Introduction
Bleeding in patients undergoing high-risk gastrointestinal endoscopy while on antithrombotic
therapy remains a major concern. Data on the need to interrupt direct oral anticoagulants
(DOACs) before an endoscopic procedure are scarce. Current guidelines recommend discontinuation
tailored to the bleeding risk of the procedure versus the thrombotic risk of the patient
based on low-quality evidence. Recent studies highlight the risk of thrombotic events
associated with transient interruption of anticoagulants with a 30-day incidence of
up to 5.4 % [1]. Early evidence suggests it may not be necessary to withhold anticoagulants for
certain therapeutic procedures such as small colorectal polypectomies [2]. Bleeding from endoscopic ultrasound-guided tissue acquisition (EUS-TA) is infrequent
occurring in 0.1 % of cases [3], and usually mild and self-limiting.
There is a lack of data on bleeding outcomes when DOAC therapy is uninterrupted with
EUS-TA. We hypothesized that EUS-TA performed without discontinuing DOACs would not
be associated with an increased risk of significant bleeding. We conducted a randomized
study in a swine model aimed to assess the safety of EUS-TA without interruption of
DOACs.
Methods
Study design
This randomized controlled pilot study was performed according to ARRIVE 2.0 recommendations
for animal studies. All experiments followed the American Association for Laboratory
Animal Science guidelines and the study protocol was approved by the Institutional
Animal Care and Use Committee (IACUC protocol #A00005862–21).
Twenty domestic pigs (weight 35–40 Kg) were randomized (1:1) to receive anticoagulant
treatment with oral apixaban (0.5 mg/Kg/12 h) or placebo. Treatment was administered
3 days before EUS-TA of the pancreas, the day of the procedure, and three days after
the procedure. The study design and timeline are detailed in [Fig. 1].
Fig. 1 Study design.
Procedure
All animals underwent EUS fine-needle biopsy (EUS-FNB) of the pancreas tail using
a 22G needle (Sharkore, Medtronic, Minnesota, United States) using the same approach
of 4 passes with 10 to-and-fro movements and 3 to 5 cc of negative pressure. Two each
of the EUS-FNB samples were submitted to cytologic and histologic evaluation. Smear
slides for cytology were prepared for rapid on-site evaluation (ROSE) while specimens
for histology were placed in formalin. Endoscopists performing EUS-TA and necropsy,
and pathologists were blinded to the treatment arm. All procedures were performed
with a linear echoendoscope (GF‐UCT260, Olympus, Tokyo, Japan). Detailed information
is displayed in Supplementary material.
Definitions and outcomes
The primary endpoint was occurrence of clinically significant bleeding events (CSBEs),
defined as a composite outcome of drop in hemoglobin level ≥ 2 g/dL [4]
[5] and evidence of intraprocedural or post-procedure bleeding. Intraprocedural extraluminal
bleeding on EUS was defined as a new hypoechoic area at the site of needle puncture.
Post-procedure bleeding was defined as clinical signs of blood loss during the survival
period and/or hematoma documented at necropsy. Bleeding at the puncture site without
a drop in hemoglobin of ≥ 2 g/dL was considered a minor adverse event.
The secondary endpoint was the adequacy of acquired specimens defined as confirmation
of pancreas tissue by cytopathological evaluation (Supplementary Material).
Statistical analyses
Randomization sequence using blocks was applied. Categorical variables are displayed
as absolute and relative frequencies. Intergroup comparisons were made using the Fisher´s
exact test. Quantitative variables are reported as mean and standard deviation or
as median and interquartile range (IQR), when appropriate. Comparisons for independent
continuous variables were performed using the Mann-Whitney U test. All analyses were
two-tailed and significance was set at P < 0.05. A sample size of 10 pigs per arm provided an E value (E value = total number
of animals-total number of groups) of 18 and allowed non-parametric tests (further
information on sample size calculation is detailed in Supplementary Material).
Results
Study population
Twenty pigs (55 % females; mean weight 38.1 Kg, standard deviation 1.7) were randomized
to oral apixaban (n = 10) or placebo (n = 10). Median hemoglobin before the procedure
was 10.9 g/dL (IQR, 10.2–11.2). Median apixaban concentration on day of EUS-FNB in
treatment pigs was 365.15 ng/mL (IQR, 252.5–484.8) with a median time between the
last apixaban dose and blood draw of 2.2 hours (IQR, 1.08–3.25). Two pigs had an apixaban
concentration > 500 ng/mL (on therapy range in humans: 20 to 500 ng/mL [6]).
Procedure and follow-up
EUS-FNB sampling of the pancreas body/tail was successfully performed in all 20 pigs
(four samples/pig) ([Fig. 2]). A total of 80 tissue samples were obtained.
Fig. 2 a Cartoon of pig pancreas anatomy. b Endoscopic ultrasound-guided fine-needle biopsy of the pancreas body/tail.
Twenty-four hours post-procedure, one animal from the apixaban group presented with
behavioral changes of reduced activity and poor oral intake which subsequently improved
within 24 hours with intravenous analgesia and liquid diet.
Endoscopic ultrasound-related bleeding
Clinically significant bleeding
Only one animal in the apixaban group presented with CSBE (10%; 95 % confidence interval
[CI] 1.8–40.4 % versus 0 %; 95 % CI 0–27.7 % in the placebo group). This animal experienced
a drop in hemoglobin level ≥ 2 g/dL (3.4 g/dL) and a large hematoma (120 × 11 mm)
at the puncture site noted at necropsy ([Fig. 3]). This pig had an apixaban concentration above the upper range (> 500 ng/mL) and
was the same animal that presented with post-procedure lethargy and poor oral intake.
No intraprocedural extraluminal bleeding was observed during EUS in this animal.
Fig. 3 Large hematoma in pig that experienced a clinically significant bleeding event.
Median drop in hemoglobin was similar between both groups, Supplementary Fig. 1.
Non-clinically significant bleeding
Minor extraluminal bleeding observed at EUS occurred in five pigs (25 %), of which
four were from the apixaban group (P = 0.3). Bleeding manifested in the form of a new hypoechoic area at the puncture
site (median size 5.5 mm [IQR, 4.8–6.3]), all of which were self-limited ([Fig. 4]; Supplementary Table S1).
Fig. 4 Intraprocedural minor bleeding at the puncture site.
Hematoma at necropsy was observed in 10 pigs (50 %) (median size 17.5 mm [IQR, 11–20]),
six of which occurred in the apixaban group (P = 0.65) ([Fig. 5]; Supplementary Table S1). Four of the 10 animals that had a hematoma at necropsy had documented intraprocedural
extraluminal bleeding.
Fig. 5 Non-clinically significant hematoma at the puncture site in animal on a apixaban and b placebo.
Cytopathological assessment
An adequate specimen for cytological evaluation and confirmation of pancreatic tissue
was acquired in all 20 animals (100 %) (median number of smear slides 5 [IQR, 4–7]).
An adequate specimen for histological assessment and confirmation of pancreatic tissue
was acquired in 19 /20 (95 %) animals. Core area median size was 4.6 mm2 (IQR, 1.2–4.6), and pancreatic tissue represented 6 % (IQR, 1.6–2.5) of the core
area ([Fig. 6], Supplementary Table S1).
Fig. 6 a, b Cytology and c, d pathology images confirming pancreatic tissue.
Discussion
Bleeding related to EUS-TA is rare and the risk in patients treated with antithrombotic
drugs is unknown. Adverse events associated with transient interruption of anticoagulants
may outweigh the potential hemorrhagic complications. In our pilot preclinical trial
with a limited sample size EUS-TA on apixaban did not significantly increase the occurrence
of CSBE or prevent adequate cytopathological evaluation.
Only one animal with levels of apixaban above the upper limit of detection experienced
CSBE. Of note, the majority of animals with a hematoma at necropsy did not have documented
intraprocedural bleeding highlighting that bleeding may be missed, may occur beyond
the endosonographic field of view, or result from delayed hemorrhage. Although more
animals in the apixaban group had minor bleeding events, there was no significant
difference between the groups. Furthermore, the difference in the median drop of hemoglobin
post-procedure between the groups was similar.
Apixaban was selected being the most frequently prescribed DOAC in clinical practice
[7]. Oral apixaban has been used in swine in two prior studies that demonstrated lower
bioavailability, shorter half-life, and a larger volume of distribution compared to
humans, and thus requirement of higher doses [8]. Specifically, a dosage of 0.5 mg/Kg/12 h was found to maintain an adequate anti-factor
Xa activity [8]. Apixaban has been associated with a lower overall bleeding risk [9], with a lower incidence of upper gastrointestinal bleeding and post polypectomy
bleeds compared to other DOACs. We determined apixaban blood concentration before
EUS-TA for the purposes of the study, even though apixaban is generally given at a
fixed dose without monitoring in clinical practice, to ensure animals had increased
anti-factor Xa activity relative to a sham.
The porcine pancreas was the target puncture site as solid pancreatic lesions are
the most common indication for EUS-TA. In this pilot study we sampled normal pancreatic
parenchyma which may result in higher incidence of adverse events compared to focal
tissue acquisition of a solid pancreatic neoplasm. Fine-needle biopsy was chosen over
fine-needle aspiration (FNA) as this approach would provide both cytological and histological
specimens. There are no reported significant differences in diagnostic yield and adverse
events between FNA or FNB techniques. The 22G needle was studied as it represents
the predominant needle of choice in clinical practice.
Current guidelines consider EUS-TA a high-risk procedure and recommend against sampling
in patients on DOACs and, if sampling is indicated, discontinuation is recommended
(last dose ≥ 48 hours before the procedure and resumption up to 48 hours after the
procedure). The available evidence on bleeding secondary to EUS-TA in patients on
antithrombotics discontinued according to guidelines is very limited, as most studies
include a small number of patients and specifically DOACs are underrepresented. The
reported bleeding rate ranges between 2.4 % to 3.5 % and, in most cases, the bleeding
was not severe and could be managed conservatively with no need for blood transfusion
[5]
[10]. A prospective multicenter study on 85 patients on antithrombotics (14 on DOACs)
reported two bleeding events in patients on antithrombotics (one on dual antiplatelet
therapy and one on warfarin, 2.4 %; 95 % CI 0.6–8.3 %). Both patients experienced
bleeding despite discontinuation of thienopyridine and warfarin [10]. Another single-center retrospective Japanese study evaluated 908 patients undergoing
EUS-FNA of which 114 were on antithrombotics (eight on DOACs). The authors observed
higher bleeding events in the antithrombotic group (3.5 %) compared to the non-antithrombotic
group (0.8 %), P = 0.02. All bleeding was post-procedure and was significant in six patients, four
from the antithrombotic group and two from the non-antithrombotic group (odds ratio:
9.59; 95 % CI 2.12–43 %). However, none of the patients required blood transfusion
and were treated conservatively [5].
Our pilot study has several limitations including a small sample size, that was intentionally
designed to minimize animal use while providing both meaningful and relevant data
consistent with the objectives of the study. Future clinical trials with a larger
population size are needed to support or refute these findings. We opted for a DOAC
with the theoretically lowest bleeding risk and as such bleeding may differ with other
DOACs. Also, societal recommendations to assess endoscopy-related adverse events is
14 days post-procedure [4]; however, bleeding related to EUS-TA generally occurs during the procedure or within
72 hours [3] which was the basis for our survival interval of 4 days.
Conclusions
There is a clinical need to minimize the risk of periprocedural thromboembolic events
in anticoagulated patients undergoing high-risk endoscopy. Continuous anticoagulation
seems safe for some endoscopic procedures, but no data is available for patients on
uninterrupted DOACs undergoing EUS-TA. Although our study showed a trend toward higher
minor bleeding events in animals on apixaban, EUS-TA with apixaban did not increase
the incidence of clinically significant bleeding or prevent adequate cytopathological
evaluation. This pilot study may set the framework for future large population human
trials.
