Liberal Vs. Restrictive Transfusion Thresholds in Leukemia Patients: A Feasibility Pilot Study

Transfusion of red blood cells (RBCs) is vitally important for the care of patients (pts) undergoing myelosuppressive therapy for acute leukemia (AL). At diagnosis and before bone marrow recovery, RBCs are given liberally with hemoglobin (Hb) transfusion triggers of 8-9 g/dL or higher, often with two RBC units given at a time. Research in a variety of non-oncologic settings suggests that lower Hb triggers (7-8 g/dL) do not increase mortality whereas higher Hb triggers (9 g/dL) may increase mortality. For pts with hematologic malignancies, the ideal Hb threshold is unknown. Also, there is literature to suggest that higher Hb values force platelets to the periphery of capillaries and reduce bleeding events. A study of red cell triggers in pts with thrombocytopenia (as in AL) is important as we may not be able to rely on trigger data from others in this setting. This feasibility study was designed to determine whether a randomized trial of Hb triggers could be performed and what challenges may be encountered with this select population.

In this prospective randomized study, we enrolled adult pts with AL (myeloid and lymphoid) undergoing intensive induction therapy to evaluate both patient and physician tolerance for a transfusion threshold of 7g/dL (LOW) compared to the standard threshold of 8g/dL (HIGH), as well as operationalize transfusion to a single unit per episode. Pts were randomized with a 2:1 ratio of LOW threshold to HIGH threshold. Pts were ineligible if there was clinical concern for acute coronary syndrome or active blood loss. Pts were monitored from the study enrollment date through the next marrow evaluation after completion of induction therapy. Vital status at Day 60, accrual rate, and safety endpoints (mortality, length of stay, infection, fatigue scores, renal, respiratory, thrombotic, bleeding events) were monitored. Total transfusions, length of inpatient stay, and percentage of pts who crossed over to HIGH threshold for symptomatic reasons were also monitored.

Between April 15, 2014 and July 23, 2015, 162 acute leukemia pts would have been eligible for this trial. Of the eligible pts, 112 (69%) were approached and 90 (of planned 90) have enrolled to date. Twenty-two pts (19.6%) of those offered the trial declined. Seventy-four pts are currently evaluable for response (13 pts have not been on-study long enough to evaluate and 3 pts are not eligible for data analysis): 51 pts in LOW arm and 23 in HIGH arm. Interim data is shown in Table 1. In the LOW arm, three pts withdrew consent due to self-reported fatigue and one patient was removed by physician due to clinical scenario. Seventeen pts (33%) in the LOW arm and six pts (26%) in the HIGH arm were transfused outside their hemoglobin trigger without meeting criteria to discontinue the study. No deaths were attributable to Hb thresholds in this study. Significant bleeding events (Grade 3 or 4 by CTCAE 4.0) were comparable with 4 in LOW arm and 3 in HIGH arm. More minor bleeding events (Grade 1 or 2 by CTCAE 4.0) were higher in LOW (24 events) than in HIGH arm (15 events).

Acute leukemia and its therapies carry a significant mortality risk and it is not clear how transfusion thresholds and number of PRBC units transfused impact this mortality. Furthermore, blood is an expensive and scarce resource. Without a clear benefit of higher Hb thresholds, the added risks and costs of transfusion may not be justified. This pilot study shows that both pts and leukemia physicians will tolerate randomization between Hb thresholds and there is not a signal for harm in either Hb threshold at present. Pts in the LOW arm receive fewer transfusions and do not experience higher fatigue scores nor increased significant bleeding events. Length of hospitalization was similar in both arms. This safety data will serve as a platform for a larger mortality study in leukemia and possibly additional studies in other oncologic diseases.