This study focused on the efficacy, safety and controllability of argatroban in patients with ARDS and extracorporeal lung assist devices (ECMO or pECLA). Overall, both anticoagulants enabled therapeutic aPTT values in critically ill patients and were suitable to operate the extracorporeal lung assist device sufficiently and safely, even for a longer period of time. No differences were found between the two groups with respect to bleeding, thrombosis or transfusion. The controllability of both anticoagulants significantly improved over the course of time, with an increasing probability to reach the targeted aPTT corridor.
The present study included patients treated for ARDS in a single center. Matching for standard parameters and severity of illness identified an appropriate control group. ARDS patients in this study were characterized by relatively severe medical conditions, as reflected by high median APACHE II and SAPS II scores at hospital admission. On average, severity scores were higher than in comparable recent studies [13] and even higher than in the ALIVE study [14]. Also, parameters of pulmonary gas exchange and invasiveness of mechanical ventilation indicated severe lung failure. This might be due to our study population, which consisted of very severely affected patients admitted to our ARDS center specifically because their medical condition is poor. On average, more than two-thirds of our patients are transferred from other hospitals [15].
Argatroban was started in 13 patients due to a clinical diagnosis of HIT, presenting with typical thrombocytopenia and positive laboratory testing for anti-pF4 antibodies. This represents about 4% of ARDS patients undergoing extracorporeal lung support, a percentage similar to that reported by others [16]. It is estimated that up to 5% of patients who receive heparin are at risk to develop HIT [4]. However, in the present study, the majority of patients who received argatroban were either non-responsive to heparin or HIT was suspected but not confirmed by laboratory testing. There are many reasons why ARDS patients on extracorporeal lung support may develop a decrease in platelet count; for example, thrombocytopenia might arise due to sepsis, platelet activation or a certain consumption by the extracorporeal circuit. However, the differential diagnosis of HIT in critically ill patients remains a challenge [17] and was not in the scope of our study. Association of a positive 4T score, anti-PF4 and aggregability tests is mandatory to confirm HIT. The threshold to change anticoagulation from standard heparin to argatroban on our ICU is relatively low. Further, patients who were once switched to argatroban upon suspected HIT with ultimately negative HIT testing were not necessarily switched back to heparin. This might explain the relatively high rate of patients receiving argatroban in our study population.
For patients with HIT, the approved initial dose of argatroban is 2 µg/kg/min. Our study population, comprising critically ill ARDS patients with concomitant organ failure, required substantially lower doses of (approximately) 0.26 µg/kg/min argatroban to achieve the therapeutic aPTT goal. This effect might be even more pronounced when the number of failing organ systems increases [18]. Our data are consistent with these findings and also support other studies, which also recommended much lower argatroban doses in critically ill patients [13, 19,20,21]. There are no specific guidelines for dosing in ARDS patients on extracorporeal gas exchange. In our center, the established starting dose is 0.3 µg/kg/min. The baseline aPTT value, the evaluation of hepatic impairment and the presence of multi-organ-system failure should be taken into account when starting argatroban.
Although major and minor bleeding complications were similar in both groups, there were significantly fewer major bleedings than minor bleedings in both groups. The rates seen in our study are similar to those reported by others [16]. Major bleedings (e.g., pulmonary or intracranial bleedings) were relatively rare events; we found four cases of severe intracranial bleeding, which corresponds to 5% of all our assessed patients. The ELSO reports rates of major intracranial hemorrhage of up to 4% [22]; in most of these cases, the outcome is deleterious. The slightly higher rate found in our study might be explained by the more severe medical condition of our patients, as evidenced by high APACHE and SAPS II scores. Also, impairment of the coagulation system due to sepsis or secondary organ failure, and additional anticoagulation, might render ARDS patients prone to intracranial hemorrhage. However, in this respect, we found no significant differences between argatroban and heparin.
Thromboembolic events are potentially harmful or even life-threatening events in patients with ARDS. Although such events are a major complication of extracorporeal lung support therapy, data on the clinical significance are sparse [23]. As compared to bleeding episodes, thromboembolism rarely occurred in the present study; also, there was no fatality due to thrombotic failure of the extracorporeal lung support device. In the study of Weingart et al., about 25% of patients on extracorporeal lung support needed at least one replacement of the membrane oxygenator [24]. In our study, about 40% of patients needed such replacements. By trend, replacements of the pECLA oxygenator occurred more often in the argatroban group than in the heparin group. We did not observe this with regard to the ECMO oxygenator. However, the overall number of affected patients did not differ significantly between the two groups. Blood flow is much lower in pECLA than in ECMO; also, there may be areas with relative stasis, which can lead to thrombus formation. Nevertheless, the desired ranges of aPTT values in patients with a running system of either kind are comparable for pECLA and ECMO. Therefore, we decided to include both devices in our study.
Interestingly, we found no significant correlation between aPTT values and bleeding or thromboembolic events. Patients with aPTT values exceeding the desired therapeutic range did not necessarily develop bleeding complications. Vice versa, patients with aPTT values below the targeted therapeutic range did not necessarily develop thromboembolic events. However, practically all the thromboembolic events occurred when minimal aPTT values were below 50 s. This is in line with Trudzinski et al., who found an aPTT level below 50 s to be predictive for thromboembolism in ARDS patients under ECMO [25]. On the other hand, almost all bleeding events occurred when the maximal aPTT of the patients was above 50–60 s. Strikingly, more than two-thirds of all bleeding events were associated with maximum aPTT values above 75 s. Although these results only reflect the population of this study, we propose that this aPTT value could be regarded as the utmost tolerable upper limit under extracorporeal lung support to avoid hemorrhage. However, management of anticoagulation in ARDS patients with extracorporeal lung support is not clearly defined. On the basis of our findings, we suggest aPTT levels around 50 s, which is in accordance with most other recommendations and ELSO guidelines suggesting a range 1.5–2.5 times baseline value [19, 21, 24,25,26]. Development of bleeding might be multifactorial in the intensive care setting and not only dependent on anticoagulation. In particular, impairment of liver function might be held responsible for dysfunction of the coagulation system [27]. Also, volume overload or right-sided heart failure is a typical complication in ARDS patients, which might ultimately lead to hepatic congestion. Overall, in the present study there was no difference between argatroban and heparin with respect to bleeding. Also, there were no significant differences in blood transfusions (a surrogate for blood loss). Our results for ARDS patients on ECMO or pECLA are similar to data from a recent retrospective study based on a prospective cohort database which found an average of one red blood cell per patient-day on VV-ECMO [28]. Thus, argatroban appears to be a safe and suitable anticoagulant in critically ill ARDS patients on extracorporeal lung support.
Finally, we evaluated the controllability of argatroban. Firstly, it enabled therapeutic aPTT values to adequately perform extracorporeal lung support, even for a long time. However, repetitive measurements of the aPTT to adjust for the aPTT goal were significantly more often necessary in the argatroban group in the first four days of therapy. Only a few comparable studies are available on this topic. For example, Treichl et al. [29] report up to three necessary dose adjustments in every patient in order to achieve a distinct aPTT goal. Most other studies do not report details on dose adjustments or aPTT monitoring. Nevertheless, in our study, the controllability of argatroban significantly improved over the first few days, with an increasing probability to reach the targeted aPTT corridor. The mechanism of ‘PTT confounding’ might have been an interfering factor. This condition is defined as a situation where patient-related clinical factors may result in changes in aPTT values that are misleading with respect to indicating the true level of anticoagulation [30]. As a consequence, dosing of argatroban might have been insufficient in some cases.
The present study has some limitations. First, it is retrospective and, although it is one of the largest studies on this topic, the sample size is relatively small. Thus, the study may be underpowered to detect significant differences in mortality, bleeding outcomes or transfusion. Therefore, generalization of these results to other patients undergoing extracorporeal lung support requires considerable caution. Also, in this study the indication for extracorporeal lung support was lung failure. Extracorporeal lung support was primarily cannulated veno-venously, and the results may not allow conclusions to be drawn in the case of a veno-arterial scenario. The patients of this study had very high severity scores. Therefore, there might be different results or thresholds in other patient populations. Another major limitation is that bleeding episodes and thromboembolic events were manually extracted from the patients’ files, which can result in underestimation of the overall incidence. There may be residual confounding that is not captured, such as confounding by indication or indicator. Finally, an additional limitation of the present study is the absence of other parameters of anticoagulation used.