Plasma PF4 is a reliable biomarker of platelet reactivity and predictive of bleeding risk in DAPT patients undergoing cardiac surgery Authors: Joseph Roux de Bezieux, Julien Amour, Aurélien Philippe, Jeanne Rancic, Nicolas Gendron, Sébastien Bertil, Marc Garnier, Pascale Gaussem, David M Smadja Published: 2026 ABSTRACT AbstractBackgroundPlatelet activation is crucial for hemostasis and thrombosis, particularly in patients on antiplatelet therapy. Managing bleeding risk during cardiac surgery requires understanding platelet dynamics, especially when antiplatelet agents are maintained until the coronary artery bypass surgery (CABG).Aimsto explore the relevance of plasma levels of sCD40L, sP-Selectin, PF4, and NAP-2 in predicting bleeding events during CABG in patients under dual antiplatelet therapy (clopidogrel+aspirin, DAPT) or aspirin.MethodsPlatelet reactivity before CABG was assessed in 143 patients by plasma levels of sCD40L, sP-Selectin, PF4, and NAP-2 as well as platelet aggregation tests and platelet membrane expression of P-Selectin. Postoperative bleeding was monitored at 24 h with the chest tube blood output.ResultsPatients receiving DAPT demonstrated markedly reduced plasma levels of PF4 and NAP-2 compared to patients on aspirin alone (PF4: 676 vs. 992 ng/ml, p < 0.001; NAP-2: 167 vs. 281 ng/ml, p < 0.001). PF4 and NAP-2 levels correlated with 10 and 20 µM ADP-induced and collagen-induced platelet aggregation (p = 0.01) but not with arachidonic acid or PAR1ap as agonists. Platelet membrane P-Selectin expression before CABG was significantly correlated with PF4 and NAP-2 levels (r = 0.26, p < 0.001 for PF4, r = 0.23, p = 0.01 for NAP-2), confirming its role in reflecting basal platelet activation. sP-Selectin and sCD40L showed no significant correlation with any clinical outcomes, treatment group or platelet activation test whereas PF4 and NAP-2 were independently correlated with increased 24-hour bleeding volume (PF4: r =-0.19 p = 0.03; NAP-2: r=-0.17 p = 0.04).ConclusionPF4 is the most platelet-specific biomarker for predicting platelet reactivity and bleeding in patients receiving DAPT until the CABG. Plasma PF4 quantification could provide a simpler alternative to aggregation tests for predicting platelet reactivity, postoperative bleeding and could guide perioperative antiplatelet management.Graphical Abstract Introduction Platelets are central for maintaining vascular integrity and contributing to thrombosis and bleeding [1]. In the perioperative setting, platelet dysfunction and hyperactivation are critical determinants of both bleeding and thrombotic complications [2]. The combination of clopidogrel and aspirin (ASA) in dual antiplatelet therapy (DAPT) is widely recognized as a cornerstone in managing acute coronary syndromes and ischemic cardiomyopathy [3]. While continuing aspirin during coronary artery bypass grafting (CABG) has been shown not to increase postoperative bleeding risk, the safety of maintaining DAPT up to the day of surgery remains contentious [4] and a 5–7 day washout period of clopidogrel is usually recommended. Indeed, DAPT alters platelet activity and exacerbates bleeding risk [5]. Although antifibrinolytic agents are known to reduce bleeding in cardiac surgery, many studies lack consistent use of these agents or standardized infusion protocols, leading to significant heterogeneity [5]We evaluated DAPT versus aspirin in a population of patients undergoing non-redux programmed coronary surgery and at a distance from any acute coronary syndrome, treated intraoperatively with tranexamic acid (TxA), and whose degree of platelet inhibition was analyzed. We showed a reduced efficacy of TxA compared with aprotinin with increased red blood cell (RBC) transfusion and possibly increased mediastinitis in the DAPT group compared with the aspirin group [6]. In this latter study, bleeding was correlated with the extent of clopidogrel antiplatelet effect, with the best correlation for the vasodilator stimulated phosphoprotein (VASP) assay. Despite the widespread use of platelet testing, the interplay between platelet biomarkers in plasma, aggregation tests, and clinical outcomes remains incompletely understood [7]. Conclusions drawn from clinical studies are undermined by methodological shortcomings, particularly the failure to account for preoperative platelet activity and the variability of the patient response to antiplatelet agents. Thus, understanding and measuring platelet activation is crucial to an effective anti-platelet therapy. The reference method is light-transmission platelet aggregation (LTA) after induction by more or less specific agonists (for example adenosine diphosphate, ADP, for clopidogrel and collagen or arachidonic acid for aspirin). For instance, in patients undergoing non-cardiac surgery treated with DAPT or clopidogrel alone, platelet reactivity to ADP assessed by LTA was significantly associated with TIMI bleeding score [8]. However, other studies showed conflicting results. In real-life patients with ACS treated with clopidogrel and newer P2Y12 inhibitors, neither the VASP nor LTA measured at the acute phase helped predicting bleeding risk [9]. Moreover, ADP-LTA and VASP assays are surrogate markers specific of clopidogrel efficacy, do not specifically reflect the effect of aspirin, and cannot be performed in all center. Point of care tests are widely used but with varying and heterogeneous results. Another strategy is to measure circulating biomarkers to reflect the global platelet reactivity, sCD40L and soluble P-selectin (sP-selectin, so-called s-CD62P) being the most often used. They are transmembrane proteins present in platelet α-granules, exposed at platelet membrane upon activation. The soluble form is generated by proteolytic cleavage of both proteins, after platelet activation. However, several studies showed mixed results when comparing platelet aggregation to soluble plasma biomarkers [10–14]. For example, s-P-selectin plasma level did not differ when comparing increasing loading doses of clopidogrel in coronary artery disease patients, contrary to P-selectin expression on ADP-stimulated platelet surface membrane [15]. neutrophil activating peptide 2 (NAP-2) and platelet factor 4 (PF4 or CXCL4) are both biomarkers from chemokine/CXCL family. NAP-2, a proteolytic fragment of platelet basic protein (CXCL-7 or PPBP), is an important marker of platelet degranulation [16–19]. PF4 is synthesized mainly by megakaryocytes with diverse biological functions, particularly in promoting thrombosis [17, 20, 21]. In essence, these biomarkers serve as pivotal indicators of platelet activation and elevated levels of these markers in circulation signal intensified platelet activity. This study aimed to evaluate these four biomarkers of platelet activation during CABG in a prospective cohort of patients treated by antiplatelet agents and to predict bleeding events. Methods Patients 150 Patients were enrolled in the ICARE study (ClinicalTrial.gov: NCT01216150; Registration Date: December 2009) from December 2009 to November 2010 in a large academic hospital [6]. The ICARE study aimed primarily to prospectively the use of TxA during primary CABG in two groups: one group with DAPT (Clopidogrel 75 mg/day, aspirin 75 mg/day) another with aspirin (ASA, 75 mg/day)) alone. When clopidogrel was stopped more than 10 days before surgery, the patients were included in the ASA group. However, patients were excluded when clopidogrel was stopped less than 10 days before surgery. All other patients had their respective treatment until operation. Other inclusion and exclusion criteria and pre- and postoperative patient management are described elsewhere [6]. The primary clinical endpoint for bleeding events was the chest tube blood output at 24 h. The present study was designed to evaluate preoperative 4 biomarkers of platelet activation in the patient cohort. This study included 143 patients as biological samples were unavailable for the remaining [6]. The plasma from 20 healthy controls (10 males and 10 females [22, 23]) was sampled and frozen around the same time as ICARE samples and kept at -80 °C until analysis. Preoperative platelet reactivity assessment Platelet aggregation analysis was conducted within three hours of blood sampling using citrated platelet-rich plasma (collected immediately after anesthesia before the CABG, obtained by centrifugation at 150 g, 10 min at room temperature) adjusted to a concentration of 250 × 10⁹ platelets/L, utilizing a RA-8 V optical aggregometer (Soderel Medical). The agonists used to induce platelet aggregation were arachidonic acid (AA, 1.5mmol/L, Helena Biosciences), horm collagen (1 µg/ml; Nycomed), PAR1 activated peptide (PAR1ap, TRAP-6 or SFLLRN; TRAP-6, 20µmol/l, Stago) and ADP (10 and 20 µmol/L, Biopool). Both maximal and residual (after 6 min) platelet aggregation (MPA and RPA) were measured [6]. P-selectin at the surface of platelets were quantified in PRP with a quantitative flow cytometry with a calibrator kit (Platelet Calibrator, Biocytex), according to the manufacturer’s instructions. Platelets were coated with the anti–P-selectin mouse IgG1 mAbs (Immunotech, CD62p, clone CLB-Thromb/6) at 10 µg/mL. For platelet responses to agonist, 100µM TRAP-6 was incubated for 15 min at 37 °C with PRP samples. 5000 events were acquired on a FACScan flow cytometer and analyzed by CellQuest software (Becton Dickinoson) [22]. Complete results are presented in the original ICARE study [6]. Plasma biomarker quantification Citrated plasma obtained from residual blood samples was double-centrifuged (2300 g for 15 min) and stored at -80 °C until assayed. sCD40L, sP-selectin, PF4 and NAP-2 were measured using Luminex kit (R&D Systems) and the Bio-Plex 200 with Bio-Plex manager 5.0 software (Bio-Rad). Measurements were performed for all patients where a plasma sample was available, which represented 143 samples. Statistical analysis Data are expressed as median [interquartile range IQR 5th -95th ]. Due to the non-normality of the data set (p < 0.001, Shapiro-Wilk test), comparison between groups was performed with the Mann-Whitney U test and correlation between two variables were assessed using Spearman’s linear regression. A multiple logistic regression was performed with high bleeding (> 1000 ml at 24 h) as the dependent variable and age, sex and PF4 level normalized as independent variables. There was no formal correction for multiples testing, as the results are to be interpreted in an exploratory context. All analyses were 2-sided with a p-value < 0.05 considered statistically significant and were done with R (version 4.3.2). Results PF4 and NAP-2 are decreased in DAPT compared to ASA alone Platelet reactivity biomarkers were measured in 85 patients from the ASA group and 58 patients from the DAPT group. Before group comparison, platelet activation plasma biomarkers levels in patients were compared to healthy controls (Fig. 1). For each biomarker, the plasma level was significantly higher in the patients before CABG, compared to the healthy patients. Biomarkers level were then compared between the ASA group and the DAPT group. Median sCD40L levels were significantly lower for DAPT patients compared to ASA patients when adjusted to platelet count (2252 pg/ml IQR 149–6576 pg/ml vs. 2513 pg/ml IQR 579–10831, p = 0.04), but this difference did not remain significant when comparing sCD40L levels without normalization (Fig. 2A), the two groups had similar platelet count (227 G/L IQR 160–392 vs. 233 G/L IQR 143–377), as described previously [6]. Median sP-Selectin levels in plasma did not differ between ASA and DAPT groups, even when adjusting for platelet count (P > 0.05 for each, Fig. 2B). Both median PF4 and NAP-2 levels were significantly lower in the DAPT group compared to the ASA group (676 ng/ml IQR 268–1904 vs. 992 ng/ml IQR 551–2757, p < 0.001 for PF4; 167 ng/ml IQR 43–439 vs. 281 ng/ml IQR 108–9000 ng/ml p < 0.001 for NAP-2). These differences remained statistically significant when both biomarker levels were adjusted for the platelet count (p < 0.001 for each). Comparison of platelet activation biomarkers between pre-CABG patients and healthy controls. For each biomarker (sCD40L: A, sP-Selectin: B, PF4: C, NAP-2; D) plasma level was compared between patients from the ICARE study (left) and 20 healthy controls (right). The whisker of the box and whisker plot represent the 2.5–97.5 percentiles. ***: p < 0.001 Comparison of platelet activation biomarkers between treatment groups. For each biomarker (sCD40L: A, sP-Selectin: B, PF4: C, NAP-2; D) plasma level was compared between the group of patients receiving dual anti-platelet therapy (ASA + clopidogrel) or only ASA (left part of each graph). The right part of the graph shows the plasma level of the biomarker normalized to the platelet count (pg/ml per million of platelets). The whisker of the box and whisker plot represent the 2.5–97.5 percentiles. ASA: aspirin; DAPT: dual anti platelet therapy; ns: not significant; *: p < 0.05; **: p < 0.01; ***: p < 0.001 PF4 and NAP-2 are correlated to ADP- and collagen-induced platelet aggregation Next, correlations between plasma biomarkers and platelet aggregation were assessed (Table 1). sCD40L and sP-selectin showed no correlation (p > 0.05 for each) to MPA or RPA for any of the agonists used (PAR1ap, ADP at 10 and 20µmol/L, AA and collagen), except for a weak correlation of sCD40L to ADP20 RPA. Conversely, PF4 and NAP-2 were significantly and positively correlated with MPA and RPA after activation by ADP at the 2 concentrations used (10 and 20 µmol/L, RPA, PF4: r = 0.32, p < 0.001, NAP-2: r = 0.32, p < 0.001) and by collagen (RPA, PF4: r = 0.21, p = 0.01, NAP-2: r = 0.23, p = 0.01). However, no correlation was observed between PF4, and NAP-2 levels and platelet aggregation induced by either AA or PAR1ap (p > 0.30). Table 1Correlation of plasma platelet activation biomarkers with agonist-induced plateletaggregationsCD40-LsP-SelectinPF4NAP-2rp-valuerp-valuerp-valuerp-value PAR1apMPA-0.070.40-0.100.25-0.070.42-0.070.43RPA-0.010.95-0.090.300.040.640.030.69ADP 10 µmol/LMPA0.070.38-0.010.870.22 0.01 0.23 0.01 RPA0.140.100.030.750.32 < 0.001 0.32 < 0.001 ADP 20 µmol/LMPA0.130.13-0.060.480.26 < 0.001 0.26 < 0.001 RPA0.17 0.04 -0.010.920.36 < 0.001 0.36 < 0.001 AAMPA0.130.11-0.030.680.080.330.060.46RPA0.120.15-0.060.510.050.570.030.75CollagenMPA0.090.28-0.010.910.20 0.01 0.23 0.01 RPA0.110.18-0.010.900.21 0.01 0.23 0.01 Spearman correlation was used between biomarker plasma level and MPA and RPA in response to different agonists. Significative p-values (p < 0.05) are highlighted in boldMPA: Maximum Platelet Aggregation; RPA: Residual Platelet Aggregation (after 6 min); PAR1ap: Thrombin Receptor Activating Peptide; ADP: Adenosine Diphosphate; AA: Arachidonic Acid Correlation of plasma platelet activation biomarkers with agonist-induced plateletaggregation Spearman correlation was used between biomarker plasma level and MPA and RPA in response to different agonists. Significative p-values (p < 0.05) are highlighted in bold MPA: Maximum Platelet Aggregation; RPA: Residual Platelet Aggregation (after 6 min); PAR1ap: Thrombin Receptor Activating Peptide; ADP: Adenosine Diphosphate; AA: Arachidonic Acid PF4 is correlated to membrane P-selectin P-selectin platelet membrane expression was measured before (at rest) and upon activation with PAR1ap (Table 2). Again, we found no correlation between plasma levels sP-Selectin and membrane P-selectin (p > 0.15), and a weak correlation of sCD40L to the percentage of activated platelets regardless of PAR1ap activation. Conversely, P-selectin expression at the membrane of resting platelets (mean fluorescence and percentage of activated platelets) was correlated with plasma levels of both PF4 (r = 0.26, p < 0.001 for mean fluorescence, r = 0.32, p < 0.001 for percentage of activated platelets) and NAP-2 (r = 0.23, p = 0.01 for mean fluorescence, r = 0.26, p < 0.01 for percentage of activated platelets), with a higher correlation for PF4. After PAR1ap-induced platelet activation, only the correlation between sPF4 levels and the percentage of activated platelets remained statistically significant (r = 0.17, p = 0.04). Table 2Correlation of platelet biomarker level to P-Selectin platelet membrane expressionmembrane platelet P-SelectinsCD40-LsP-SelectinPF4NAP-2rp-valuerp-valuerp-valuerp-valueGeenceRest0.070.390.030.680.26 < 0.001 0.23 0.01 PAR1ap activated0.110.19-0.080.320.140.100.110.17Percentage of activated plateletsRest0.17 0.04 0.030.700.32 < 0.001 0.26 < 0.01 PAR1ap activated0.18 0.03 -0.010.900.17 0.04 0.140.10Spearman correlation was used between biomarker plasma level and platelet membrane P-Selectin mean fluorescence and percentage of activated platelets, at rest or after activation by PAR1ap Significative p-values (p < 0.05) are highlighted in boldPAR1ap: PAR1 activated peptide Correlation of platelet biomarker level to P-Selectin platelet membrane expression Spearman correlation was used between biomarker plasma level and platelet membrane P-Selectin mean fluorescence and percentage of activated platelets, at rest or after activation by PAR1ap Significative p-values (p < 0.05) are highlighted in bold PAR1ap: PAR1 activated peptide PF4 is the best biomarker correlated to bleeding and other clinical outcomes The association between the four platelet activation biomarkers and the primary clinical endpoint of the ICARE study, the chest tube blood output at 24 h, was assessed (Table 3). No significant correlation was found between bleeding and sCD40L or sP-selectin plasma levels (p = 0.81 and p = 0.26 respectively). Conversely, the chest tube blood output at 24 h negatively correlated with plasma levels of PF4 and NAP-2 (r=-0.19, p = 0.03 for PF4 and r=-0.17, p = 0.04 for NAP-2). The level of each biomarker was compared in patients who had or did not have a specific clinical event: RBC transfusion in the first 24 h postoperatively, platelet transfusion in the first 24 h, surgical re-exploration in the first 24 h, ventilation duration longer than 6 h, mediastinitis and readmission to the intensive care unit. Preoperative sCD40L was lower for patients who had ventilation for more than 6 h (p = 0.04). Preoperative PF4 was significantly lower in patients who subsequently returned to the operation room (n = 5, p = 0.03), as well as those who needed RBC transfusion (n = 19, p = 0.03) in the first 24 h, although this is secondary outcome related to bleeding severity. Furthermore, in a multivariate logistic regression analysis, PF4 levels emerged as the sole significant predictor variable (odds ratio: 0.50, p = 0.02) for high bleeding risk, defined as a chest tube blood output exceeding 1000 ml at 24 h, while age and sex were considered as additional independent variables. Table 3Platelet biomarker level according to clinical events Clinical eventsCD40-L (pg/ml)sP-Selectin (pg/ml)PF4 (ng/ml)NAP-2 (ng/ml)yesno p yesno p yesno p yesno p RBC Transfusion at 24 h559 [152–1889]561 [152–1073]0.4721,681 [20549–22814]19,984 [19139–20828]0.45760 [704–815]926 [828–1024] 0.03 192 [173–212]221 [183–258]0.10Platelet transfusion at 24 h567 [162–1875]522 [28-1073]0.3121,010 [19973–22047]19,970 [19103–20836]0.52828 [778–878]905 [805–1004]0.22210 [194–225]221 [183–259]0.33Surgical re-exploration at 24 h564 [162–1701]375 [286–568]0.1618,813 [18142–19484]19,984 [19087–20881]0.59553 [509–596]913 [818–1008] 0.03 167 [153–180]221 [184–257]0.06Ventilation duration >6 h516 [145–1459]621 [273–2282] 0.04 21,262 [20213–22311]19,955 [19165–20745]0.25895 [789–1001]913 [823–1003]0.05213 [173–253]221 [186–255]0.70Mediastinitis556 [158–1533]713 [289–2296]0.1918,601 [17513–19688]19,984 [19110–20858]0.71772 [633–911]896 [805–987]0.67245 [185–305]221 [187–254]0.74Readmission in ICU561 [153–1495]561 [289–2715]0.4918,274 [17207–19341]20,106 [19240–20972]0.44656 [538–773]925 [834–1017]0.13245 [193–298]221 [187–254]0.52 Correlation to bleeding volume at 24 h rp-valuerp-valuerp-valuerp-value-0.020.810.020.26-0.19 0.03 -0.17 0.04 Spearman correlation was used between biomarker plasma level and total postoperative bleeding volume after 24 h. Mann-Whitney U tests were performed between plasma level of the four biomarkers for patients who subsequently did or did not have one 2 of the clinical events in the table. Data are expressed as median [IQR 5th − 95th]. Significative p-values (p < 0.05) are highlighted in boldRBC: red blood cell; ICU: intensive care unit Platelet biomarker level according to clinical events Spearman correlation was used between biomarker plasma level and total postoperative bleeding volume after 24 h. Mann-Whitney U tests were performed between plasma level of the four biomarkers for patients who subsequently did or did not have one 2 of the clinical events in the table. Data are expressed as median [IQR 5th − 95th]. Significative p-values (p < 0.05) are highlighted in bold RBC: red blood cell; ICU: intensive care unit Discussion This study provides significant insights into the relationship between plasma biomarkers of platelet activation, platelet aggregation tests, and clinical outcomes in patients undergoing CABG. The results highlight the potential of soluble PF4 as a relevant biomarker in understanding platelet behavior and predicting perioperative complications, particularly in the context of antiplatelet therapy. Many authors reported that platelet reactivity after cardiovascular or neurological procedures is an independent predictor of bleeding events [24–27]. Others did not find an association between the pharmacodynamics of clopidogrel and the bleeding risk [28, 29]. However, most of these studies used assays evaluating primarily the effect of anti-P2Y12 agents. On the contrary, our results showed a correlation between PF4 level and collagen-and ADP-induced aggregation, respectively reflecting ASA and clopidogrel effect. Therefore, PF4 emerged as a relevant marker, with decreased levels observed in patients treated with DAPT compared to ASA. These findings align with existing literature that identifies PF4 as a critical component of platelet activation, released upon alpha-granule secretion during platelet degranulation [30]. The marked decrease in PF4 in DAPT-treated patients suggests a less pronounced platelet activation state, which is likely compounded by the pharmacological effects of clopidogrel and aspirin. Interestingly, sCD40L and sP-selectin levels in plasma did not significantly differ between the treatment groups. This discrepancy may reflect differing release kinetics or sensitivity to antiplatelet agents, warranting further investigation. First, we showed that sP-selectin plasma level did not differ between the DAPT group and the ASA group, whereas sCD40L was lower in the DAPT group when plasma level was normalized to platelet count. Both sPF4 and sNAP-2 were lower before and after adjusting to platelet counts. In line with this result, Kaufman et al. found that sP-selectin dropped in patients receiving ASA but was not a useful marker in stable CAD patients receiving clopidogrel [31]. Seyfarth showed that sP-selectin plasma level did not differ when comparing 300 or 450 mg loading doses of clopidogrel in coronary artery disease patients, contrary to P-selectin expression on ADP-stimulated platelet surface membrane [15]. The correlation of plasma PF4 and NAP-2 to platelet membrane P-selectin before activation by PAR1ap as well as the correlation between plasma PF4 and NAP-2 levels and platelet aggregation, particularly ADP- and collagen-induced responses, underscores the interconnected nature of platelet activation pathways [32]. ADP is a well-established mediator of platelet aggregation, acting through P2Y12 and P2Y1 receptors, while collagen binds to receptors GPIV and α2β1 [33]. The association of PF4 and NAP-2 with these aggregation pathways reinforces their role in amplifying platelet reactivity, which may be clinically significant in DAPT-treated patients [34]. This in contrast with plasma levels of sP-selectin, which did not show any correlation with platelet membrane P-selectin or platelet aggregation in response to agonists. We observed no correlation between PF4 plasma levels and aggregation in response to either PAR1ap or AA. For patients under aspirin, cyclooxygenase-1 activity is fully suppressed and the response to AA is very low, limiting residual variability [35]. On the other hand, PAR1ap induces a near-maximal platelet activation regardless of treatment, which reduces the variability between patients and weaken correlations with plasma biomarkers [36, 37]. The observed link between PF4 levels and bloodloss highlights the role of platelet activation in preventing bleeding [38]. DAPT, by reducing thromboxane A2 and ADP-dependent pathways, impairs platelet aggregation and thrombus formation, and PF4 can be a biomarker to help assess the efficacy of this inhibition. Plasma biomarkers such as PF4 and NAP-2 could be incorporated into perioperative risk assessment algorithms, with other clinical and biological markers, complementing functional platelet tests. By identifying patients at higher risk for bleeding or thrombotic complications, clinicians could tailor antiplatelet therapy and transfusion strategies to individual needs. While no assay is currently available to measure plasma level of those biomarkers in a clinical kit context, ELISA kits would be adapted to clinical use. Other biomarkers have been suggested to evaluate platelet reactivity. The serum level of thromboxane B2 (TxB2), the stable metabolite of TXA2, is an excellent surrogate marker of ASA effect that also reflects the inhibition of platelet amplification upon clopidogrel by a smaller but detectable decrease. However, the pre-analytical conditions and method of quantification are complex. In a recent randomized study on healthy volunteers treated by DAPT or clopidogrel [39], neither treatment significantly influenced beta thromboglobulin or sP-selectin levels. DAPT abolished and clopidogrel reduced thromboxane B2. Overall, the effects on markers of platelet reactivity and coagulation activation (D-dimer) of DAPT compared with clopidogrel were comparable. However, the ARIES study highlighted a non-inferiority of placebo compared to aspirin in patients with an Heartmate3 LVAD, with more patients free of hemocompatibility-related adverse events at 12 months, mainly driven by a decrease in bleeding events [40]. Those results remain valid, even when considering only the patients on ASA before implantation, due to a prior history of arterial vascular disease [41]. Comparing plasma levels of these biomarkers in light of treatments could be valuable to optimize antiplatelet strategies for patients on mechanical support. Other biomarkers have been suggested such as mean platelet volume, with a prognostic value similar to that of troponin for long-term ischemic but not bleeding outcomes in patients with acute coronary syndrome [42]. Many studies have also suggested quantification of platelet microparticules but the method is time-consuming, the results may vary according to the cytometer used and on the analysis of the different microparticule populations. Same for platelet-leukocyte aggregates that need a rigorous laboratory methodology for reliable and reproducible results, as shown elsewhere [43]. An 2024 international consensus statement recommend the use of platelet function and genetic testing for a guided selection of oral P2Y12 inhibitors, and the use of ticagrelor or prasugrel for clopidogrel poor-responders [44]. However, currently very limited data exist on plasma platelet biomarkers for patients on ticagrelor, with lower sP-selectin levels compared to patients on clopidogrel in diabetic patients [10]. Studying platelet plasma biomarkers for patients on ticagrelor or prasugrel compared to other antiplatelet agents such as clopidogrel and aspirin, would help understand how they can be used to predict clinical events and personalize antiplatelet strategies. Despite the findings, this study presents some limitations. First, some of the clinical events were rare: for example, only 5 returns to the operating room out of 143 patients. Therefore, those results should be taken with caution. Furthermore, the biomarkers measured here, and the correlation found are based on a very specific set of patients under antiplatelet therapy. Moreover, the sample size, while sufficient for identifying key associations, may limit the generalizability of findings to broader populations or other surgical contexts. Furthermore, the higher treatment failure rate of clopidogrel compared to ticagrelor or prasugrel makes interpretation more difficult. Further investigations should focus on confirming these results in extensive, multicenter studies while delving into the underlying mechanisms that connect PF4, platelet aggregation, and bleeding and to define an appropriate threshold. Additionally, studies evaluating the impact of targeted interventions on these biomarkers could provide valuable insights into improving perioperative outcomes. All in all, this study underscores the importance of plasma biomarkers, particularly PF4, in understanding platelet activation and predicting clinical outcomes. Plasma PF4 may complement platelet reactivity assessment in research settings. Further studies are needed to define its potential clinical utility and validate its role across broader patient populations. Author contributions Joseph Roux de Bézieux: Performed experimental manipulations, data acquisition, contributed to data analysis and wrote the paper.Julien Amour: Contributed to study conception and design; assisted in data interpretation and drafting of the manuscript.Aurélien Philippe, Jeanne Rancic, Nicolas Gendron, Sébastien Bertil, Marc Garnier: Contributed to data collection, validation, and critical revision of the manuscript.Pascale Gaussem: Conception and design of the study and drafting of the manuscript.David M. Smadja: Principal investigator, overall study supervision, project administration, and wrote the manuscript. Funding Grant from Inserm. Data availability Data available upon request to corresponding author. Declarations Ethics declaration Study obtained ethical committee approval from Comité de Protection des Personnes VI - Ile de France Registration Date: December 2009 and has been published on ClinicalTrials.gov ID NCT01216150. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Joseph Roux de Bezieux and Julien Amour share co authorship. Pascale Gaussem and David M. Smadja contributed equally to this work. References Associated Data Data Availability Statement Data available upon request to corresponding author.