Our results show a lower rate of bleeding in patients treated with argatroban when compared to heparin administration. Moreover, we found a significantly lower frequency of hemorrhage only at the tracheal stoma site. A comparable incidence of gastrointestinal tract, intramuscular, upper respiratory tract, cannulation site, intracranial, and pulmonary bleedings has been demonstrated in both groups. However, a slightly lower frequency of bleeding events was observed in the upper respiratory tract and at the cannulation site, although this difference was not statistically significant. Data suggested that argatroban may be associated with an increased risk of parenchymal bleeding (in the brain and lungs), while unfractionated heparin appeared to correlate more with bleeding at sites of invasive procedures (cannulation site, tracheal stoma, upper respiratory tract) but other risk factor could have interfered with this result.
However, among previous studies, only Fisser and coworkers have directly compared the incidence of bleeding at different sites between anticoagulant treatments, but they registered not statistically significant difference between argatroban and unfractionated heparin [7]. We must consider the possibility that this observation could be influenced by a bias in our study design.
In a recent meta-analysis, that has compared both clinical and biochemical outcomes of patients treated with non-heparin anticoagulants because of the spreading out of an acute HIT, argatroban was judged successful in obtaining a full platelets recovery. Indeed, in this meta-analysis, the effectiveness and safety outcomes were similar among various non-heparin anticoagulants [10].
As suggested by Martucci et al. in the recent publication of PROTECMO study, maintaining the aPTT at the lower end of the recommended range may be more effective than switching anticoagulants [11].
Even though the sample size was very small, our data show an overall similar risk of bleeding events between unfractionated heparin and argatroban. However, the total number of bleeding events in the UFH group (58 events over 401 days of ECMO) was significantly higher compared to the ARG group (21 events over 648 days of ECMO).
This result has been obtained despite the huge difference in platelet count between groups, suggesting that low platelet count would not be a major risk factor for bleeding on veno-venous ECMO.
Furthermore, we have investigated the incidence and severity of thrombocytopenia as a secondary outcome of interest in our study, and we have been able to demonstrate a positive association between argatroban infusion and platelet count compared to heparin. Moreover, heparin therapy was associated with a more severe initial drop in platelet count compared to argatroban infusion. However, the platelet count at the time of drug switch was significantly lower compared to the beginning of heparin infusion.
Wilcoxon rank-sum tests revealed a statistically significant difference in platelet count, platelet count at baseline (T0), platelet count at day 5 (T5), change in platelet count, and d-dimer levels.
The baseline platelet counts in patients starting heparin versus argatroban differed due to several factors. First, the platelet counts of patients receiving argatroban were already reduced by previous anticoagulant treatment, by the approximately 11 days of extracorporeal membrane oxygenation support, and by a longer duration of COVID-19 infection. For these reasons, this difference should be confirmed by other prospective studies.
Similarly, the statistically significant difference in platelet count observed on day 5 of infusion could be an effect of the longer duration of ECMO support and COVID-19 infection.
For this reason, we evaluate the platelets count drop between day 5 and day 0 of infusion. It revealed a statistically significant difference, that could suggest a protective effect on platelets of argatroban compared to heparin, but surely confirmed that argatroban is not inferior. This evidence could be ascribed to heparin associated thrombocytopenia (HAT or Type 1 HIT), but in our study a reduced PLT drop in ARG group could be addressed to preexisting thrombocytopenia, or even other condition which may have had influence but have not been investigated. This opinion is confirmed by linear regression of PLT T5 on PLT T0 and anticoagulant, that reveals a correlation of PLT in day 5 to initial PLT count more than to anticoagulant.
However, Fisser demonstrated a reduced consumption of platelet packs per day in patients managed with argatroban, suggesting that this drug could play a role in preserving platelet count during veno-venous ECMO [7].
As the same, in a recent study on 57 patients, Menninger et al. demonstrated that argatroban was superior to heparin alone in preserving oxygenator from exhaustion and in reducing blood products transfusion rate [9].
The difference in the platelet drop between two groups is interesting and is sustained by the linear regression analysis conducted on relationship between median PLT and both ECMO days and LOS.
This linear regression analysis reveals that in UFH group there was a negative correlation between ECMO days or LOS and PLT count. In our opinion, this evidence could be sustained by the fact that patient with shorter run of ECMO and shorter stays generally survived and were easily weaned. On the other hand, patients with longer runs and hospitalization time had more complications, and lower median thrombocytopenia could have been the effect of several factors (i.e. sepsis or circuit and oxygenator degeneration).
Differently, in ARG group, linear regression analysis reveals a positive correlation between median PLT count and ECMO days or LOS. In our opinion this phenomenon could be the result of two different mechanism. First, argatroban does not sustain a dose dependent platelet aggregation like heparin does activating spleen clearance. Second, argatroban has a protective effect on oxygenator and circuit lasting as other authors have already demonstrated, despite in our study ECMO runs under argatroban were significantly longer than under UFH [7, 9].
In this study, there was not statistically difference in mortality between groups, even if in ARG group 16 patients died compared to 8 patients in UFH group. This difference could easily be explained considering that worser patients with a longer ECMO run switched from UFH to ARG group increasing mortality in the second one, because runtime is considered an independent variable for mortality during ECMO support.
The recent annual ELSO guidelines report clots occurrence in the oxygenator in nearly 13% of patients. Therefore, a continuous pharmacologic anticoagulation is suggested to guarantee a correct use of the extramembrane oxygenation support by counterbalancing the effects of exposure to the non-endothelial surface of the ECMO circuit, inhibiting activation of platelets, coagulation, and inflammatory pathways without increasing the risk of bleeding.
Despite the possible advent of these collateral effects due to the application of ECMO circuits, ELSO guidelines recommend the administration of a systemic anticoagulation therapy for all those patients undergoing ECMO support. ELSO guidelines still consider as first choice anticoagulant therapy unfractionated heparin, but they do not exclude the possibility to use direct thrombin inhibitors (DTIs) as second choice drugs, moreover in specific conditions like demonstrated type 2 HIT. Between DTIs, literature explored efficacy of bivalirudin, lepirudine and argatroban and ELSO guidelines still consider an off-label use for this anticoagulants. Argatroban has been highlighted as a drug able to induce and even maintain a therapeutic anticoagulation in case of either confirmed or suspected type 2 HIT. Activated partial thromboplastin time (aPTT) is used to monitor argatroban therapy using a range of 1.5 to 3 times the patient's baseline aPTT, up to a maximum of 100 s [13, 14].
However these guidelines do not exclude the possibility of a management without any systemic anticoagulation therapy, if bleeding cannot be controlled neither through a pharmacological approach, nor throughout surgical methods [15, 16].
Although ECMO protocols, endorsed by the Extracorporeal Life Support Organization (ELSO) have routinely included the use of systemic anticoagulation with the primary goal of minimizing circuit thrombosis [8, 9], Olson et al. systematized the evidence about anticoagulant – free veno-venous ECMO support, concluding that the incidence of thrombosis was comparable to patients receiving continuous systemic anticoagulation [11, 12].
They found an overall incidence of thrombosis (patient or circuit) of 21.9% and an overall incidence of bleeding of 32.8% among patients of reviewed studies [17, 18].
In a health assessment study, Cho et al compared course costs of patient in argatroban and unfractionated heparin, showing that the mean cost per course with argatroban was significantly lower than one with unfractionated heparin, without an increasing in adverse events correlated to anticoagulant therapy [8].
Contrasting with data reported in literature, in this study we observed a lower incidence of circuit thrombosis and clotting of the oxygenator compared to previous studies, even if median time of use of circuit and oxygenator was longer and median aPTT ratio was near to 1,5 times normal ratio15. This evidence suggests that anticoagulation in veno-venous ECMO could be used at lower dosage for prophylaxis [9, 10, 18, 20].
Data on Table 1 demonstrated that there were not differences between groups characteristics except for ECMO days and LOS – ICU. This difference is due to the clinical protocol of anticoagulant management in our institute, where in case of progressive platelets decrease, intensivists could suspend heparin in the suspect of HIT to protect residual platelets and start argatroban infusion. Statistical analysis on this phenomenon has been considered useless because it represents a bias, patients enrolled in ARG group had necessarily a longer ECMO run and total LOS due to the number of days under heparin and after under argatroban.
Limitations
Our study is not free from some limitations.
Its first limitation is the design: it is a retrospective, monocentric study and as in other existing studies the treatment strategy was not decided before study started, and results could be warped.
Another limitation is the relatively small sample, that is due to the difficulties of recruiting we had to face during the critical pandemic event.
One more limitation is due to our internal clinical protocol for anticoagulant management therapy, all patients enrolled for anticoagulation with argatroban was previously treated with unfractionated heparin.
Moreover, patients who underwent argatroban infusion due to lowering platelets count, suffered from a mean length of stay and ECMO treatment of more than 15 days, and direct comparison of drug impact is skewed. However, our results suggest that also in a disadvantageous situation, argatroban has been useful to maintain sufficient total platelet count without a significantly increase of bleedings.