The interaction of thrombocytopenia, hemorrhage, and platelet transfusion in venoarterial extracorporeal membrane oxygenation: a multicenter observational study

Hemorrhage is an important and frequent complication during extracorporeal membrane oxygenation (ECMO). Although changes in anticoagulation strategies and improved circuit technology have led to a reduction in its occurrence rate, it remains a significant problem attributing to worse patient outcomes [1]. Active bleeding is one of the main indications to supply coagulation factors or transfuse blood products, including platelets. Guidelines by the Extracorporeal Life Support Organization (ELSO) state that as a spontaneous bleeding can occur at a platelet count of 20·10⁹/L, the advised threshold for platelet transfusion is 80·10⁹/L [2]. However, evidence-based guidelines are lacking, resulting in a wide range of thresholds used in daily practice [3].

Thrombocytopenia, defined as a platelet count < 150·10⁹/L [4], is common in patients supported with venoarterial ECMO (VA ECMO). Approximately one in five patients develops a thrombocytopenia during their ECMO support, independent of ECMO duration [5, 6]. Its etiology is complex and multifactorial, consisting of platelet activation and consumption by the extracorporeal circuit, hemodilution, hemorrhage, and the usage of platelet count and function altering medication such as platelet inhibitors and heparin [2, 7]. Importantly, thrombocytopenia is found to be a risk factor for hemorrhage and even mortality [8, 9].

In VA ECMO, the platelet transfusion rate has been described in a few retrospective studies, showing a range of 4% up to 62% [8, 10, 11], and has been associated with an increased risk of mortality [8]. As such, hemorrhage, thrombocytopenia and platelet transfusion are intertwined and associated with mortality. However, up to now, most studies are limited to a single-center design, therefore limiting generalizability. Moreover, correction for confounding factors is often limited or insufficiently described. Although in 2020 a meta-analysis was performed, conclusions were limited due to the high level of heterogeneity, as well as the very limited number of studies describing thrombocytopenia or platelet transfusion in VA ECMO [5].

Therefore, the aim of this international multicenter study was to ascertain the course and occurrence rate of thrombocytopenia, and to assess the association between thrombocytopenia, hemorrhage and platelet transfusion during VA ECMO.

Of the total of 433 patients, 419 were eligible for further analyses (Additional file 1: S3. Flowchart). At ICU admission, median platelet count was 179·10⁹/L [119–253·10⁹/L]. An overview of baseline characteristics can be found in Table 1. To highlight, most patients had a peripheral cannulation configuration (N = 356, 86%), and over half of the patients was cannulated using a surgical technique (N = 227, 56%). The nadir platelet count at the first day of ECMO showed an average 49·10⁹/L [6–102·10⁹/L] decrease when compared to the last known value before cannulation.

Nearly all patients developed a thrombocytopenia during ECMO (398/418, 95%), of which two third already during the first day (273/406, 67%, missing = 13). During ECMO, lowest platelet count was < 50·10⁹/L in 179 patients (43%), 50–100·10⁹/L in 159 patients (38%), and 100–150·10⁹/L in 60 patients (14%). An overview of the degree of thrombocytopenia is shown in Fig. 1. The largest proportion of severely thrombocytopenic patients was found at day 5, when 70 out of the remaining 244 patients had a platelet count < 50·10⁹/L (29%). With the exception of the yet severely thrombocytopenia patients, the platelet course showed an initial decrease in platelet count, followed by a stabilization and further platelet recovery, as shown in Fig. 2. This same trend was found independent of overall hemorrhage and transfusion status (Additional file 1: Fig. S1).

In comparison with the patients with a nadir platelet count > 100·10⁹/L, patients that developed a severe thrombocytopenia during ECMO had a higher lactate level, higher SOFA score and lower platelet count before ECMO initiation (all adjusted P < 0.05). In addition, with a median of 6 days [4–9], the total days of ECMO were longest in patients with a severe thrombocytopenia during ECMO (adjusted P < 0.01). Unadjusted 28-day mortality was highest in patients that suffered a severe thrombocytopenia, significantly lower than patients with other degrees of thrombocytopenia (P < 0.01).

Thrombocytopenia, hemorrhage and platelet transfusion are common in patients supported with VA ECMO. This study aimed to describe the occurrence rate and course of thrombocytopenia during VA ECMO, and further assess the association between thrombocytopenia, hemorrhage and platelet transfusion. We report the following clinically relevant findings. First, thrombocytopenia has a high frequency during VA ECMO, whereas almost all patients suffered from a thrombocytopenia. Second, after a global initial decrease in platelet count, stabilization occurred. Third, half of the patients received a platelet transfusion, of which the frequency and number of transfusions administered increased along with the severity of the thrombocytopenia. Severe thrombocytopenia has the strongest association with receiving a platelet transfusion, also after adjustment for confounders. Lastly, the presence of hemorrhage increased the odds to receive a platelet transfusion in all degrees of severity of thrombocytopenia. However, also in the absence of bleeding, a significant part receives a platelet transfusion.

A striking result of this study is that, contrarily to current literature, the incidence of thrombocytopenia in our cohort is significantly higher than previously described. In the systematic review and meta-analysis of Jiritano et al., a pooled incidence was found of 23.2%, based upon the results of six studies [5, 9, 13‐17]. However, all these studies were performed in a single-center design, and the heterogeneity among these studies was high. For example, the definition of thrombocytopenia was not always provided [14, 15]. In addition, patient populations consisted of ECPR, refractory cardiogenic shock, bridging for left ventricular assist device and “ECMO in general” [9, 13‐17]. This discrepancy in patient populations and methodology could explain part of the differences found with our cohort. Another possible explanation could lie in the fact that the hemorrhage rate reported in our cohort is higher than what is currently stated in large ELSO cohorts, despite using the same definition [1].

We confirm the platelet course consisting of an initial decrease, followed by a stabilization in platelet count, in case a severe thrombocytopenia was not yet present at ECMO initiation [9, 18]. This initial decrease in platelet count can be multi-factorial in etiology, in which VA ECMO adds to the equation by increased platelet consumption in the device itself as well as due to increased shear stress [19‐21]. The recovery may be the result of changing patient conditions and the fact that platelets have a mean lifespan of 7 to 10 days [22]. Factors complicating this recovery can include hemorrhage, which is a serious problem that can occur at any moment during ECMO. In addition, previous studies have described an increased incidence of heparin-induced thrombocytopenia (HIT) when compared to similar patient groups [12]. Currently, studies evaluating the degree and type of anticoagulation during ECMO are being performed, adding information in finding the optimal balance among anticoagulation, hemorrhage, and possibly related thrombocytopenia [23].

In our cohort, just over half of the patients received one or more platelet transfusions during VA ECMO, an occurrence rate that is in line with a recent single-center study [8]. Of note, a considerable part of the non-bleeding patients receives a platelet transfusion, of whom a large part at the time of severe thrombocytopenia. In general, indications for platelet transfusion consist of the treatment or prophylaxis of bleeding, such as in case of an invasive procedure or in the presence of severe thrombocytopenia [24‐26]. It can be hypothesized that in the transfused non-bleeding patients, the transfusion was administered prophylactically. The benefits of prophylactic platelet transfusion can be questioned. Questions have been raised regarding the increment of platelet transfusion, and whether the benefits and possible risk reduction in bleeding outweigh the downsides of platelet transfusion. Downsides of platelet transfusion include the risk of transfusion-related sequelae in an already vulnerable patient population, next to increasing scarcity and costs of transfusion products [24]. Therefore, a transfusion should be carefully considered. Recently, a randomized controlled trial (RCT) was published focusing on prophylactic platelet transfusion before central venous catheter placement in severely thrombocytopenic critically ill patients in the ICU and hematology ward. They found that withholding platelet transfusion resulted in more placement-related bleeding [27]. As such, a personalized approach may be preferable, differing per patient group and prophylactic indication.

The minimum threshold for platelet transfusion in bleeding and non-bleeding critically ill patients has been a topic of discussion. In ECMO, the variance in the range of thresholds as well as the absolute threshold have been shown to be higher when compared to other critically ill patient populations not supported with ECMO [3]. Our study confirms this wide range of thresholds, reflected by the eight different thresholds that were used in our 16 centers. Evidence from randomized trials is scarce, and current guideline recommendations are based on observational studies [25]. Again, which degree of thrombocytopenia to accept, as well as when to transfuse, is a delicate balance. No studies have been performed assessing the threshold for platelet transfusion during VA ECMO, and to our knowledge, none are scheduled in the near future. In addition, during ECMO, there is not only a shortage in absolute platelet count, but impaired platelet function may also be present [28]. Combining both an absolute threshold with platelet function tests may be helpful in the decision when to transfuse.

To our knowledge, this study is the first multicenter collaboration and one of the largest cohort studies focusing on thrombocytopenia during VA ECMO. In addition to the observational data, institutes’ protocols were taken into account, confirming the wide range of thresholds applied in current practice. However, some limitations should be acknowledged. As the data are retrospective, it should be emphasized that no conclusions regarding causality can be drawn. Additionally, no data were collected regarding the indications for platelet transfusion (i.e., prophylactic for a certain procedure), the occurrence of transfusion-related complications such as transfusion-associated circulatory overload or transfusion-related acute lung injury, as well as HIT. Lastly, no data on platelet function or drugs possibly influencing platelet function were collected, as protocols lack standard use of those tests, and differences between centers in the usage and type of function testing were considered too large.

The occurrence of thrombocytopenia is considerably higher than currently recognized in VA ECMO. Severe thrombocytopenia is an important factor for platelet transfusion, also in the absence of bleeding. It is clear that future research in VA ECMO should focus on the etiology of thrombocytopenia, including the influence of medication during ECMO, as well as evaluating the indications and thresholds for platelet transfusion in bleeding and non-bleeding patients on VA ECMO.