Platelet transfusions in cancer patients with hypoproliferative thrombocytopenia in the intensive care unit

Thrombocytopenia is a frequent finding in critically ill patients, either present at the time of intensive care unit (ICU) admission or further acquired during the ICU stay [1]. Thrombocytopenia carries major implications in terms of diagnostic investigations, prognosis, and specific treatment with particular emphasis on platelet transfusions. The mechanisms leading to thrombocytopenia in the critically ill are mostly peripheral through consumption or destruction [2]. The poor prognostic value of thrombocytopenia has been largely emphasized in the ICU. Most interestingly, it has been shown that relative thrombocytopenia, as defined by a 30 % drop in platelet count regardless of baseline and final values, shares similar pathophysiological and prognostic significance with absolute thrombocytopenia [3, 4]. This suggests that the process leading to platelet consumption is more important than the platelet count by itself. With respect to the risk of bleeding, it is striking that the risk of bleeding in ICU patients is poorly related to the depth of thrombocytopenia [2, 4, 5]. However, such studies were performed in cohorts of critically ill patients in whom peripheral mechanisms leading to thrombocytopenia were largely predominant [2, 5]. In contrast, a number of ICU critically ill patients primarily exhibit decreased platelet production as a consequence of underlying malignant diseases and their myelosuppressive treatments. The relation between platelet count and bleeding in this setting has not been addressed.

The only goal of platelet transfusions is to prevent or to treat bleeding. To this day, indications of platelet transfusions in the ICU remain highly speculative. Recently, the French Intensive Care Society provided some expert recommendations guiding platelet transfusions in the ICU [6]. The platelet count threshold for securing invasive procedures or treating severe bleeding was set at 50 × 10⁹/L. As for prophylactic platelet transfusions, the guidelines did not retain any indication in patients with peripheral thrombocytopenia and proposed a platelet count threshold of 20 × 10⁹/L for hypoproliferative thrombocytopenia. Of note, such a restrictive transfusion policy was only based on studies performed in hematologic patients in stable conditions and may not apply to critically ill patients. We herein aimed to address the current practices in platelet transfusion and the risk of bleeding in cancer patients with hypoproliferative thrombocytopenia in a medical intensive care unit.

A number of studies have underscored the high prevalence and the poor prognostic value of thrombocytopenia in the ICU [1–3, 5, 6, 11–15]. Beyond absolute thrombocytopenia, decline in platelet count over the first days in the ICU, the so-called relative thrombocytopenia, also represents a reliable prognostic factor in addition to single measurements [2, 3, 11]. This suggests that the prognostic value of thrombocytopenia might not only be fully related to its depth, but also to the underlying pathophysiological process leading to platelet consumption or destruction [16]. It is noteworthy that all studies about thrombocytopenia in the ICU and the eventual indication of platelet transfusions have been performed in general cohorts of patients for whom decrease in platelet counts was mainly related to peripheral mechanisms [2–5, 17]. To the best of our knowledge, this is the first study that specifically addresses the transfusion management in cancer patients with hypoproliferative thrombocytopenia in the ICU.

The risk of bleeding events in the setting of thrombocytopenia underlies the use of prophylactic platelet transfusions. The current guidelines about the management of platelet transfusions in hypoproliferative thrombocytopenia are mainly derived from studies in patients with hematological malignancies requiring chemotherapy or undergoing allogeneic or autologous hematopoietic stem cell transplantation [18, 19]. Studies in hematology have been able to correlate the nadir platelet count with bleeding, the risk of severe hemorrhage clearly becoming much higher when platelet count drops below 10 × 10⁹/L [20–22]. In addition, various associated conditions are likely to increase the risk not only of bleeding in oncologic and hematologic thrombocytopenic patients, including the duration of thrombocytopenia, but also altered functional status, bone marrow metastasis, allogeneic bone marrow transplantation, a recent history of bleeding, hypoalbuminemia, and treatment with drugs affecting platelet function [21, 23]. Over the last two decades, several studies have addressed the indications of prophylactic platelet transfusion in clinically stable patients without active bleeding. Compared to the once traditional trigger of 20 × 10⁹/L, these studies have proven the safety and the cost-effectiveness of a more restrictive trigger of 10 × 10⁹/L in patients with prolonged thrombocytopenia related to acute myeloid leukemia [24–27]. An exclusive therapeutic policy was recently proposed to restrict platelet transfusion to documented bleeding [22, 28]. However, such a restrictive transfusion policy was associated with an increased risk of moderate-to-severe bleeding in patients with prolonged and profound thrombocytopenia as seen in acute leukemia.

The French Intensive Care Society recently provided expert recommendations in the management of platelet transfusions in the ICU [6]. Prophylactic platelet transfusion was recommended for central thrombocytopenia when platelet count fell below 20 × 10⁹/L. Accordingly, the large majority (80 %) of prophylactic platelet transfusions in the present study were performed at platelet counts thresholds lower than 20 × 10⁹/L. However, this recommendation was derived from studies performed in hematologic patients in stable conditions, and whether it can be safely implemented in critically ill cancer patients had not been addressed previously. Indeed, the quantitative relation between platelet count and hemorrhage remains unclear in critically ill patients in contrast to thrombocytopenic patients in stable conditions. Observational studies retrieved an increased risk of bleeding for severe thrombocytopenia (platelet count <50 × 10⁹/L), whereas the risk associated with moderate (50–100 × 10⁹/L) and mild (100–150 × 10⁹/L) thrombocytopenia was inconsistent [2, 4, 5, 13, 29]. Whether thrombocytopenia was directly involved in bleeding through the nadir platelet count or indirectly as a marker of severity remains unclear. Altogether, these results and ours suggest that ICU-acquired severe bleeding is a multifactorial process that is not only dependent on a sole platelet count. Most importantly, some additional factors such as uremia, other coagulation disorders, or anticoagulant treatments are frequently associated with the development of bleeding in the ICU, and should certainly be taken into account in the decision-making process of platelet transfusion in thrombocytopenic patients [12, 21].

A major obstacle to broad platelet transfusion lies in the limited availability of platelet concentrates. In addition, platelet transfusion may not be innocuous in the critically ill as supply of exogenous platelets may sustain or even exacerbate some pathophysiological thrombotic processes [15, 30–32]. Therefore, prophylactic platelet transfusion is usually not recommended in patients with exclusive peripheral thrombocytopenia [6]. As mentioned above, it is likely that both central and peripheral mechanisms are involved in thrombocytopenia of cancer patients, who commonly exhibit risk factors for platelet transfusion refractoriness such as spleen enlargement, bleeding, fever, infection, disseminated intravascular coagulation as well as the number of previous transfusions [33]. The low platelet recovery we observed in all three indications of transfusions probably resulted from fast consumption or destruction of exogenous platelets.

This study has several limitations. It was performed in a single tertiary care center with a particular case-mix of patients, and the period of inclusion was 7 years. However, our transfusion practices, based on previous studies in hematology and confirmed by the release of expert recommendations for critically ill patients, remained similar over the study period. Despite the fact that we primarily selected patients on the occurrence of platelet transfusions, this represents our common attitude in profound hypoproliferative thrombocytopenia and this probably allowed an exhaustive identification of patients. The major limit of the study obviously lies in its retrospective design, the results being highly dependent on the quality of records in the medical charts. However, we believe that automated extraction and quantitative variables from computerized charts as well as manual inquiry for other qualitative variable allowed us to perform a reliable collection of data. In contrast, duration, depth, and management of thrombocytopenia prior to ICU admission could not be accurately collected. The retrospective design may also have prevented a completely reproductive assessment of platelet recovery following transfusion. Indeed, we chose to collect the last platelet count prior to transfusion, and the next morning’s resulting platelet count. So the time interval between both platelet counts, as well as the timing of the transfusion in between, is likely to impact the difference. In the same way, this questions the exhaustiveness of collection of bleeding events. Undoubtedly, grade 3 and 4 bleeding events were significant enough to be reported in the charts, but grade 1 and 2 events might have been inconsistently mentioned. Finally, although our results suggest that the depth of thrombocytopenia was not an accurate predictor of severe bleeding, the retrospective design of the study and the quite low number of patients who bled precluded developing a more comprehensive and reliable multivariate predictive model of bleeding.

In accordance with recent expert recommendations, most prophylactic platelet transfusions were given using restrictive thresholds of 10–20 × 10⁹/L in critically ill cancer patients with hypoproliferative thrombocytopenia. As a matter of fact, the risk of ICU-acquired severe bleeding appears hardly predictable with the depth of thrombocytopenia. Further prospective interventional studies are needed in order to reliably identify safe thresholds for platelet transfusion in critically patients with hypoproliferative thrombocytopenia in relation with the complex prediction of severe bleeding.