A Randomized Controlled Trial on the Efficacy of High-Dose Granulocyte Transfusion Therapy in Neutropenic Patients with Infection

Bacterial and fungal infections continue to be a major problem in patients with prolonged severe neutropenia. Early controlled trials suggested that granulocyte transfusions were modestly effective in this setting, but the doses provided were later considered inadequate. Recent studies have shown that the dose can be increased substantially by administering G-CSF ± dexamethasone to granulocyte donors. Although these cells circulate in neutropenic recipients and appear to function normally, the evidence for clinical efficacy has been inconclusive.

We report here the result of the RING study, a recently completed randomized controlled trial on the efficacy of high-dose granulocyte transfusion therapy, carried out as part of the NHLBI Transfusion Medicine/Hemostasis Clinical Trials Network. Fourteen clinical sites participated. Eligible subjects were those with neutropenia (ANC<500) and proven/probable/presumed bacterial or fungal infection. Subjects were randomized to receive either 1) standard antimicrobial therapy or 2) standard antimicrobial therapy plus daily granulocyte transfusions from normal donors stimulated with G-CSF (450µg) and dexamethasone (8mg). The primary end point was a composite one; survival plus a microbial response, both evaluated 42 days after randomization. Microbial response was determined by a blinded adjudication panel.

The target sample size was 236 subjects, designed to provide 80% power to detect a 20% difference in success rates between treatment and control groups; however, only 114 subjects could be enrolled. Patient infections were 36% invasive fungal, 27% invasive bacterial, 11% fungemia, and 26% bacteremia. Subjects in both arms were well matched in terms of demographics, underlying disease, types and sites of infection, and severity of illness. Fifty six subjects were randomized to the granulocyte arm; 51 received at least one transfusion; the mean time from eligibility to the first transfusion was 2.3 +/- 1.2 days. Among these 51 subjects, the median number of transfusions was 5 (quartiles 3 and 9), given over a median of 6 days (quartiles 4 and 11). The median number of granulocytes administered per transfusion was 54.9 x10⁹ (quartiles 26.1 x10⁹, 72.5 x10⁹). Fourteen percent of these patients had > Grade 3 hypoxemia develop during or within six hours after a granulocyte transfusion, requiring ventilation in one patient (2%). No deaths were attributed to adverse effects associated with the transfusions.

Among subjects with sufficient data to determine the primary outcome, success rates were 42% (20/48) and 43% (21/49) for the granulocyte and control groups, respectively (p> 0.99) on Intention to Treat (ITT) analysis, and 49% (17/35) and 41% (16/39), respectively, for subjects who adhered to their assigned treatments (Per Protocol (PP) analysis)(p=0.64). There was also no significant difference between treatment groups in a model of the primary outcome that adjusted for baseline prognostic factors (e.g. ventilator use, high Zubrod score). Differences in primary end point success rates for granulocyte and control arms did not differ significantly for any infection type whether analyzed by ITT or PP. Outcomes for patients who received the first transfusion within 2 days of eligibility were similar to outcomes for patients who received the first transfusion later. For patients who received at least three granulocyte transfusions, those who received an average dose per transfusion of >50x10⁹ granulocytes had a higher success rate (57.7%)(n=26) than those receiving <5x10⁹ cells per transfusion (11.1%)(n=9)(p = 0.04); while supporting the hypothesis that dose is critical to the outcome, this result needs be interpreted with caution because of the low numbers and poor outcome of the low dose group. There was no significant difference between the granulocyte and control arms on overall survival to either 42 or 90 days after randomization.

Because of incomplete patient enrollment, the power of this study to detect a 20% difference in overall success rates was reduced to approximately 40%. Thus it is possible that a true convincing favorable effect was missed, particularly, as suggested, in the subset of patients who received daily transfusions containing at least 50x10⁹ granulocytes per transfusion.