Thrombus remodelling by reversible and irreversible P2Y12 inhibitors

Abstract Pharmacological inhibition of the platelet ADP-receptor P2Y12 is a cornerstone in the prevention of atherothrombotic events in adult patients with acute coronary syndrome (ACS). Thienopyridines such as clopidogrel and prasugrel exert their antithrombotic effect by means of active metabolites that irreversibly inhibit P2Y12. Due to the short half-life of these metabolites, a subpopulation of ADP-responsive platelets will form in between dosing. With increased platelet turnover rate or poor patient compliance, the fraction of ADP-responsive platelets will increase, potentially increasing the risk for new thrombotic events. In contrast, the reversible P2Y12 inhibition produced by direct-acting ADP blockers such as ticagrelor and cangrelor inhibit the entire platelet population. In this study, we evaluated the impact of these pharmacological differences on thrombus formation in an ex vivo flow chamber model. A customized image analysis pipeline was used for automatized, large-scale identification and tracking of single platelets incorporated into the thrombus, enabling quantitative analysis of the relative contribution of inhibited and uninhibited platelets to thrombus growth and consolidation. Comparative experiments were conducted using the irreversible and reversible P2Y12 inhibitors prasugrel active metabolite (PAM) and ticagrelor, respectively. Our results show that PAM inhibited thrombus platelet recruitment more gradually than ticagrelor, with a slower onset of inhibition. Further, we show that the presence of a small fraction (<10%) of uninhibited platelets did not abrogate the antithrombotic effect of PAM to any significant extent. Finally, we demonstrate a gradual enrichment of inhibited platelets in the thrombus shell due to selective recruitment of inhibited platelets to the thrombus periphery.


Introduction
ADP receptor antagonists block the activation of platelet P2Y 12 receptors by adenosine diphosphate (ADP), thereby putting a break on an important signaling pathway driving recruitment and activation of platelets during thrombus formation [1][2][3].As a testament to the beneficial effects of this unique antithrombotic mechanism, ADP-receptor antagonists have found widespread use for secondary prevention of acute coronary syndromes (ACS) and ischemic stroke, supported by a wealth of encouraging clinical data [4,5].The thienopyridines (ticlopidine, clopidogrel, and prasugrel) were the first group of ADP receptor antagonists approved for clinical use.Thienopyridines are administered orally as prodrugs and undergo metabolic conversion to produce short-lived active metabolites, which irreversibly inhibit P2Y 12 signaling by catalyzing the formation of a disulfide bridge near the ADPbinding site of the receptor [6].This means that reticulated platelets formed during the time interval in between drug administrations will retain P2Y 12 signaling capacity during a significant fraction of their circulating lifetime with the recommended once daily dosing regimen [7,8], resulting in the formation of a "mosaic" platelet population comprising platelets with both low (P2Y 12 -) and high (P2Y 12 +) ADP reactivity.In the average individual with a platelet life span of 7-10 days in the circulation, it is likely that P2Y 12 + platelets make up approximately 10% of the entire platelet population immediately before the administration of the next daily dose of a thienopyridine [9].However, the proportion of uninhibited, P2Y 12 +, platelets can be expected to be significantly higher in patients with an increased platelet turnover rate due to conditions that are known to be associated with cardiovascular disease, such as diabetes mellitus, immune thrombocytopenia (ITP), renal failure, essential thrombocythemia, and metabolic syndrome [7,9].
There is an inverse correlation between the fraction of newly formed reticulated platelets and the response to prasugrel in patients with ACS or ST elevation myocardial infarction (STEMI), as measured by multiple electrode aggregometry or the VerifyNow™ P2Y 12 platelet function assay [10,11].This indicates that the presence of a fraction of newly formed, P2Y 12 + platelets is enough to markedly reduce the antithrombotic efficacy of thienopyridines.To explore the underlying mechanisms for this finding, Hoefer et al. performed in vitro experiments on platelet aggregates formed after ADP stimulation of washed platelet suspensions containing a 80/20 mixture of P2Y 12 +/P2Y 12 -platelets, showing that P2Y 12 + platelets were overrepresented in the core of platelet aggregates [9].In a follow-up study, this finding was corroborated ex vivo using platelets from patients with coronary artery disease (CAD) treated with thienopyridines [7].These studies suggest that P2Y 12 + platelet subpopulations can establish "nuclei" that support ADP-induced platelet aggregation in the face of thienopyridine treatment.
The direct-acting ADP blockers ticagrelor and cangrelor represent a newer class of ADP-receptor antagonists that provide immediate P2Y 12 inhibition after systemic exposure.Ticagrelor, the only orally administered direct-acting ADP receptor antagonist, has an elimination half-life of 7-9 h after oral administration.Hence, patient adherence to the recommended twice daily dosing regimen ensures continuous systemic exposure.The reversible inhibitory activity observed for both ticagrelor and cangrelor also confers the advantage of ensuring consistent P2Y 12 inhibition across the entire platelet population, irrespective of platelet turnover rate [12].Although the clinical data is somewhat ambiguous, experimental evidence have suggested that these unique properties could translate into superior therapeutic efficacy compared to thienopyridines, especially in patients with a high fraction of reticulated platelets [9,10,13].However, recently published clinical data show that this potential therapeutic benefit may come at the expense of increased bleeding complications, especially in the elderly population [14,15].
Ex vivo flow chamber systems have previously been used to quantify the effects of P2Y 12 inhibition on thrombus formation at varied shear rates [16,17].In this study, our aim was to expand on previous studies by providing a head-to-head comparison of how irreversible versus competitive P2Y 12 inhibition affects thrombus formation in a whole-blood flow chamber model of thrombosis.In addition, we wanted to experimentally assess how different proportions of P2Y 12 + platelets affect the antithrombotic properties of thienopyridines.

Sample preparation and inhibitor treatment
Venous whole blood was drawn from healthy volunteers into hirudin sampling tubes (Roche) Diagnostics, Mannheim, Germany) after informed consent, in accordance with the approval by the regional ethics committee in Linköping (Dnr 2012/382-31 and the Declaration of Helsinki).Only information about gender and age was linked to the samples, thereby ensuring anonymization of the samples.The blood was used within 4 h and was stored as recommended by Roest et al (Roest et al., 2011).All experiments were conducted using anti-coagulated blood without any centrifugation steps.Ticagrelor and PAM were dissolved in DMSO, with equal amounts of DMSO added in controls.The final DMSO concentrations in all samples were 0.2%.Ticagrelor was used at final concentrations of 1.5, 0.4, and 0.2 µM.
An illustration of the experimental design for the fractional inhibition and platelet labeling is provided in Figure 1.In experiments using fractional P2Y 12 inhibition, one aliquot of the sample was treated with PAM at a final concentration of 3 µM.This was followed by 3-h incubation at RT to ensure complete degradation of the active metabolite before further use.The validity of this approach was confirmed using light transmission aggregometry, showing no difference in the response to ADP stimulation in platelets suspended in plasma treated with PAM for 3 h in comparison with untreated plasma (data not shown).Two aliquots from the inhibited and the non-inhibited samples, respectively, both corresponding to 5% of the final sample volume, were then labeled with a CD41 antibody conjugated to either CF488 or AF 555.Directly before each flow chamber experiment, these two differently labeled aliquots were then mixed gently with unlabeled samples containing inhibited and non-inhibited platelets in appropriate proportions to produce the desired fractional ratios of uninhibited (P2Y 12 + ) and inhibited (P2Y 12 -) platelets in the final sample.This approach allowed for separate detection and quantification of the inhibited and the uninhibited platelet fractions during experiments.To detect and eliminate any differences caused by the different dyes, dual labeling was used for all samples and the fluorescent label was alternated between the inhibited and uninhibited platelet fractions.For samples with 0% and 100% inhibited platelets, measurements represented averages for the two labels.Incubation with inhibitors was performed before labeling to ensure equal dosing in labeled and unlabeled samples.

Whole blood flow chamber thrombosis model
A collagen strip (250 µm wide) was coated on the glass slide with collagen solution (500 µg/mL) for 10 min, and a straight PDMS channel (height; 100 µm, width; 400 µm) was placed perpendicular over the collagen strip.Flow chambers were blocked with BSA (1 g/L) for 15 min prior to use.The blood was drawn through the flow chamber at a flow rate of 20 µl/min with a Fusion 200 syringe pump from KR Chemyx (Stafford, TX, USA) for a shear rate of 400 s −1 .

Image acquisition and data analysis
Z-stack time-lapse images were captured with a wide-field 20x objective (NA 0.8) on a Zeiss Axio Observer Z1 with a Colibri LED-module and a Neo 5.5 sCMOS camera (Andor Technology Ltd., UK) controlled by µManager software (Vale lab, UCSF, CA, US).Z stacks covering the entire thrombus volume with 2 µm between slices were acquired every 30 s, for a total of 20 min.The resulting 4D image data was processed using a customdesigned data analysis pipeline according to a previously described protocol [18,19].Image segmentation with respect to individual stained platelets allowed for quantitative analysis of platelet numbers, positions, and movements.

Graphing & statistics
For bivariate relational data, line graphs were used to visualize dynamic trends in means of continuous variables over time, with each assessed time point (frame) displayed individually on the x-axis.To reduce the impact of intermittent optical artifacts, a rolling time window of four frames was used when calculating mean values for the platelet population in each individual experiment.Means from individual experiments were then treated as single data points in the aggregate analysis as displayed on the y-axis.To visualize trends that were more persistent over time, histograms were constructed by binning individual time points into 120 s time intervals comprising 4 consecutive time points.In these analyses, data were presented as point graphs, with connecting lines to facilitate detection of trends over time.Simple linear regression with least square fitting was used to validate the association between two continuous numerical variables across multiple levels.For groups comprising n treatment conditions, pairwise comparisons of means were performed using one-sided t-tests (n = 2) or ANOVA followed by Tukey's HSD test (n > 2).95% confidence intervals for the means were calculated by bootstrapping and included in figures as indicated.

Effects of reversible and irreversible P2Y 12 inhibitors on thrombus platelet counts
To investigate dose-response effects of the reversible P2Y 12 inhibitor ticagrelor, experiments were performed using low (0.2 µM), intermediate (0.4 µM) and high (1.5 µM) ticagrelor dosing.For PAM, experiments were conducted using a low, intermediate, and high fraction of drug-treated platelets (50%, 80%, and 90%), as well as with 100% drug-treated platelets.As shown in Figure 2 and Supplemental Figure 1, thrombus platelet recruitment was visibly reduced by P2Y12 inhibition with both ticagrelor and PAM.At the highest dosing level of ticagrelor (1.5 µM), only a monolayer of adherent platelets could be observed on top of the collagen-coated area of the flow chambers indicating complete inhibition of secondary mediator driven recruitment of platelets.A similar pattern could be observed when ≥90% of the platelet population was treated with PAM, although a slight thickening of the platelet lining could be distinguished in the edges of the flow chambers, even when 100% of the platelet population was druginhibited.To quantify how P2Y12 inhibitors affect the number of platelets that are incorporated in thrombi over time, we counted the number of stained platelets in thrombi every 30 s during 20 min (Figure 3a).P2Y 12 inhibition caused a gradual divergence of thrombus platelet counts relative to control over time, with a clearly discernible separation of counts for all treatments from around 180 s.For ticagrelor, a significant (P < .001)dose response effect was observed on mean thrombus platelet counts (Figure 3b, lower panel), with a reduction of 28.6%, 52.3%, and 68.5% for low, intermediate, and high dosing levels, respectively (P < .001).Likewise, there was a significant (P < .001)association between thrombus platelet count and the fraction of platelets inhibited with PAM in the range 0-90% (Figure 3b, upper panel).However, no significant change in mean platelet counts could be detected when the fractional inhibition with PAM was increased from 90% to 100% (mean thrombus platelet counts of 3229 and 3180, P = .52,Figure 2b).This finding indicates that the antithrombotic efficacy of thienopyridines is preserved in the presence of a small subpopulation of P2Y 12 + (uninhibited) platelets.Interestingly, the effects of PAM and ticagrelor on thrombus platelet counts differed substantially during the earliest stages of thrombus development (Figure 3c).In the 0-120 s time interval, P2Y 12 inhibition with PAM had small negative or even slightly positive effects on thrombus platelet counts, with inhibitory efficacies of −5.1%, −3.5%, 11.7%, and 6.2% for 50%, 80%, 90%, and 100% inhibited platelets.After this initial phase, the inhibitory efficacy gradually increased to levels approaching those for the corresponding ticagrelor dosing, eventually reaching 20.9%, 49.4%, 63.5%, and 64.8% for 50%, 80%, 90%, and 100% inhibited platelets at the last assessed time interval (1,080-1,200 s).In contrast, ticagrelor immediately produced robust reductions of thrombus platelet counts, with inhibitory efficacies at 0-120 s ranging from 24%, 40.2%, and 46.9% for low, intermediate, and high dosing, and then only increased moderately over time, reaching 32.3%, 59.6%, and 79.2% at the last assessed time interval.

Effects of reversible and irreversible P2Y 12 inhibitors on thrombus size and platelet contraction
To quantify how P2Y 12 inhibitors affect the spatial expansion of growing thrombi, we calculated the platelet center of mass in the vertical (z) axis (thrombus center height) and the mean distance to the center of mass in the horizontal plane (thrombus center width, Figure 4).In our experimental model, spatial thrombus growth occurred almost exclusively in the z-axis.This is illustrated by the observation that thrombus center height in control experiments increased more than tenfold, from 1.4 to 15.7 µm (Figure 4a), in lieu of P2Y 12 inhibition.In contrast, mean width only increased by 29% (from 129 to 166 µm, Figure 4c).In an aggregate analysis of all time points, ticagrelor produced a significant (P < .001)and dose-dependent inhibition of thrombus center height, with AUC reductions of 23.2%, 48.4%, and 59.6% for low, intermediate, and high dosing, respectively (Figure 4b).Increasing fractions of PAM-treated platelets reduced thrombus center height to a similar extent as ticagrelor, with inhibitions of 27.3%, 48.4%, and 58.4% for 50%, 80%, and 90% inhibited platelets (P < .001).Neither P2Y 12 inhibitor had any significant effect on thrombus center width (P = .29,Figure 4e-f).
When analyzing different time periods separately, we observed a markedly reduced effect of P2Y 12 inhibitors on thrombus center height during the early stages of thrombus formation.Instead, trends toward positive effects on thrombus center height were observed for all dose levels of ticagrelor and for 3 out of 4 PAM treatment groups during the first assessed time interval 0-120 s (Figure 4c) although the increase was not statistically significant.As shown in Figure 5g, this finding can at least partially be ascribed to diminished platelet movements directed downwards toward the collagen-coated surface of the flow chambers.Mean platelet contraction, i.e. the average net movement of platelets toward the thrombus center of mass, was dose or fraction-dependently reduced by both ticagrelor and PAM to similar extents (P < .001for both treatments, Figure 4g-h).

Effects of fractional P2Y 12 inhibition on platelet subpopulations -paradoxical enrichment of P2Y 12 -platelets in the thrombus shell
Previous studies have indicated that subpopulations of uninhibited platelets can function as nucleation sites for platelet aggregation during the treatment with irreversible P2Y 12 inhibitors [7,9].To investigate the potential effects of this phenomenon on thrombus formation in our model, we quantified the spatial distribution of P2Y 12 -and P2Y 12 + platelets in thrombi formed from a P2Y 12 mosaic platelet population.To mimic scenarios that would be feasible in patients treated with thienopyridines, experiments were performed with 0, 10, 20, and 50% P2Y 12 + (uninhibited) platelets.When comparing the total number of platelets from each of the two labeled subpopulations in thrombi over time, we found no signs of "nucleation," i.e. enrichment of P2Y 12 + platelets during the initial stages of thrombus formation (Figure 5a-b).Instead, a progressive enrichment of inhibited platelets was observed from around 240 s in experiments with 50% and 80% PAM inhibited platelets.As shown in Figure 5d, this overrepresentation of PAMtreated (P2Y 12 -) platelets was more pronounced in the thrombus surface (platelets located <2 µm distance from the thrombus surface).To find out if differential recruitment of P2Y 12 -and P2Y 12 + platelets could be one causal factor driving this unexpected phenomenon, we quantified the average number of new platelets that were recruited to the thrombus surface every 120 s (Figure 5e).As expected, the total number of newly recruited platelets was progressively reduced in thrombi formed with higher fractions of PAM-treated platelets.However, when comparing the recruitment of platelets from the respective subpopulations, we found an overrepresentation of P2Y 12 -platelets at every measured time interval in experiments with 50% and 80% PAM-treated platelets (Figure 5f), with a progressive increase in P2Y 12 -/P2Y 12 + ratio from 0.53/0.47 to 0.63/0.37 during experiments.
Differences in contractile movements could plausibly lead to a gradual redistribution of platelet subpopulations in thrombi.To test the validity of this hypothesis in our experiments, we compared the individual trajectories of P2Y 12 -and P2Y 12 + platelets in our experiments.On average, platelets from both subpopulations exhibited a negative cumulative displacement in the z-axis (i.e.moved downwards toward the collagen surface), but P2Y 12 + platelets traveled significantly longer distances toward the collagen-coated bottom than their P2Y 12 -counterparts (−0.52 µm and −0.21 µm, P < .001).In addition, P2Y 12 + platelets displayed a more contractile phenotype, as demonstrated by a greater decrease in the average distance between untreated platelets and their 5 closest neighbors from the first to the last observation of individual platelets (Figure 5g).

Discussion
Previous preclinical work on the role of P2Y12 during thrombus formation has provided a robust rationale for transitioning from irreversible to reversible ADP blockers in the treatment of thrombotic disorders.It is therefore not surprising that the clinical approval of ticagrelor for secondary prevention of myocardial infarction, supported by the impressive results of the PLATO trial [20], was heralded with much enthusiasm in the research community.Since then, however, clinical experience has been somewhat sobering, as a growing body of evidence suggests that the therapeutic benefit of ticagrelor vis-a-vis members of the thienopyridines are undermined by an elevated risk of serious bleeding complications [14,21].Prompted by these recent developments, the aims of this study were to provide a comprehensive head-tohead comparison of the antithrombotic activity of ticagrelor and prasugrel, the two most recently developed reversible and irreversible P2Y 12 inhibitors in current clinical use.In addition to studying bulk thrombus properties (e.g.thrombus size), our experimental and analytical methodology also enabled us to detect large numbers of individual platelets and track their movements inside thrombi.This unique analytical feature allowed for a detailed assessment of how the antithrombotic properties of thienopyridines are affected by the presence of a fraction of uninhibited platelets in the platelet population, as is invariably the case in between dosing among patients.
In our study, we identify several differences in the global effects of ticagrelor and prasugrel on thrombus formation, some of which have not previously been reported.Perhaps, the most interesting of these findings with respect to its implications for hemostatic function, are the differences we found in the timing of onset of inhibition for the respective drug.It is common knowledge that platelet aggregates are the primary components of the primary hemostatic plug that rapidly form the first line of defense against excessive bleeding.It is therefore possible that the more immediate inhibitory effect of ticagrelor on thrombus buildup could contribute to the increase in bleeding complications associated with its use in recent large clinical studies.Regarding antithrombotic potency, our failure to find any significant differences in the sizes and platelet counts of thrombi formed from a 100% P2Y 12 + platelet population as compared to a chimeric platelet population with 90% P2Y 12 + and 10% P2Y 12 -platelets could help to explain why differences in antithrombotic efficacy between reversible and irreversible P2Y 12 inhibitors have been smaller than perhaps expected, as >90% is likely to be the ratio of drug-treated to untreated platelets that would be present at most times in patients with a normal platelet turnover rate.
Another aim of this study was to provide a detailed assessment of how the drug-induced P2Y 12 mosaicism that is observed in patients treated with thienopyridines affects thrombus growth and composition.We predictably found a negative correlation between the percentage of P2Y 12 -inhibited platelets and the total number of new platelets that were recruited to thrombi, a trend that was increasingly strong during later time points of the experiments.This is likely caused by accelerated "thrombus cooling," i.e. decreased paracrine signaling, in the P2Y 12 mosaic thrombus platelet population, causing rapidly diminished GpIIb/ IIIa activation in platelets located at the thrombus surface.This "cooling effect" reduces the availability of immobilized fibrinogen at the platelet surface, depriving bypassing platelets of a scaffold to which they can attach.The enrichment of P2Y 12platelets we observed in the shell of thrombi formed from P2Y 12 chimeric platelet populations could also be related to this "cooling" phenomenon.Considering the short time scale on which platelet adhesion and disaggregation occurs, the 30 s time interval between each cycle of thrombus imaging in our experiments gives plenty of time for platelets to complete a cycle of engagement and disengagement with thrombi without ever being detected, especially in the turbulent and highly dynamic shell of thrombi.Thus, it is possible that the apparent decrease in recruitment of P2Y 12 + platelets observed in this study is caused by an increased propensity of P2Y 12 + platelets to rapidly disengage with thrombi after making contact with the thrombus surface.
Upon attachment, platelets rapidly activate their contractile apparatus by a Rho kinase-dependent mechanism.In a process once dubbed "internal contraction" [22], platelets pull themselves closer to adjacent platelets to increase thrombus packing density [23].As P2Y 12 has been shown to potentiate this morphological transition [24], lack of P2Y 12 signaling in drug-treated platelets can be anticipated to decrease these contractile movements.Indeed, this is supported by our findings that P2Y 12 + platelets exhibit a more contractile phenotype than their drug-treated neighbors after recruitment to the thrombus shell (Figure 5g).Based on our observation that P2Y12+ platelets exhibit a more contractile phenotype, we hypothesize that the observed enrichment of drug-treated platelets in the thrombus shell of thrombi formed by P2Y 12 mosaic platelet populations is caused by a progressive "truncating effect" of drug-inhibited platelets with low surface exposure of immobilized fibrinogen.In a decidedly "cool" thrombus environment, such as the surface of a thrombus formed by a P2Y 12 mosaic platelet population, the low local availability of immobilized fibrinogen provides few anchoring points for platelet surface glycoproteins.It is likely that the increase in mechanical strain caused by platelet internal contraction could increase the propensity of P2Y 12 + platelets to pull themselves loose and detach from the thrombus surface.However, as direct experimental proof of this concept would require experiments performed with a higher time resolution than what is achievable using conventional fluorescence microscope systems, it is outside the scope of the present study.

Figure 1 .
Figure 1.Overview of experimental design and methods for fractional inhibition and fractional platelet labeling.Illustration of the workflow used in the study including the steps used to obtain blood containing a 80/20 mix of PAM-inhibited and uninhibited platelets as well as fractional platelet labeling (5% of total platelets for each subpopulation).Created with BioRender.Com.

Figure 2 .
Figure 2. Visualization of platelet accumulation.Representative examples of platelet accumulation on collagen strips after 20 min blood flow at 400s −1 .Colours indicate distances between individual platelets and the thrombus surface (µm).

Figure 3 .
Figure 3. Effects of ticagrelor and PAM on thrombus platelet counts.Thrombi were formed on collagen strips using blood treated with ticagrelor or with fractional inhibition by prasugrel active metabolite (PAM) as indicated in the figure legends.Estimators of platelet counts are expressed as means of absolute numbers (a-b) and average inhibitory efficacies (c-d).n = 9-10 (independent donors).Shaded areas (a) or error bars (b-d) are 95% confidence intervals calculated by bootstrapping, * P < .05,** P < .01,*** P < .001.

Figure 4 .
Figure 4. Effects of ticagrelor and PAM on thrombus size.Thrombi were formed on collagen strips using blood treated with ticagrelor or with fractional inhibition by prasugrel active metabolite (PAM) as indicated in the figure legends.Estimators for thrombus height (a-d), width (e-f) and platelet contraction were expressed as means of absolute values (a-b & e-f) and mean inhibitory efficacies (c-d & g-h).n = 9-10 (independent donors).Shaded areas (a) or error bars (b-d) are 95% confidence intervals calculated by bootstrapping, * P < .05,** P < .01,*** P < .001. .

Figure 5 .
Figure 5. Surface enrichment of P2Y 12 -platelets.Treatment with ticagrelor or fractional inhibition with prasugrel active metabolite (PAM) was used as indicated in the figure legends.(a-b) Thrombus platelet counts for the two fluorescently labeled platelet subpopulations, each representing 5% of the entire platelet population used in experiments.When fractional P2Y 12 inhibition with PAM was used, dashed lines represent the drug-treated (P2Y 12 -) platelet subpopulation as indicated.(c) Platelets incorporated into thrombi were dichotomized by the shortest distance between each individual platelet and the thrombus surface, with a cutoff of 2 µm used to separate the interior and surface regions of thrombi.(d) Comparative analysis of the number of PAM-treated (P2Y 12 -) and untreated (P2Y 12 + ) platelets incorporated into the interior (bottom row) and surface (top row) of the thrombus.(e) Numbers of newly recruited platelets in the surface region of thrombi formed from platelet populations with 0-50% P2Y 12 + platelets.(f) Numbers (left panel) and ratios (right panel) of P2Y 12 -/P2Y 12 + platelets that were recruited to the surface of thrombi formed from platelet populations with 50% and 80% P2Y 12 -platelets during the indicated time intervals.(g) Cumulative platelet displacement in z axis (left panel) and change in average platelet distance to the 5 closest neighbors between the first and last observation (right panel) in thrombi formed from platelet populations with 50% and 80% inhibited platelets.n = 9-10 (independent donors).Error bars are 80% confidence intervals calculated by bootstrapping, * P < .05,** P < .01,*** P < .001.