Granulocyte transfusions are inconsistently used despite observational evidence of their efficacy (Seidel et al, 2008). The Resolving Infection in Neutropenia with Granulocytes (RING) trial was incomplete due to poor enrollment and thus was underpowered to assess efficacy of granulocytes in adults with neutropenia and severe infection unresponsive to anti-microbials (Price et al, 2015). Post-hoc analysis showed a group of patients had better clinical responses with higher doses of granulocytes. We hypothesized that patients treated at Stanford University with higher doses of granulocytes had better clinical outcomes than those patients who received lower doses of granulocytes.
Granulocyte transfusions are available at Stanford University from well repeat platelet donors who have recent negative infectious disease markers for transfusion-transmitted infections. Donors receive stimulation with steroids about 12 hours prior to collection.
After IRB approval, a single center retrospective cohort study was done to assess all patients who had received granulocytes at Stanford University from August 2006 to June 2018. We queried the electronic health record and transfusion service information system for patient demographics, diagnosis, weight, dose and frequency of granulocyte infusion(s), time to ANC recovery >500 and the primary outcome of survival to hospital discharge. There were 45 patients identified within the transfusion service information system, but only 27 had available granulocyte dosing and clinical records for review.
To evaluate for most important clinical parameters, random forest classification and regression was performed against survival to hospital discharge and time to neutropenia recovery post initiation of granulocyte infusions respectively (randomForest R package). Cox proportional hazard (coxph) models were determined using mean granulocyte dose per infusion (cells/kg/infusion), total granulocyte dose per admission (cells/kg), age, and duration of neutropenia against survival to discharge and resolution of neutropenia (survival and survminer R packages). Kaplan-Meier survival analysis was performed based on high and low mean granulocyte dosing (<0.6e9cells/kg, which is equivalent to 4 x10e10 in a 70kg patient).
Twenty-seven patients age 3 to 80 years (median 38 years) with various hematologic disorders were treated with granulocyte transfusions (median 4 infusions, range 1 - 14). Fifteen of 27 (56%) patients survived to hospital discharge. Random forest classification identified mean granulocyte dose per infusion as the most influential feature in predicting survival to hospital discharge by both mean decrease in accuracy and mean decrease in Gini. Additional influential parameters included total granulocyte dose per kg, age, and duration of neutropenia. Random forest regression could not identify a clear feature that was most influential in predicting time to resolution of neutropenia. Coxph models did not identify a significant feature that predicted survival to discharge. Although all 3 patients who received high dose infusions survived compared to 50% surviving in the low dose group (median survival of 33 days), the results did not achieve significance (p = 0.15). Coxph models identified mean granulocyte dose per kg as significantly associated with resolution of neutropenia (HR 26.15, 95% CI 1.65-413.83, p = 0.021) whereas total granulocyte dose per kg over all infusions was associated with prolonged neutropenia (HR 0.69, 95% CI 0.52-0.92, p = 0.012).
We present a series of granulocyte transfusions in 27 neutropenic patients with life-threatening infections not responsive to standard therapies. Though there was not a statistically significant difference in survival between those infused with high versus low dose granulocytes, the mean granulocyte dose infused per kg was associated with a resolution of neutropenia. This suggests that if one prescribes granulocytes for these patients, they should be higher dose (at least 0.6e9cells/kg).
1. Seidel MG, Peters C, Wacker A, et al. Study of granulocyte transfusions in neutropenic patients. Bone Marrow Transplant 2008;42:679-86.
2. Price TH, Boeckh M, Harrison RW, et al. Efficacy of transfusion with granulocytes from G-CSF/dexamethasone-treated donors in neutropenic patients with infection. Blood 2015;126:2153-61.
Disclosures
No relevant conflicts of interest to declare.
Matlab Based Potent Algorithm for WBc cancer Detection and classification
