. Pharmacodynamic interaction of warfarin with cranberry but not with garlic in healthy subjects . Br J Pharmacol . 2008 ; 154 ( 8 ): 1691 ± 700 . Epub 2008/06/03. https://doi.org/10.1038/bjp. 2008 .210 PMID: 18516070 .
27. Lee YH , Lee BK , Choi YJ , Yoon IK , Chang BC , Gwak HS . Interaction between warfarin and Korean red ginseng in patients with cardiac valve replacement . Int J Cardiol . 2010 ; 145 ( 2 ): 275 ± 6 . Epub 2009/11/ 17. https://doi.org/10.1016/j.ijcard. 2009 . 09 .553 PMID: 19913311 .
28. Abdul MI , Jiang X , Williams KM , Day RO , Roufogalis BD , Liauw WS , et al. Pharmacokinetic and pharmacodynamic interactions of echinacea and policosanol with warfarin in healthy subjects . Br J Clin Pharmacol . 2010 ; 69 ( 5 ): 508 ± 15 . Epub 2010/06/25. https://doi.org/10.1111/j.1365- 2125 . 2010 . 03620 . x PMID : 20573086 .
29. Zhou Y , Zeng R . Effects of Ginkgo biloba extract on anticoagulation and blood drug level of warfarin in healthy wolunteers . Zhongguo Zhong Yao Za Zhi . 2011 ; 36 ( 16 ): 2290 ± 3 . Epub 2011/11/22. PMID: 22097347 .
30. Hussain MS . Patient counseling about herbal-drug interactions . Afr J Tradit Complement Altern Med . 2011 ; 8 ( 5 Suppl) : 152 ± 63 . https://doi.org/10.4314/ajtcam.v8i5S.8 PMID: 22754069 .
31. Roskos LK , Schneider A , Vainshtein I , Schwickart M , Lee R , Lu H , et al. PK-PD modeling of protein drugs: implications in assay development . Bioanalysis . 2011 ; 3 ( 6 ): 659 ± 75 . https://doi.org/10.4155/bio. 11.28 PMID: 21417734 .
32. Yuan CS , Wang CZ , Wicks SM , Qi LW . Chemical and pharmacological studies of saponins with a focus on American ginseng . J Ginseng Res . 2010 ; 34 ( 3 ): 160 ±7. https://doi.org/10.5142/jgr. 2010 . 34 .3.160 PMID: 21359120 .
33. Lee SM , Kim SC , Oh J , Kim JH , Na M. 20 (R)-Ginsenoside Rf : A new ginsenoside from red ginseng extract . Phytochemistry Letters . 2013 ; 6(4):620±4 . https://doi.org/10.1016/j.phytol. 2013 . 08 .002
34. Paeng CH , Sprague M , Jackevicius CA. Interaction between warfarin and cranberry juice . Clinical Therapeutics . 2007 ; 29 ( 8 ): 1730 ±5. https://doi.org/10.1016/j.clinthera. 2007 . 08 .018 PMID: 17919554 . Language : English. Entry Date: 20080314 . Publication Type: journal article .
35. Sylvan L , Justice NP . Possible interaction between warfarin and cranberry juice . Am Fam Physician . 2005 ; 72 ( 6 ):1000; author reply Epub 2005/09/30. PMID: 16190498 .
36. Janetzky K , Morreale AP . Probable interaction between warfarin and ginseng . Am J Health Syst Pharm . 1997 ; 54 ( 6 ): 692 ± 3 . Epub 1997/03/15. PMID: 9075501 .
37. Yue QY , Bergquist C , Gerden B . Safety of St John's wort (Hypericum perforatum) . Lancet . 2000 ; 355 ( 9203 ): 576 ±7. https://doi.org/10.1016/S0140- 6736 ( 05 ) 73227 -X
38. Barnes J , Anderson LA , Phillipson JD . St John's wort (Hypericum perforatum L.): a review of its chemistry, pharmacology and clinical properties . J Pharm Pharmacol . 2001 ; 53 ( 5 ): 583 ± 600 . Epub 2001/05/24. PMID: 11370698 .
39. Pham DQ , Pham AQ . Interaction potential between cranberry juice and warfarin . American Journal of Health-System Pharmacy . 2007 ; 64 ( 5 ): 490 ±4. https://doi.org/10.2146/ajhp060370 PMID: 17322161 . Language : English. Entry Date: 20070525 . Revision Date: 20120302 . Publication Type: journal article .
40. Zikria J , Goldman R , Ansell J . Cranberry juice and warfarin: when bad publicity trumps science . Am J Med . 2010 ; 123 ( 5 ): 384 ± 92 . https://doi.org/10.1016/j.amjmed. 2009 . 08 .019 PMID: 20399311 .
41. Fasinu PS , Bouic PJ , Rosenkranz B. An overview of the evidence and mechanisms of herb-drug interactions . Frontiers in Pharmacology . 2012 ; 3 APR.
42. Izzo AA , Di Carlo G , Borrelli F , Ernst E. Cardiovascular pharmacotherapy and herbal medicines: the risk of drug interaction . Int J Cardiol . 2005 ; 98 ( 1 ):1± 14 . Epub 2005/01/29. https://doi.org/10.1016/j. ijcard. 2003 . 06 .039 PMID: 15676159 .
43. Ramsay NA , Kenny MW , Davies G , Patel JP . Complimentary and alternative medicine use among patients starting warfarin . Br J Haematol . 2005 ; 130 ( 5 ): 777 ± 80 . Epub 2005/08/24. https://doi.org/10. 1111/j.1365- 2141 . 2005 . 05689 . x PMID : 16115136 .
44. Vaes LP , Chyka PA . Interactions of warfarin with garlic, ginger, ginkgo, or ginseng: nature of the evidence . Ann Pharmacother . 2000 ; 34 ( 12 ): 1478 ± 82 . Epub 2001/01/06. https://doi.org/10.1345/aph. 10031 PMID: 11144706 .
45. Koch E. Inhibition of platelet activating factor (PAF)-induced aggregation of human thrombocytes by ginkgolides: considerations on possible bleeding complications after oral intake of Ginkgo biloba extracts . Phytomedicine . 2005 ; 12 ( 1 ±2): 10 ± 6 . Epub 2005/02/08. https://doi.org/10.1016/j.phymed. 2004 . 02 .002 PMID: 15693702 .
46. Bal Dit Sollier C , Caplain H , Drouet L . No alteration in platelet function or coagulation induced by EGb761 in a controlled study . Clinical and laboratory haematology . 2003 ; 25 ( 4 ): 251 ± 3 . Epub 2003/08/ 02. PMID: 12890165 .
47. Gorski JC , Huang SM , Pinto A , Hamman MA , Hilligoss JK , Zaheer NA , et al. The effect of echinacea (Echinacea purpurea root) on cytochrome P450 activity in vivo . Clin Pharmacol Ther . 2004 ; 75 ( 1 ): 89 ± 100 . https://doi.org/10.1016/j.clpt. 2003 . 09 .013 PMID: 14749695 .
48. Umegaki K , Saito K , Kubota Y , Sanada H , Yamada K , Shinozuka K. Ginkgo biloba extract markedly induces pentoxyresorufin O-dealkylase activity in rats . Jpn J Pharmacol . 2002 ; 90 ( 4 ): 345 ± 51 . Epub 2002/12/26. PMID: 12501011 .
49. Li T , Yu T , Hawkins BS , Dickersin K. Design , Analysis, and Reporting of Crossover Trials for Inclusion in a Meta-Analysis . PLoS One . 2015 ; 10 ( 8 ):e0133023. https://doi.org/10.1371/journal.pone.0133023 PMID: 26284684 .
50. Dhariwal K , Jackson A . Effect of length of sampling schedule and washout interval on magnitude of drug carryover from period 1 to period 2 in two-period, two-treatment bioequivalence studies and its attendant effects on determination of bioequivalence . Biopharm Drug Dispos . 2003 ; 24 ( 5 ): 219 ± 28 . https://doi.org/10.1002/bdd.359 PMID: 12784322 .