Incidence, kinetics, and clinical impact of thrombocytopenia in venovenous ECMO: insights from the multicenter observational PROTECMO study

Abstract

Background: Thrombocytopenia is a recognized risk factor for bleeding during extracorporeal membrane oxygenation (ECMO). This study determines the incidence, risk factors, and clinical relevance of thrombocytopenia and platelet transfusions during venovenous (VV) ECMO.

Methods: The multicenter, prospective observational PROTECMO study included 652 adult patients who received VV ECMO for respiratory failure. Thrombocytopenia was classified as mild (100-149·10⁹/L), moderate (50-99·10⁹/L), or severe (< 50·10⁹/L). Bleeding events were evaluated using a modified Bleeding Academy Research Consortium score. Cox proportional hazards and logistic regression analyses were done to identify predictors, and quantify the association between platelet counts and bleeding risk.

Results: A total of 182 patients (27.9%) had thrombocytopenia at baseline (mild in 14.7%, moderate in 8.7%, and severe in 4.4%). Thrombocytopenia during ECMO, at least once in 80.2% of patients, was mild in 21.3% of cases, moderate in 32.2%, and severe in 26.7%. A 10·10⁹/L decrease in platelet count was associated with a 3.7% (95% CI: 2.4-5.0%) increase in risk of bleeding. There was no strong evidence of nonlinear relationship within the platelet count range between 25,000 and 300,000. This relation remained consistent across all ECMO weeks. Mild thrombocytopenia increased the risk of experiencing a bleeding event by 61% (hazard ratio (HR) 1.611, 95% CI 1.230-2.109, p = 0.0005), while moderate and severe thrombocytopenia increased the risk by roughly 90% (moderate: HR 1.944 (CI 1.484-2.545), p < 0.0001; severe: HR 1.876 (CI 1.275-2.7680), p = 0.0014). The risk for thrombocytopenia < 100·10⁹/L during ECMO significantly increased with ICU days prior to ECMO start, postoperative admission, immunocompromised state, renal replacement therapy, septic shock, low hemoglobin, and circuit exchange.

Conclusions: Thrombocytopenia is highly prevalent in VV ECMO, and associated with a significant increase in the risk of bleeding, and a reduction in 6-month survival, particularly at platelet counts below 100·10⁹/L. Further research is needed to better define the outcomes associated with specific thresholds for transfusion of platelets.

Keywords: Anticoagulation; Bleeding; Intensive care; Platelet kinetics; Predictors; Thrombocytopenia; VV ECMO.

Fig. 1

a. Progression of Thrombocytopenia Classification Over Time. This stacked bar chart displays the progression of thrombocytopenia severity during 28 days of follow-up. The y-axis represents the number of patients stratified by thrombocytopenia severity; the x-axis indicates the follow-up period in days. b. Platelet Trends and Bleeding Incidence over 28 Days of Follow-Up. The bar chart represents the mean platelet count (y-axis on the left) over the follow-up period (x-axis in days), stratified by bleeding status. The black line with markers shows the rate of bleeding patients as a percentage (y-axis on the right) at each follow-up day. Numerical values displayed on the bars correspond to the mean platelet counts for the respective groups. c. Association between Platelet Count and Hazard Ratio for Bleeding. The relationship was assessed using restricted cubic splines in a Cox proportional-hazard model, with platelet count as a continuous variable. The solid line represents the estimated hazard ratio, and the shaded area indicates the 95% confidence interval. d. Platelet Count Trajectories by Bleeding Status over 28 Days. This line graph illustrates the trajectory of platelet count (y-axis) over a 28-day follow-up period (x-axis) in patients grouped by bleeding status. The blue line represents patients who did not experience any bleeding event during the follow-up period, while the green dashed line represents patients who experienced at least one bleeding event. Data points and error bars for each day represent mean platelet count with its 95% confidence interval