PATCHing platelet data to improve transfusion

In the current issue of Blood, Baharoglu et al present an important secondary analysis of the Platelet Transfusion in Cerebral Hemorrhage (PATCH) trial to evaluate the risks of platelet transfusion among patients with platelet inhibitor–associated hemorrage.¹  Although platelet transfusion is recognized as an effective prophylactic and therapeutic treatment of bleeding, the sparse evidence comparing restrictive to liberal platelet transfusion strategies results in wide variations in transfusion practice.²  One area of significant controversy is the transfusion of platelets to patients with intracranial hemorrhage while taking antiplatelet therapy. The AABB platelet guidelines,³  published in 2015 before the primary PATCH trial⁴  results were reported, highlight the lack of data in this area and do not make a recommendation for transfusion in this clinical scenario.

Despite this apparent equipoise, the primary analysis from the PATCH trial,⁴  which identified a twofold increase in 3-month mortality among patients who were transfused with platelets, was highly controversial, in part due to the intrinsic desire for physicians to provide aggressive, rational treatments to patients presenting with intracranial hemorrhage.⁵  Had PATCH reported that platelet transfusions had no impact on patient outcomes, the finding may have been easier for many stakeholders to accept. The fact that the trial identified harm to patients who were transfused left many experienced physicians uncertain of the findings, and wondering how the provision of platelets to a bleeding patient could worsen outcomes. In an accompanying editorial,⁵  PATCH was critiqued for a relatively modest enrollment (n = 190), imbalance between the study arms, and stringent inclusion criteria that, as with most clinical trials, made it difficult to judge applicability of the findings to individual, real-world patients. Clinical practice guidelines published by the British Society for Haematology and the American Society of Clinical Oncology after the release of the PATCH data do not make specific, strong recommendations against transfusion in this setting, suggesting that the trial did not significantly change practice.^6,7  Nonetheless, the primary analysis was clearly important, as it was the first (and only, to date) randomized clinical trial to address this question. In addition, the harm signal that was identified establishes the need to critically examine the rationale for platelet transfusion in this setting.

The PATCH trial demonstrated that individuals who received platelet transfusions had greater intracranial hemorrhage volume and perihematomal edema. The secondary analysis in the current issue of Blood assesses whether the increased risk associated with platelet transfusion could be due to potentially proinflammatory effects of platelets or differences in patient characteristics between the trial arms. The investigators found no differences in outcomes observed that were associated with platelet type (apheresis platelets vs buffy coat platelets) or duration of platelet storage. However, the authors show the negative outcomes associated with platelet transfusion could potentially be attributable to a lack of balance between the transfusion arm and the control arm. Patients assigned to the invention group had significantly larger intracerebral hemorrhage, perihematomal edema, and total lesion volume than those in the control group at baseline. In addition, more patients in the transfusion arm were treated with a drug regimen that results in greater platelet inhibition (dual therapy or adenosine 5′-diphosphate inhibitor therapy). These baseline differences may have increased the risk of hemorrhage expansion, irrespective of transfusion strategy, and provide a credible alternative hypothesis for the observed differences in outcomes between the study groups.

Ultimately, determination of the optimal transfusion strategy for patients who develop intracranial hemorrhage while taking antiplatelet drugs will require additional observations on the outcomes of patients who are treated with (or without) platelets, in a wide range of clinical scenarios. Perhaps one lesson from the PATCH trial is that even well-designed and expertly executed clinical trials may only take us part of the way. The best approach to answering complex clinical questions may be identified by analyzing a patchwork of clinical trial data combined with real-world observations via evaluation of large clinical datasets. At present, large quantities of important clinical data are not consistently captured by electronic medical record systems.⁸  In the future, we hope that the use of well-designed common data models will allow researchers to capture and organize a large amount of clinical data, including transfusion data, so that real-world observations can complement clinical trial data in the search for answers to difficult questions.