Non-invasive dentistry method for clinical determination of bleeding time: evaluation in subjects with and without direct oral anticoagulant therapy

Abstract

See More

Assessing bleeding duration in patients on direct oral anticoagulants (DOACs) is crucial for evaluating reversal strategies, yet no standardized methods are currently available for clinical trials. We investigated a non-invasive, reproducible bleeding model based on a routine dental cleaning procedure, aiming to provide a clinically meaningful endpoint for testing DOAC antidots.

We enrolled 90 subjects in this prospective observational pilot study: 49 not receiving DOACs (group 1; 30 female, 19 male; mean age: 40.7 ± 16 years) and 41 on DOACs (group 2; 22 female, 19 male; mean age: 71.3 ± 20.4 years). Mouth-rinse samples were collected before and at 5-minute intervals for up to 60 min following bleeding-on-probing assessment, periodontal staging, and dental cleaning. Red blood cell (RBC) counts in these samples were determined using a hemocytometer. Bleeding cessation was defined as an RBC count below 150 cells/µL. Reaching this threshold within 15 min was defined as the clinical endpoint (EP).

In group 1, 83.7% reached the EP (group 1a (without gingivitis): 90.9%; group 1b (with gingivitis): 77.8%). In group 2, only 39% reached the EP (group 2a (without gingivitis): 60%; group 2b (with gingivitis): 32.3%; p < 0.001 vs. group 1). The most relevant comparison for future antidote trials is between group 1b (non-DOAC with gingivitis) and group 2b (DOAC with gingivitis; p < 0.001).

This model offers a standardized, non-invasive, and clinically relevant approach to monitor bleeding in DOAC-treated patients. The defined clinical endpoint (≤ 150 RBC/µL after 15 min) could serve as a useful surrogate for evaluating the efficacy of temporary DOAC reversal agents.

Introduction

Bleeding time (BT), for example as determined by Duke’s filter paper method, is a clinical test historically used to assess the ability of platelets to form a hemostatic plug following vascular injury. It essentially measures how long it takes for bleeding to stop and has traditionally been applied to evaluate primary hemostasis, particularly platelet function [1]. The determination of bleeding time in clinical studies for antidotes to anticoagulants is fraught with several challenges, including technical variability, limited sensitivity, ethical concerns, and poor correlation with clinical outcomes [2, 3].

First, BT is subject to substantial technical variability. Methods based on skin incisions differ in depth, length, and pressure, and are performed at different anatomical sites (e.g., forearm vs. earlobe), resulting in inconsistent and poorly reproducible outcomes [1, 4]. Additionally, BT is not sensitive to most anticoagulants such as warfarin or direct oral anticoagulants (DOACs), and often fails to detect the subtle hemostatic changes induced by these drugs or their reversal agents [3,4,5].

Furthermore, BT results vary significantly between individuals due to factors such as skin thickness, vascular structure, and even ambient temperature. This inter-individual variability makes it difficult to interpret BT results reliably or to use them as endpoints for evaluating antidote efficacy [6]. Compounding these issues are ethical concerns: the procedure is invasive, with risks including prolonged bleeding, infection, scarring, and impaired wound healing. These risks are particularly relevant in elderly or vulnerable patient populations and often raise concerns during protocol review by institutional review boards.

Importantly, there is limited evidence that BT correlates with actual clinical bleeding risk or outcomes, especially in anticoagulated patients [7]. For this reason, BT is no longer considered a reliable marker of bleeding risk or of antidote effectiveness in reducing bleeding complications. Its limited sensitivity, combined with ethical and logistical drawbacks, makes it an impractical and potentially unacceptable tool for repeated measurements in clinical trials. Moreover, reliance on BT may not align with current regulatory expectations or Good Clinical Practice (GCP) standards, which increasingly favor more specific, standardized, and non-invasive coagulation tests.

Given these limitations, there is a clear need for a safer, more reproducible, and clinically relevant method to assess bleeding in the context of anticoagulant therapy and reversal. To address this gap, we developed and tested a novel, minimally invasive bleeding model based on a routine dental cleaning procedure—a procedure known to trigger mild, localized gingival bleeding in most individuals.

In our approach, participants underwent a standardized dental cleaning following periodontal assessment and probing for bleeding. We collected mouth-rinse samples before and at 5-minute intervals up to 60 min afterward, quantifying red blood cell (RBC) counts per µL using a hemocytometer—a simple microscopic counting chamber. Bleeding cessation was defined as the point when RBC counts fell below a predefined threshold, and we defined a clinical endpoint as achieving this level within 15 min.

This method offers several advantages: it mimics a natural mucosal bleeding event, minimizes patient risk, and allows for serial sampling and objective quantification of bleeding over time. Our primary objective was to determine whether this model could distinguish between subjects taking DOACs and those not on anticoagulants, while also evaluating the potential influence of gingival inflammation on bleeding duration.

Subjects and methods

This cross-sectional non-interventional observational study was conducted with subjects who came to the dental office for dental hygienic cleaning procedures to collect pilot information about the differences in the time to cessation of bleeding in different patient cohorts, which were selected based on differences in their Bleeding on Probing (BOP) score, the stage of periodontal disease, and on the use of oral anticoagulants. It was conducted in accordance with all applicable ethical and regulatory standards. The protocol was approved by the responsible institutional review board (Ethikkommision der Landesärztekammer Rheinland-Pfalz, Application No.: 2021–16124). Subjects having routine appointments in the dental ward were invited to participate into this protocol, and they signed informed consent before any additional study measure. The targeted study population consisted of a healthy control cohort (~ 30 subjects without DOAC and gingivitis), a non-DOAC cohort with gingivitis (~ 10 subjects), and at least ten subjects each taking one of three investigated DOACs (rivaroxaban, apixaban, edoxaban). One enrolment goal was to get an equal gender distribution also within the subgroups.

To be eligible for this study, subjects had to be at least 18 years of age or older, and if taking a DOAC, they had to receive a stable DOAC-regimen with either apixaban, rivaroxaban, or edoxaban for at least 7 days prior to the scheduled dental procedure. The last DOAC dose had to be taken at least 2 h before the dental cleaning procedure. Subjects were excluded if they had taken any investigational drug during a clinical research study within the previous 3 months, or if they were taking any other anticoagulant drug than the three allowed DOACs (e.g. antiplatelet medication, such as acetylsalicylic acid, prasugrel, or ticagrelor). Subjects selected for the healthy control group were not allowed to have any known bleeding disorder or coagulation disorder.

After signature of informed consent, a baseline mouth rinse sample was collected (rinse for 15 s, with 10 mL of 0.9% NaCl-solution) as a pre-procedure baseline assessment of red blood cell count. Bleeding from the gums was determined by a dentist using a Bleeding on Probing (BOP) assessment [8]. It required a gentle insertion of a periodontal probe tip into the sulcus of each tooth (1–2 mm deep), followed by a gentle sweep around from proximal surface to proximal surface. Any bleeding seen within 10 to 15 s after removing the probe tip was recorded. Studies have shown that the absence of BOP indicates a state of health that will likely continue for several months after this examination. BOP in more than 10% of the examined pockets was considered indicative for presence of gum inflammation (gingivitis).

Thereafter, the status of periodontal disease was assessed by a dentist in accordance with the 2018 periodontitis staging standard [9]. In addition, active matrix- metalloproteinase (aMMP-8) levels in gingival crevicular fluid (GCF) were measured using the dentoELISA. (dentognostics GmbH, Solingen, Germany, cut-off > 20 ng/mL), as an indicator for an ongoing destructive collagenolytic process in the mouth [10]. Consecutively, the participants were subjected to the routine dental cleaning procedure (mechanical cleaning) with tartar removal, which was allowed to last for not more than 30 min. Additional mouth rinse samples were taken at timepoints 0, 3, 6, 9, 12, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60 min post-procedure. Timepoints are relative to the end of the cleaning procedure.

The mouth rinse samples were centrifuged, and the pellet was dissolved in 2 mL of physiological saline solution. The number of red blood cells (red blood cell count, RBC) was counted under the microscope using a hemocytometer (Neubauer chamber) by the same laboratory technician throughout the study. The Neubauer hemocytometer and the cover glass were cleaned with ethanol and the cover glass was placed on top of the hemocytometer’s chambers to stop the sample from evaporating. 10 µl of the mouth rinse sample was loaded into both counting chambers with a micropipette and the hemocytometer was placed under the microscope. The cells were counted, focusing on the grid lines of the counting area with a 5-10x objective and the cell concentration was calculated.

The statistical analysis was performed by means of descriptive statistics. Subjects were stratified in accordance with their BOP score [8], and in addition according to their DOAC uptake status. The threshold to describe an “ongoing dental hygiene procedure-induced bleeding” was to be defined with the help of the results of the pre-procedure samples. The results were tabulated and the mean and standard deviation for the different subgroups was determined for all feasible observation parameters. Differences between the sub-groups were analysed by means of appropriate parametric and non-parametric methods.t was expected that coagulation in healthy subject not taking DOACs definitely achieved after 40 min. Therefore, the cut-off number of erythrocytes was determined by using the 40 min sample of the control group: Cut-off (RBC/µL) = mean RBC(Group 1a)_40 min + 2*SD. The clinical endpoint (EP) for the analysis was defined as the number of subjects in each group reaching the cut-off RBC until time-point 15 min.

Results

A total of 94 subjects were enrolled into the study of whom four subjects were excluded from the analysis for different reasons (wrong cleaning procedure; 1 case, prior uptake of aspirin: 1 case, no measurable bleeding after the dental cleaning procedure: 2 cases). Finally, 90 subjects performing the study per protocol were included into the analysis of whom 49 were not taking DOACs (Group 1: 30 female, 19 male, age: 40.7 ± 16.0 yrs.), and 41 were taking DOACs (Group 2: 22 female, 19 male; age: 71.3 ± 20.4 yrs.).

In group 1, 22 subjects had no indication of gingivitis (Group 1a: BOP < 10%), while 27 subjects had a BOP-score of ≥ 10% (Group 1b). Based on the results of group 1a, the RBC cut-off indicating bleeding cessation was determined to by 150/µL The receiver operator curve (ROC) for subjects reaching the cut-off for cessation of bleeding (RBC < 150/µL) over time is shown in Fig. 1.

Fig. 1

Percentage of non-DOAC subjects reaching the cut-off for cessation of bleeding over time (≤ 150 RBC/µL) after the end of the dental cleaning procedure (groups 1, 1a, and 1b)

Subjects without gingivitis (BOP < 10%) reached the endpoint earlier than subjects with gingivitis (p < 0.001). For the later comparison with DOAC patients, however, the non-DOAC control with gingivitis (Group 1b) represents the more suitable comparator group, because a theoretical inhibition of anti-coagulation in the DOAC group would in the majority of the cases lead to a Non-DOAC situation, but with gingivitis.

Based on the BOP status, the DOAC study cohort (Group 2: n = 41), was stratified into the following subgroups: Subjects without gingivitis (BOP < 10%) on DOAC treatment (Group 2a, n = 10) and subjects with gingivitis (BOP ≥ 10%) on DOAC treatment (Group 2b, n = 31). The receiver operator curve (ROC) for subjects on DOAC reaching the cut-off for cessation of bleeding (RBC < 150/µL) over time is shown in Fig. 2.

Fig. 2

Percentage of non-DOAC subjects reaching the cut-off for cessation of bleeding over time (≤ 150 RBC/µL) after the end of the dental cleaning procedure (groups 2, 2a, and 2b)

It can be seen that a far lower number of subjects on DOACs reached the clinical endpoint of achieving the RBC cut-off within 15 min, which is indicating prolonged bleeding time and actually the desired effect of the drugs. The overall statistical analysis also providing the demographic breakdown of the individual subgroups is provided in Table 1.

Table 1 Demographic data and result summary for the different subgroups (EP: reaching the RBC cut-off value (< 150/µL within 15 min)

Subjects in group 1 were younger and healthier than patients in group 2. In the healthy control group without DOAC treatment and without gingivitis (group 1a), the defined bleeding cessation endpoint was reached by more than 90% of the participants. The comparison of interest for potential interventional studies with DOAC antidotes is the comparison between non-DOAC subjects with gum inflammation (group 1b) and DOAC-patients with gum inflammation (group 2b). While 77.8% of the non-DOAC subjects reached the clinical endpoint, this was only the case for 32.3% of the DOAC-patients with gum inflammation (p < 0.001). When adjusting for age, number of teeth, and other potential confounders (e.g., systemic inflammation, oral hygiene status), there was no difference in the study outcome.

The DOAC patients without gingivitis (n = 31) were numerically almost equally distributed amongst the three allowed DOACS, rivaroxaban (n = 10), apixaban (n = 12), and edoxaban (n = 9). There was a remarkable difference between the results with edoxaban and the two other DOACs. EP was reached by 55.6% of subjects with edoxaban, while only 30.0% and 16.7% achieved EP with rivaroxaban and apixaban, respectively. The EP ORC analysis for the three DOACs is provided in Fig. 3.

Fig. 3

Percentage of subjects taking the different investigated anticoagulants reaching the primary endpoint cut-off (≤ 150 RBC/µL) after the end of the dental cleaning procedure

Subjects taking edoxaban reached EP earlier than subjects taking rivaroxaban and apixaban (15.4 ± 10,1 min vs. 33.0 ± 20.8 min (p < 0.05) and 38.6 ± 18.6 min (p < 0.01)).

The procedures were well tolerated, and no adverse event or serious adverse event was reported from any participant.

Discussion

This pilot study explored the feasibility of using bleeding after a routine dental hygienic cleaning procedure as a clinical model for assessing the efficacy of direct oral anticoagulants (DOACs) and potentially their reversal by antidotes. The main finding was that patients on DOAC therapy with gingivitis exhibited significantly longer bleeding times than non-DOAC subjects with comparable periodontal inflammation, as defined by a red blood cell (RBC) count of ≤ 150/µL at 15 min post-procedure.

These findings are consistent with the well-established risk of prolonged bleeding following invasive dental procedures in patients treated with anticoagulant or antiplatelet medication [11,12,13,14]. While numerous guidelines exist for the dental management of patients on warfarin or aspirin [11,12,13,14], there remains a lack of national dental clinical practice guidelines for the newer DOACs [15]. Our results contribute to filling this gap by offering a pragmatic approach to assess DOAC activity through dental bleeding. Further subgroup analysis confirmed the model’s sensitivity, with prolonged bleeding observed in patients taking rivaroxaban and apixaban, consistent with their known anticoagulant effects [16].

In routine dental care, DOACs are typically continued before treatment, offering a natural setting to observe their effects on post-procedural bleeding. This allowed us to investigate whether a simple mouth rinse test following standard dental cleaning could serve as a practical bleeding model—especially for evaluating DOAC activity or reversal. Our cross-sectional pilot study tested this approach and showed that the method allows for quantifiable, time-resolved assessment of bleeding cessation.

A key result was the significant difference in bleeding duration between DOAC-treated patients and non-DOAC controls, particularly in those with gingival inflammation. While 77.8% of non-DOAC subjects with gingivitis reached the clinical endpoint (red blood cell count ≤ 150 cells/µL after 15 min), only 32.3% of DOAC-treated patients with gingivitis did so (p < 0.01). This suggests that the proposed endpoint may be a reliable marker for detecting the anticoagulant effect of DOACs and for assessing the action of potential reversal agents.

Among DOAC subgroups, patients on rivaroxaban and apixaban displayed the expected pattern of prolonged bleeding. However, the edoxaban subgroup showed no statistically significant difference from non-DOAC patients with gingivitis. As no laboratory markers of anticoagulation (e.g., PT, aPTT, anti-Xa) were collected in this observational study, this finding remains difficult to interpret. It is conceivable that the dosing regimen (30–60 mg once daily) in our edoxaban group did not achieve an anticoagulant effect comparable to that of apixaban or rivaroxaban. This interpretation aligns with previous studies showing a lower bleeding risk for edoxaban compared to rivaroxaban (HR 0.74), though no significant difference was observed between edoxaban and apixaban [17].

An additional possible explanation relates to the limitations of the method used by Pesce et al., who assessed bleeding time using standardized skin incisions on the forearm [18]. While their data suggested longer bleeding times in DOAC-treated patients, they did not differentiate between individual DOACs. Our study, using a mucosal bleeding model with objective red blood cell quantification, may therefore offer greater sensitivity and drug-specific resolution.

We also observed that DOAC patients had significantly fewer teeth than non-DOAC controls (24.1 ± 4.3 vs. 27.4 ± 3.6; p < 0.001), likely reflecting the older age and more advanced periodontal disease in this population. Still, all groups exhibited signs of active periodontal inflammation, confirmed by elevated aMMP-8 levels—an established marker of collagenolytic activity and predictor of periodontitis progression [19, 20]. This underlines the need for preventive dental strategies in patients with systemic conditions requiring anticoagulation, such as diabetes, atherosclerosis, or coronary artery disease [21, 22].

Our findings suggest that bleeding induced by routine dental cleaning offers a minimally invasive and reproducible model to assess anticoagulant effects in vivo. The procedure is well-tolerated, avoids the ethical and logistical challenges of skin incision-based BT tests, and allows for standardization through consistent sampling at fixed intervals. The method does not require advanced laboratory equipment—only a microscope and hemocytometer—and may be simplified by focusing on a single 15-minute post-procedure time point.

This study is subject to limitations typical of exploratory pilot trials. These include the absence of coagulation biomarkers, the small size of the investigated subgroups, and the non-interventional nature of the trial. These factors limit the generalizability of our findings and the ability to fully interpret the edoxaban results.

However, as a proof-of-concept, the study successfully introduces a clinically relevant, non-invasive bleeding model that may be suitable for future trials—particularly those assessing the efficacy of DOAC antidotes. The results allow for power calculations and refined protocol development in subsequent confirmatory research.

In conclusion, this pilot study proposes a novel, standardized method for evaluating bleeding cessation after dental cleaning as a proxy for DOAC activity. The defined endpoint (RBC count ≤ 150/µL at 15 min) distinguished between subjects on DOACs and controls. Future trials incorporating this model, together with biochemical markers and larger sample sizes, may help establish its role in evaluating anticoagulant efficacy and reversal.