Effect of leukocyte compatibility on neutrophil increment after transfusion of granulocyte colony-stimulating factor–mobilized prophylactic granulocyte transfusions and on clinical outcomes after stem cell transplantation

Granulocyte transfusions improved the survival rates of neutropenic dogs with Pseudomonas sepsis.1 In preclinical models, granulocyte component cell dose was a critical determinant of neutrophil increments and of survival.1Leukocyte incompatibility also affected outcomes, resulting in poorer absolute neutrophil count (ANC) increments and reduced granulocyte migration and function.2-4 In a meta-analysis of randomized human trials, both component cell dose and leukocyte compatibility were important determinants of clinical efficacy of granulocyte transfusions.5 In human trials of granulocyte components collected without granulocyte colony-stimulating factor (G-CSF) mobilization, leukocyte incompatibility resulted in poorer ANC increments and reduced granulocyte migration and function.6-8 Unfortunately, the development of alloimmunization after granulocyte transfusions was common.9 10 

The benefit of G-CSF to mobilize granulocytes includes increased component cell dose by a factor of 2- to 5-fold over corticosteroids.11 In addition, G-CSF inhibits apoptosis and prolongs survival of neutrophils in vitro12 and, thus, may further sustain neutrophil increments after transfusion. We and others11,13,14 demonstrated that the transfusion of G-CSF–mobilized, human leukocyte antigen (HLA)-matched granulocytes resulted in significant and sustained ANC increments. Mean peak ANC increments between 631/μL and 1195/μL were observed after granulocyte transfusions administered to recipients of non-alloimmunized neutropenic bone marrow transplants.13The duration of post-transfusion recipient mean ANC at or above the baseline (before transfusion) value was 25 to 37 hours.13Such granulocytes were also functional, as demonstrated by the localization of indium-labeled granulocytes to sites of inflammation for at least 48 hours after transfusion.15 

Granulocyte component cell dose improved with the use of G-CSF as the mobilization agent; however, the effect of leukocyte compatibility on outcomes with G-CSF–mobilized granulocyte transfusions is unclear. The objectives of this trial were to determine the effect of leukocyte compatibility on ANC increments and clinical outcomes after the transfusion of prophylactic G-CSF–mobilized granulocyte components into neutropenic recipients of autologous PBSC transplants. For each recipient, 4 granulocyte components were to be collected from a single donor. Granulocyte components were transfused into recipients on transplant days 2, 4, 6, and 8. Leukocyte compatibility between donor and recipient was determined by lymphocytotoxicity assays, HLA-A and HLA-B typing, and leukoagglutination cross-match. The results of this trial are reported herein.

The protocol was approved by the Institutional Review Board of Washington University School of Medicine and accrued 25 donor–recipient pairs between March 1997 and July 1998. Informed consent was required for entry of recipients and donors into the study. Eligibility criteria for donors included age greater than 15 years, ANC 1500/μL or greater, platelet count 100 000/μL or greater, hematocrit 30% or greater, and, if female, negative pregnancy test result. Every donor was ABO-compatible with, and a first-degree relative (biologic sibling, parent, or adult offspring) of, a recipient. Otherwise, donors satisfied standard blood donation criteria as determined by the U.S. Food and Drug Administration and as interpreted by the Blood Bank director. Recipients between the ages of 15 and 70 years who were undergoing autologous PBSC transplantation for malignant disease were eligible. Stored autologous PBSC products for each recipient had to contain a minimum of 2 × 10⁶CD34⁺ cells/kg actual body weight for subsequent transfusion. The PBSC products of 1 recipient contained 1.4 × 10⁶ CD34⁺ cells/kg, but the recipient was permitted to participate in this protocol. Recipients had to be free of active infection at study entry.

Within 7 days of initiation of the transplant-conditioning regimen, donors and recipients underwent baseline history taking and physical examination. Blood was obtained for complete blood count with differential, ABO and Rh typing, and infectious serology testing, as previously reported.13 Type and cross of donor and recipient blood were performed to confirm ABO compatibility.

At study entry only and before the conditioning regimen was initiated, leukocyte compatibility between donor and recipient was determined by 4 methods. HLA-A and HLA-B typing of donor and recipient were performed using a standard serologic typing technique,16 and a match grade was assigned from 0 (0 of 4 HLAs matched) to 4 (4 of 4 HLAs matched). A screening lymphocytotoxicity assay (s-LCA) was performed on each recipient. Serum samples from each of the recipients were tested for class I and II HLA antibodies against T- and B-cell targets by standard microlymphocytotoxicity testing with a panel of 24 to 30 cells.17 Recipient serum was added to a standard panel of screening cells and incubated to allow antigen–antibody interaction. Rabbit serum was added to the cell–serum suspension as a source of complement. Cell membrane injury that resulted from an interaction between antibody-bound cells and complement was visualized by the cells' inability to exclude a vital dye (ethidium bromide; Mallinckrodt, Paris, KY), and the percentage of cell death was scored from 1 (cell viability same as control) to 8 (essentially all cells killed). Scores of 6 or greater indicated the presence of an HLA antibody. Panel reactivity was calculated as percent reactive antibody: number of positive panel cells divided by total number of panel cells times 100. Positive s-LCA was defined as percent reactive antibody greater than or equal to 10%. The sensitivity of the technique was enhanced by using the Amos 3-wash method and by the addition of anti-human globulin before the addition of complement.18Appropriate positive and negative controls were concurrently performed. Using the same methods, a lymphocytotoxicity cross-match assay (c-LCA) between recipient serum and donor cells was performed and scored as positive or negative. A leukoagglutination cross-match was performed by standard methods using recipient serum and granulocyte-enriched cell suspension obtained from defibrinated blood of the donor.19Appropriate positive and negative controls were concurrently performed. Reactions were scored based on clumping according to the following scale: 0 or negative (no visible microscopic agglutination), 1+ (10% to 20% agglutination), 2+ (30% to 40% agglutination), 3+ (50% to 80% agglutination), and 4+ (90% agglutination).

To mobilize granulocytes, donors were given G-CSF at 10 μg/kg subcutaneously 8 to 12 hours before each of 4 scheduled granulocyte apheresis collections. Leukapheresis of the donor began the evening (5PM) of post-transplant day 1 and was repeated the mornings (9 AM) of days 4, 6, and 8. Venous access for each apheresis procedure was by peripheral vein-to-peripheral vein technique. Granulocyte collections were performed with an automated continuous-flow blood cell separator by means of Standard Procedure 2 (CS-3000 Plus; Baxter, Deerfield, IL) modified by the use of an interface offset setting of 35.20 For each collection, 7 L was processed by continuous-flow centrifugation at a rate of approximately 50 mL per minute. Clotting was prevented and red cell sedimentation was facilitated by the use of sodium citrate (30 mL of 46.7% trisodium citrate) in 500 mL of 6% hydroxyethylstarch (McGraw Laboratories, Glendale, CA) to the donor line at a ratio of 1 part anticoagulant to 10 to12 parts blood.

To evaluate the cellular composition of each leukapheresis component, the white blood cell (WBC) count and differential were determined on a 3-mL sample, by using a counter (Coulter, Hialeah, FL). The total number of granulocytes in each leukapheresis component was calculated on the basis of these data and the total component volume. Each component was irradiated (2500 cGy, 137 Cs source) before transfusion to the recipient, to prevent transfusion-associated graft-versus-host disease.

Donors were evaluated daily for any side effects caused by either the G-CSF or the leukapheresis procedures. Donor monitoring, toxicity grading, and criteria for removal of donors from the study were previously reported.13 On the day of each scheduled leukapheresis, the donor's hematocrit and platelet count had to be 30% or greater and 75 000/μL or greater, respectively, for the donor to be eligible for the procedure.

Recipients received transplant conditioning regimens consisting of high-dose multi-agent chemotherapy only (n = 15 patients) or with total body irradiation (n = 10 patients) given over 4 to 6 days, followed by infusion of thawed autologous PBSC products on transplant day 0. Beginning 4 hours after PBSC transplantation, each recipient received G-CSF at 5 μg/kg subcutaneously and given daily until the ANC recovered to greater than or equal to 1500/μL for 3 consecutive days.

Within 10 hours of completion of the first leukapheresis procedure and within 2 hours of completion of subsequent leukapheresis procedures, the apheresis components were transfused into the recipient over the course of 1 hour. An extended interval between completion of leukapheresis and component transfusion was required for the first procedure to permit the completion of infectious serology tests on the component to determine whether it was acceptable for transfusion according to American Association of Blood Banks' (AABB) guidelines.21 Apheresis components collected on subsequent days were transfused promptly if results of infectious serology tests on the immediately prior component were acceptable, as recommended by AABB guidelines.21 Based on this schedule, granulocyte components were transfused into recipients on day 2 in the early morning and on days 4, 6, and 8 in the early afternoon. The recipient's ANC and platelet count were determined immediately before each component transfusion (the baseline) and then at 1, 4, 8, 12, 24, 36, and 48 hours after each component transfusion.

Routine medications were given to recipients 15 to 30 minutes before each granulocyte component transfusion to prevent transfusion reactions (fever and hives); they included acetaminophen (650 mg by mouth) and diphenhydramine (25 mg by mouth or intravenously [IV]). During and after each granulocyte transfusion, recipients were observed and vital signs were monitored for any evidence of acute toxicity resulting from the transfusion. If fever (temperature 38.3°C or higher) or respiratory symptoms (cough, shortness of breath, or chest tightness) developed during or within 1 hour of a transfusion, hydrocortisone (100 mg IV) was also given to that recipient before subsequent transfusions. If fever or respiratory symptoms developed with 2 transfusions and after the addition of hydrocortisone, no further transfusions were given. Respiratory distress requiring supplemental oxygen (caused by a decline in O₂ saturation to less than 90%) or hypotension (systolic blood pressure less than 90 mm Hg) during or within 1 hour of a transfusion was also an indication not to administer subsequent scheduled components.

Prophylactic antibacterial and antifungal agents were not given to recipients. Infection was presumed in recipients with fever (temperature 38.3°C or higher) when the ANC was less than 1000/μL, and it was managed initially with empiric vancomycin (1 g IV bid) and cefepime (1 g IV tid) after evaluation (by examination, blood cultures, and other indicated tests), followed by amphotericin B (0.5 mg/kg per day) if febrile neutropenia persisted or recurred after 3 or more days of antibacterial therapy. Amphotericin B was administered 8 to 12 hours before or after the transfusion of granulocyte components to reduce the likelihood of adverse pulmonary reactions reported when the 2 were given concurrently.22 Documented bacteremia, tissue (eg, pneumonia), or catheter-related infections were treated with a 14-day course of antimicrobials (minimum, 7 days, IV). Recipients received prophylactic irradiated, WBC-filtered, single-donor platelet concentrate transfusions if the platelet count was less than 10 000/μL and 2 U WBC-filtered packed red cells if the hemoglobin level was less than 8.0 g/dL. Blood counts with differentials and platelet counts were performed daily during the hospital stay and BIW afterward until the platelet count was greater than 50 000/μL after transplantation.

The primary objective of this phase 2 study of 25 donor–recipient pairs was to determine the effect, if any, of leukocyte compatibility on peak ANC increments (over baseline) after the transfusion of prophylactic G-CSF–mobilized granulocyte components into neutropenic recipients of autologous PBSC transplants. Because apheresed granulocyte components contain large numbers of platelets,13 the absolute and corrected platelet count increments (CCI) were also determined 1 hour after the granulocyte transfusions, and the effect of leukocyte compatibility on these outcomes were assessed. CCI was calculated according to the following formula23: CCI = [absolute increment × body surface area (m²)] / [number of transfused platelets × 10¹¹].

Secondary endpoints of the study were to determine the effect of leukocyte compatibility on selected clinical outcomes in PBSC recipients including the number of febrile days, platelet transfusion requirements, hematologic recovery, and adverse granulocyte transfusion reactions (including fever, respiratory distress, and hypotension).

The kinetics of the recipient ANC and platelet count with transfusions of G-CSF–mobilized granulocyte components on transplant days 2, 4, 6, and 8 were assessed by descriptive statistics (mean, range), as were clinical outcomes. Comparisons between groups of data were performed using either an unpaired t test or a Mann-Whitney Utest, as appropriate, with P = .05 defined as significant. Statistical analysis was performed using a software program (Statview SE + Graphics; Abacus Concepts, Berkeley, CA).

Twenty-five donor–recipient pairs were accrued to this study, of which 2 were nonevaluable because of poor venous access and the inability to perform apheresis in 1 donor and a side effect of G-CSF in 1 donor. An anaphylactoid reaction developed in the latter donor within 1 hour of the first dose of G-CSF and has been reported in detail elsewhere.24 The donor recovered without complications and did not undergo apheresis or receive additional G-CSF injections. The relationship of donor to recipient was sibling in 12 pairs, adult offspring in 11, and parent in 2. Donors' mean age and weight were 34 years (range, 16 to 57 years) and 84.3 kg (range, 56.8 to 133.3 kg), respectively, and the male-to-female ratio was 14:11. Recipients' diagnoses included lymphoma (11 patients), breast cancer (9 patients), myeloma (4 patients), and acute leukemia (1 patient). Recipients' mean age and weight were 47 years (range, 19 to 64 years) and 79.8 kg (range, 54.9 to 116.9 kg), respectively, and the male-to-female ratio was 11:14.

Of the 23 evaluable donors, the G-CSF doses to be given on transplant days 1, 3, 5, and 7 were administered to 23, 23, 22, and 20 donors, respectively. Four scheduled doses of G-CSF were not administered to 3 donors as a result of a decision to discontinue subsequent scheduled granulocyte collections because of the development of adverse transfusion reactions in 3 recipients with prior infusions of granulocyte components. Of the 23 evaluable donors, the leukapheresis procedures scheduled to occur on days 1, 4, 6, and 8 were performed in 22, 23, 22, and 20 donors, respectively. Five scheduled leukapheresis procedures were not performed either because of adverse transfusion reactions in recipients with prior infusions of granulocyte components (4 patients) or with a red blood cell component given just before a scheduled granulocyte transfusion (1 patient).

After a single dose of G-CSF, the mean donor ANC increased from 3770/μL (2072 to 8458/μL) at baseline to 14 764/μL (3568 to 22 081/μL) just before the first leukapheresis on day 1. The mean donor ANC before the subsequent leukapheresis procedures increased from 27 785/μL (16 687 to 47 830/μL) on day 4; to 30 748/μL (15 333 to 52 628/μL) on day 6; to 45 384/μL (26 396 to 64 490/μL) on day 8. The mean donor platelet count decreased from 215 000/μL (158 000 to 312 000/μL) at baseline to 135 000/μL (102 000 to 231 000/μL) before the fourth scheduled leukapheresis on day 8. The mean donor hemoglobin remained stable throughout (data not shown).

The granulocyte cell dose of components increased with each successive leukapheresis procedure (Table1). The mean granulocyte cell dose (×1010) of components collected on day +1 and transfused on day +2 was 5.6 (1.9-9.3), whereas the component cell dose collected and transfused on day +8 was 9.9 (1.8-16.5). The platelet dose of granulocyte components decreased with each successive leukapheresis procedure (Table 2). The mean component platelet dose (× 1011) was 4.1, 3.3, 2.7, and 2.4 in components collected on days +1, +4, +6, and +8, respectively.

ANC increments (over baseline) observed in the 23 evaluable recipients after transfusions of G-CSF–mobilized granulocyte components are shown in Table 1. Overall, the mean ANC increments were of borderline significance and were sustained for a short term after each of the 4 scheduled days of granulocyte transfusions. For the entire group, the peak mean ANC increment was 1191/μL at hour 4 after transfusion on day 2; 477/μL at hour 8 after transfusion on day 4; 462/μL at hour 8 after transfusion on day 6; and 616/μL at hour 1 after transfusion on day 8. The ANC increments observed at 36 and 48 hours after transfusion of the granulocyte components on day 8 most likely represented endogenous recovery of neutrophils rather than increments resulting from the granulocyte transfusions.

For each recipient, the peak ANC increment occurred at different intervals after the granulocyte transfusions. Thus, when considering this variable, the mean peak ANC increment for the entire group was 1410/μL (109 to 3551/μL) on day 2; 541/μL (0 to 1561/μL) on day 4; 537/μL (0 to 2895/μL) on day 6; and 734/μL (0 to 4100/μL) on day 8 (Table 1). In comparison with peak ANC increments observed after the transfusion of granulocyte components on day 2, granulocyte transfusions administered on subsequent days resulted in lower peak ANC increments in spite of the transfusion of larger granulocyte component cell doses.

Table 3 summarizes the results of the leukocyte compatibility tests. Eight of the 23 recipients had a positive s-LCA, whereas only 1 of 23 recipients had a positive c-LCA against donor cells. In the leukoagglutination cross-match, no clumping was observed in the 20 evaluable tests. HLA-A and HLA-B typing demonstrated that 19 of 23 donor–recipient pairs were 2 of 4 antigen matches. Only 1 donor–recipient pair was an HLA 4/4 antigen match. Given the results of the leukocyte compatibility tests, the effect of leukocyte compatibility on ANC increments after the transfusion of G-CSF–mobilized granulocyte components was analyzed based on a positive or negative s-LCA only. With each of the other 3 tests of leukocyte compatibility, similar analysis could not be performed because of the skewing of most or all data points into 1 potential outcome for each test.

The relationship between the results of the s-LCA and the mean peak ANC increment after transfusions of G-CSF–mobilized granulocyte components is shown in Table 4. On days 2 and 4, the mean peak ANC increments were comparable between the patient cohorts with a positive and a negative s-LCA. However, on days 6 and 8, the mean peak ANC increments after granulocyte transfusion were lower in the cohort with a positive s-LCA in comparison with the cohort with a negative s-LCA. These differences were of borderline significance (P = .05 and P = .06 for days 6 and 8, respectively). The poorer peak ANC increments observed on the latter 2 days of granulocyte transfusions in the cohort with a positive s-LCA (day 6, 246 vs 724/μL, P = 0.05; day 8, 283 vs 1079/μL,P = .06) occurred even though comparable granulocyte component cell doses were administered to the 2 cohorts.

The granulocyte components contained approximately the number of platelets present in a unit of single-donor apheresed platelet components and resulted in significant platelet increments after transfusion (Table 2). Within the limits of the study design, a positive s-LCA did not predict for poorer absolute platelet count increments or CCI at 1 hour after transfusion with granulocyte components administered on days 2, 4, 6, or 8. On each of the 4 days of granulocyte transfusions, comparable absolute platelet increments and CCI occurred 1 hour after transfusion between the recipients with a positive s-LCA and those with a negative s-LCA (Table 2). The platelet doses in granulocyte components were not significantly different between these 2 patient cohorts.

Adverse reactions to granulocyte transfusions occurred in 3 of the 23 evaluable recipients. Each adverse reaction resolved without serious or lasting complications. In one recipient, febrile reactions developed with granulocyte transfusions given on days 2 and 4, and subsequent scheduled granulocyte transfusions were withheld. In another recipient, febrile reactions and hives developed with granulocyte transfusions given on days 4 and 6, and the subsequent scheduled granulocyte transfusion was withheld. Respiratory distress (shortness of breath, wheezing) developed in 1 patient, who required supplemental oxygen for low O₂ saturation (89%) with the granulocyte transfusion given on day 6, and the transfusion scheduled on day 8 was withheld. Thus, of the 87 granulocyte transfusions administered, 5 (5.7%) were associated with adverse reactions.

There was no apparent association between the results of leukocyte compatibility tests and the development of adverse reactions to granulocyte transfusions. Of the 3 recipients in whom adverse transfusion reactions developed, none had a positive s-LCA. Of the 20 recipients in whom adverse transfusion reactions did not develop, 8 had a positive s-LCA. All 3 recipients with adverse transfusion reactions were matched with the donor at 2 of the 4 HLA loci and had negative results in the c-LCA and leukoagglutination assays.

Hematologic recovery was delayed in the recipients of granulocyte transfusions with a positive s-LCA, in spite of transplantation of PBSC components with comparable CD34⁺ cell doses between the cohorts with a positive and a negative s-LCA (Table5). The terminal portion of neutrophil recovery was significantly longer in recipients with a positive s-LCA than in those with a negative s-LCA, a finding that could not be explained by differences in duration of G-CSF administration. The mean duration to achieve an ANC greater than 1000/μL and an ANC greater than 1500/μL was 5.7 and 7.2 days longer in recipients with a positive s-LCA. G-CSF was administered for 17 (range, 12 to 30) days and 14.2 (range, 11 to 18) days in the recipients with a positive s-LCA and a negative s-LCA, respectively (P = .10). The absolute mean number of days to platelet recovery was longer in recipients with a positive s-LCA than in those with a negative s-LCA, though the difference was not statistically significant.

Between days 2 and 10, the interval during which granulocyte transfusions were administered, the number of febrile days was significantly greater in the recipients with a positive s-LCA (Table5). During this interval, recipients with a positive s-LCA experienced a mean of 6.3 (range, 5 to 8) febrile days compared with 4.1 (range, 0 to 6) febrile days in recipients with a negative s-LCA (P = .01). Recipients with a positive s-LCA also required more days of IV antibiotics and amphotericin B and more platelet transfusions than those with a negative s-LCA (Table 5). Between day 2 and neutrophil engraftment, recipients with a positive s-LCA received 10.5 (range, 8 to 13) days of IV antibiotics compared with 7.3 (range, 0 to 11) days of IV antibiotics for recipients with a negative s-LCA (P = .01). This difference could not be explained by a greater proportion of the positive s-LCA cohort on antibiotics at the initiation of granulocyte transfusions on day 2 because the number of days on IV antibiotic administration from the start of the preparative regimen until day 1 was similar between the 2 cohorts. Three patients received amphotericin B after transplantation, and all doses were given between days 6 and 11. The number of days of amphotericin B administered to the positive s-LCA cohort was 1.3 (range, 0 to 5 days), whereas the negative s-LCA cohort did not require amphotericin B (P = .02). The numbers of single-donor apheresed platelet transfusions administered between day 2 and the day of platelet engraftment were 7.6 (range, 1 to 18 days) and 3.4 (range, 0 to 10 days) in recipients with a positive and a negative s-LCA, respectively (P = .05).

In preclinical models and human trials, the primary limitations of granulocyte transfusions included low component cell dose and leukocyte incompatibility.25 Compared with other agents, mobilization of granulocyte components with G-CSF improved component cell dose.11 In addition, transfusion of HLA-matched and G-CSF–mobilized granulocyte components into non-alloimmunized, neutropenic recipients resulted in significant, sustained ANC increments.11,13 14 It can be speculated that the transfusion of G-CSF–mobilized granulocyte components with large cell doses may result in significant ANC increments in recipients with preformed leukocyte antibodies by a saturation effect or through some as yet unknown mechanism. However, the effect of leukocyte compatibility on outcomes with G-CSF–mobilized granulocyte transfusions is unclear.

In this study, the presence of HLA antibodies in recipient serum at baseline (before transfusion), as determined by a positive s-LCA, predicted for poorer peak ANC increments with the latter 2 of 4 G-CSF–mobilized granulocyte transfusions administered on alternate days beginning day 2 to patients who underwent autologous PBSC transplantation. On days 2 and 4, the mean peak ANC increments after granulocyte transfusion were comparable between the cohorts with a positive and a negative s-LCA; however, on days 6 and 8, the mean peak ANC increments after granulocyte transfusion were lower in the cohort with a positive s-LCA. This observation should be confirmed in a larger study, given that the latter differences in peak ANC increments were of borderline statistical significance and that the sample size of the study was small. Previous studies in humans demonstrated an adverse effect of alloimmunization on outcomes with granulocyte transfusions collected at steady state or after mobilization with agents other than G-CSF. In the presence of preformed leukocyte antibodies directed against donor cells, Graw et al26 and Goldstein et al8 observed significantly lower WBC increments and recoveries after transfusion into neutropenic recipients of granulocyte components obtained from chronic myelogenous leukemia donors. McCullough et al15 observed a significant reduction in the intravascular recovery of normal donor granulocytes transfused into recipients with preformed anti-leukocyte antibody and noted that incompatible granulocytes failed to localize to sites of infection. The latter finding was confirmed by Dutcher et al7 in a study of larger numbers of patients. These outcomes in human trials of granulocyte transfusions mirrored those observed in animal studies, in which the presence of alloimmunization also resulted in poorer ANC increments, reduced granulocyte migration, shortened recipient survival time, and increased frequency of infection.2-4,27 Our observations support the hypothesis that the adverse effect of leukocyte incompatibility on outcomes with granulocyte transfusions applies also to components mobilized with G-CSF. Within the range of cell doses (1.3 to 16.5 × 1010) transfused in this study, it appears unlikely that the larger granulocyte component cell doses that occur with G-CSF mobilization can overcome the limitation of leukocyte incompatibility. Although even larger component cell doses may effectively address the problem of leukocyte incompatibility, a collection of such components would require higher doses of G-CSF, the addition of other mobilization agents, or more effective techniques to collect granulocytes. These requirements may expose donors to greater risks or more side effects. A more practical strategy may be to select leukocyte-compatible donors or to focus further research of G-CSF–mobilized granulocyte transfusions on patients who are not alloimmunized.

Preclinical studies clearly established that to ensure the survival of neutropenic dogs with Pseudomonas sepsis, transfused granulocyte components must contain a minimum cell dose, 1 of which resulted in significant ANC increments after transfusion.1Lower component cell doses resulted in no ANC increments and no clinical benefit. These data support the notion that significant post-transfusion ANC increments may be a key condition of demonstrating reproducible improvements in clinical outcomes with granulocyte transfusions in humans and that variables that reduce the ANC increment would likely adversely influence the efficacy of granulocyte transfusions. Our observations of an adverse effect of leukocyte incompatibility on post-transfusion peak ANC increments may have implications regarding the design and analysis of clinical studies attempting to determine the efficacy of G-CSF–mobilized granulocyte transfusions.

Neutropenic recipients of autologous PBSC transplants transfused with G-CSF–mobilized granulocyte components on alternate days over 1 week had significant peak ANC increments after transfusion if the baseline recipient serum s-LCA and c-LCA were negative, even when the donor and recipient were HLA mismatched. Recipients with a baseline serum positive s-LCA and negative c-LCA had significant peak ANC increments after G-CSF–mobilized granulocyte transfusions given on days 2 and 4; however, poor peak ANC increments occurred in the same recipients after granulocyte transfusions given on days 6 and 8. The mechanism of this observation is unclear, but it may be explained by the rapid development of donor leukocyte-specific alloimmunization in recipients with a baseline serum positive s-LCA that did not occur in recipients with a negative baseline s-LCA. Comparable peak ANC increments in the positive and the negative s-LCA cohorts after granulocyte transfusions given on days 2 and 4 likely occurred because only 1 of the 23 evaluable recipients had a positive c-LCA at baseline against donor cells. If this hypothesis is supported by the results of future planned studies, in which recipient serum samples will be serially monitored after transfusion, the presence of a baseline positive s-LCA and a negative c-LCA appeared to predict for rapid development of donor leukocyte-specific alloimmunization after transfusion of G-CSF–mobilized granulocyte transfusions. Conversely, alloimmunization after granulocyte transfusion was unlikely to develop in recipients with negative s-LCA and c-LCA, at least in the 1-week interval during which the transfusions were administered. Our results do not address whether such recipients would later demonstrate evidence of donor-specific alloimmunization. However, results of small studies in which recipients were given either prophylactic or therapeutic granulocyte transfusions mobilized with dexamethasone found that alloimmunization developed frequently and was associated with the number of prior granulocyte transfusions.9 10 

A number of laboratory methods have been used to assess leukocyte compatibility between granulocyte donors and recipients.²⁸These methods include lymphocytotoxicity assays against donor cells or against a panel of HLA-defined cells, leukoagglutination cross-match, granulocyte-specific antibody assays, and HLA typing. Several studies examined the effect of each method on outcomes, with granulocyte transfusions mobilized with agents other than G-CSF, and reported mixed and occasionally contradictory results. If the baseline recipient lymphocytotoxicity cross-match and leukoagglutination assays were negative, Graw et al26observed poorer WBC recovery after granulocyte transfusion in recipients who were HLA disparate with donors. Goldstein et al8 observed an adverse effect of preformed leukocyte antibodies as determined by leukoagglutination and screening lymphocytotoxicity assays on outcomes after granulocyte transfusions. Dutcher et al7 demonstrated that alloimmunization, as defined by a positive lymphocytotoxicity cross-match with the donor, a positive leukoagglutination assay, or both, prevented the migration of transfused granulocytes to sites of infection. In contrast to these reports, McCullough et al6 found that granulocyte agglutinating antibodies, but not granulocytotoxic or lymphocytotoxic antibodies, adversely affected transfused granulocytes. In our study, though we observed a significant relationship between the results of the baseline recipient s-LCA and peak ANC increments after G-CSF–mobilized granulocyte transfusions, we were unable to further elucidate the effect, if any, of the results of the baseline recipient c-LCA, leukoagglutination assay, or donor–recipient HLA disparity. Determination of which method or combination of methods more accurately predicts leukocyte compatibility between donor and recipient is an important area for future investigation, even with G-CSF–mobilized granulocyte transfusions.

Compared with granulocyte transfusions, alloimmunization is perhaps better understood with platelet transfusions. The mechanism of alloimmunization to platelets usually results from the development of antibodies to class I HLA antigens in response to leukocytes contaminating platelet transfusions.29 Granulocytes express class I, but not class II, HLA antigens, and several studies provide evidence to support that antibodies to class I HLA antigens also cause alloimmunization to granulocyte transfusions.7-10 With this information in mind, we anticipated in this study that the presence of HLA antibodies in recipient serum, as determined by a positive s-LCA, would also result in evidence of alloimmunization against contaminating platelets transfused in the granulocyte component. However, a positive s-LCA did not predict for poorer platelet increments at 1 hour after transfusion of granulocyte components on either of the 4 days they were administered. After the final scheduled granulocyte transfusion, it is possible that alloimmunization to platelets occurred more frequently in the cohort with a positive s-LCA, but this was not prospectively addressed in this study. In addition, studies with larger numbers of patients may be required to show a significant effect of the s-LCA result on alloimmunization against platelets contained in granulocyte components.

Granulocyte transfusions are associated with a number of potential adverse reactions, including fever, chills, hives, respiratory distress, or hypotension. In several reports, adverse reactions to granulocyte transfusions were associated with the presence of leukocyte antibodies.9,10 In this study, adverse reactions to G-CSF–mobilized granulocyte transfusions occurred in only 3 of the 23 evaluable recipients and in only 5.7% of all granulocyte transfusions administered. This compares to a prior report in which the frequency of adverse reactions to G-CSF–mobilized granulocyte transfusions that were HLA matched was 3.4%.13 In contrast to previous reports,9 10 we observed no obvious association between the results of baseline leukocyte compatibility tests and development of adverse reactions to granulocyte transfusions. However, all recipients in our study were premedicated with acetaminophen and diphenhydramine. The frequency of adverse reactions may have been greater if no premedications had been given.

In this study, the presence of HLA antibodies in recipient serum at baseline, as determined by a positive s-LCA, predicted for delayed neutrophil engraftment after autologous PBSC transplantation, in spite of the transplantation of PBSC components with comparable CD34⁺ cell doses between the cohorts with a positive and a negative s-LCA. Because neutrophil engraftment occurred 9 or more days after transplantation and after the final granulocyte transfusion, the more significant ANC increments observed in the negative s-LCA cohort with the latter 2 granulocyte transfusions did not confound the interpretation of neutrophil engraftment in this cohort and cannot explain our unique observation. The mechanism of the adverse effect of a positive s-LCA on neutrophil engraftment after autologous transplantation is unclear but may be explained by the positive s-LCA cohort reflecting a group of patients who were more heavily pretreated. Such patients would likely have received more prior blood product transfusions, which may explain why patients in this cohort had preexisting HLA antibodies. The presence of preexisting HLA antibodies in the positive s-LCA cohort was also associated with a greater number of febrile days, days of IV antibiotics, and platelet transfusions compared with the cohort with a negative s-LCA. Additional studies of larger numbers of patients should be performed to confirm the adverse effect of a positive s-LCA on neutrophil engraftment and on clinical outcomes. If these observations are confirmed, the s-LCA may be used to risk-stratify patients who undergo autologous PBSC transplantation based on the likelihood for adverse clinical outcomes with the transplantation procedure.

The data from this study support the conclusion that leukocyte incompatibility adversely affects ANC increments after G-CSF–mobilized granulocyte transfusions and clinical outcomes after PBSC transplantation. Future trials designed to determine the clinical efficacy of G-CSF–mobilized granulocyte transfusions should consider the potential adverse effect of leukocyte incompatibility on outcomes. Strategies effective in reducing the incidence of alloimmunization after platelet transfusion, such as ultraviolet irradiation, should be evaluated with G-CSF–mobilized granulocyte transfusions.23