Granulocyte transfusions in hematologic malignancy patients with invasive pulmonary aspergillosis: outcomes and complications

Abstract

Background

Granulocyte transfusions (GTXs) have been used successfully as an adjunctive treatment option for invasive infections in some neutropenic patients with underlying hematologic malignancy (HM).

Patients and methods

We sought to determine the impact of GTX as an adjunct to antifungal therapy in 128 patients with HM and prolonged neutropenia (≥14 days) with a proven or probable invasive aspergillosis (IA) infection by retrospectively reviewing our institutional database.

Results

Fifty-three patients received GTX and 75 did not. By univariate analysis, patients with invasive pulmonary aspergillosis who received GTX were less likely to respond to antifungal therapy (P = 0.03), and more likely to die of IA (P = 0.009) when compared with the non-GTX group. Among patients who received GTX, 53% developed a pulmonary reaction. Furthermore, IA-related death was associated with the number of GTX given (P = 0.018) and the early initiation of GTX within 7 days after starting antifungal therapy (P = 0.001). By multivariate competing risk analysis, patients who received GTX were more likely to die of IA than patients who did not receive GTX (P = 0.011).

Conclusions

Our study suggests that GTX does not improve response to antifungal therapy and is associated with worse outcomes of IA infection in HM patients, particularly those with pulmonary involvement.

introduction

Neutropenia is a leading risk factor for the acquisition of an invasive aspergillosis (IA) in patients who have an underlying hematologic malignancy (HM) and have undergone hematopoietic stem cell transplantation (HSCT), and when persistent indicates poor prognosis [1, 2]

In the 1970s, granulocyte transfusions (GTXs) were introduced to treat neutropenia in HM patients with severe infections [3]. However, the efficacy of this therapeutic modality was questioned, especially after some patients had severe pulmonary reactions during GTX; therefore, enthusiasm for the treatment declined [4–6].

In the 1990s, with the introduction of a granulocyte colony-stimulating factor (GCSF), it became possible to stimulate bone marrow to produce a high concentration of granulocytes for collection and transfusion [7]. This has increased the number of granulocytes available for transfusion and subsequently rekindled interest in using GTX for neutropenia in patients with an IA.

Several studies have demonstrated the relative safety of GCSF-stimulated GTX, particularly in pediatric patients [8–10]. In addition, prospective, controlled trials have suggested that prophylactic high-dose GTX might reduce the risk of death from infection in neutropenic patients [11]. However, the role of GTX as an adjunct treatment of documented IA in neutropenic patients has not been well examined [12, 13]. Most of the available data supporting the use of GTX in the treatment of fungal infections are based on case reports, cohort studies, and anecdotal experience [14–16].

In this study, we examined the use of GTX as an adjunct to antifungal therapy for the treatment of documented IA in patients with an underlying HM by comparing the outcomes of patients who received GTX with the outcomes of those who did not. We evaluated the response to antifungal agents, overall and IA-related mortality rates, and adverse effects associated with GTX.

design and methods

We conducted a retrospective chart review at The University of Texas MD Anderson Cancer Center (RCR02–297; Houston, TX) between June 1993 and March 2010. Using Infectious Diseases/Infection Control Department database, we identified all patients with an underlying HM who had neutropenia for ≥14 days following diagnosis of a documented proven or probable IA and were treated with antifungal agents active against IA. We used standardized forms to collect relevant demographic and clinical information of patients from the institutional patient records. Patient characteristics include age, gender, type of underlying HM, persistent neutropenia, signs and symptoms of IA infection at the initiation of antifungal treatment, use of steroids or tacrolimus, type of immunotherapy used during infection including granulocyte colony-stimulating factor (GCSF), granulocyte-macrophage colony-stimulating factor (GM-CSF) and γ-interferon, intensive care unit (ICU) admission, use of mechanical ventilation, use of antifungal prophylaxis before the IA, and antifungal therapy regimen used. We also collected information about HSCTs within 1 year before IA infection diagnosis or during antifungal treatment and the presence of graft-versus-host disease (GHVD). We also recorded the Aspergillus species and the antifungal therapy used.

GTX donors were patients' relatives or unrelated healthy individuals and were screened according to the US Food and Drug Administration criteria. The donors' immune system was primed and granulocyte specimens were subsequently collected [17]. Briefly, the donors were given a single dose of dexamethasone and GCSF 24 h before neutrophil specimens were collected. Centrifuge leukapheresis yielded ∼5.5 × 10¹⁰ neutrophils per transfusion [17]. The GTXs, after having been irradiated at a dose of 25 cGy, were given daily or on alternating days, depending on the availability of the donors.

definitions

Proven and probable IA were defined according to the criteria of the European Organization for Research and Treatment of Cancer and Mycoses Study Group [18]. The day of diagnosis is the day on which the first microbiologic or histopathologic evidence of IA was collected. Persistent neutropenia is defined as ANC of <500 cells/μl³ that persisted during antifungal therapy following the diagnosis of IA. Primary antifungal therapy was defined as the first antifungal regimen (single agent or combination) that was administered for at least seven consecutive days. Salvage therapy is any regimen administered after primary antifungal therapy.

A favorable response to therapy was defined as a complete or partial clinical, radiologic, or microbiologic resolution of the IA infection. Failure to respond was characterized by progressive or unchanged clinical or radiologic parameters.

Overall mortality included all patient deaths within 12 weeks of initiating antifungal therapy. IA-related mortality included deaths in patients with documented antemortem or postmortem radiographic, microbiologic, or histologic findings suggestive of active IA that did not have a sustained favorable response to treatment.

statistical analysis

Categorical variables were compared using chi-square or Fisher's exact tests and continuous variables compared using Wilcoxon rank-sum tests. Multiple logistic regression analysis was used to evaluate the independent effect of GTXs on patients' favorable response, as necessary. In addition, competing risk analyses were carried out to evaluate the effect of GTX on IA-related mortality, using death due to other causes as a competing event. All tests were two-sided tests with a significance level of 0.05. The competing risk analyses were carried out using the statistical software R version 2.8.1 (R Development Core Team. 2008) and all other statistical analyses were carried out using SAS version 9.1 (SAS Institute Inc., Cary, NC).

results

We identified 128 HM patients with a proven or probable IA and prolonged neutropenia (duration ≥14 days) of whom 53 (41%) received GTX and 75 (59%) did not. The majority were leukemia patients (89% in the GTX group versus 84% in the non-GTX group). Demographic and clinical characteristics were compared between the two groups of patients (Table 1). There was no significant difference in patients' median age (44 years in the GTX group and 54 years in the non-GTX group, P = 0.14). Of the 27 HSCT patients, those in the GTX group were less likely to have undergone allogeneic HSCT than were those in the non-GTX group (67% and 100%, P = 0.029). Patients in the GTX group were less likely to have had invasive pulmonary infection (60% versus 83%, P = 0.005) and more likely to have had localized infection (26% versus 8%, P = 0.005) than patients in the non-GTX group. Most patients had been neutropenic at the onset of IA (89% for GTX and 81% for non-GTX). The median duration of neutropenia during infection was 24 days for GTX and 26 days for the non-GTX group, and 60% of patients in the GTX group and 52% of patients in the non-GTX group were persistently neutropenic.

Table 1.

Characteristics of neutropenic hematologic malignancy (HM) patients who did or did not receive a GTX to treat invasive aspergillosis (n = 128)

Characteristics	GTX (n = 53) N (%)	Non-GTX (n = 75) N (%)	P
Age (years), median (range)	44 (9–75)	54 (7–83)	0.14
Gender, male	32 (60)	50 (67)	0.47
Duration of neutropenia during infection (days), median (range)	24 (14–92)	26 (14–197)	0.86
Type of invasive aspergillosis infection
Invasive pulmonary infection	32 (60)	62 (83)	0.005
Disseminated infection	7 (13)	7 (9)	0.49
Localized infection (including sinus infection)	14 (26)	6 (8)	0.005
Type of HM
Leukemia	47 (89)	63 (84)	0.45
Lymphoma	4 (8)	10 (13)	0.30
Myeloma	1 (2)	1 (1)	>0.99
Hematopoietic stem cell transplant within prior year or during infection	9 (17)	18 (24)	0.34
Type of transplant			0.029
Allogeneic	6/9 (67)	18/18 (100)
Autologous	3/9 (33)	0/18 (0)
Graft-versus-host disease (GHVD)	2 (4)	11 (15)	0.07
Neutropenia at the onset of invasive aspergillosis infection	47 (89)	61 (81)	0.26
Persistent neutropenia	32 (60)	39 (52)	0.35
Symptoms of invasive aspergillosis infection at baseline
Fever	44 (83)	58 (77)	0.43
Cough	24 (45)	46 (61)	0.07
Hemoptysis	4 (8)	11 (15)	0.22
Shortness of breath	20 (38)	35 (47)	0.31
Pleuretic chest pain	5 (9)	12 (16)	0.28
Sinus symptoms	13 (25)	9/74 (12)	0.1
Aspergillus species isolated
Aspergillus fumigatus (alone or with others)	14/49 (29)	19/69 (28)	0.9
Aspergillus terreus (alone or with others)	20/47 (43)	18/67 (27)	0.08
Aspergillus flavus (alone or with others)	13/47 (28)	20/67 (30)	0.8

Therapeutic interventions were compared between the two groups (Table 2). Eighty-three percent of patients in the GTX group and 85% of patients in the non-GTX group had been treated with steroids during the month preceding IA diagnosis or during infection. Patients in GTX and non-GTX groups were similar in having received immunotherapy during infection (92% versus 84%, P = 0.15). The two groups were comparable in receiving GCSF and γ-IFN treatment, but patients in the GTX group were more likely to have received GMSF treatment (43% versus 23%, P = 0.013). Patients who received GTX were more likely to have received antifungal prophylaxis before therapy (83% versus 64%, P = 0.018). The two groups were comparable in using a lipid amphotericin B-containing regimen, echinocandins-containing regimen, and azole-containing regimen.

Table 2.

Interventions used for and therapeutic outcomes of neutropenic hematologic malignancy (HM) patients with invasive aspergillosis who did and did not receive a GTX (n = 128)

Interventions or outcomes	GTX (n = 53) n (%)	Non-GTX (n = 75) n (%)	P
Steroid use	44 (83)	64 (85)	0.72
Tacrolimus use	6 (11)	13 (17)	0.35
Immunotherapy during infection	49 (92)	63 (84)	0.15
Granulocyte-macrophage colony-stimulating factor (GM-CSF)	23 (43)	17 (23)	0.013
Granulocyte colony-stimulating factor (GCSF)	41 (77)	58 (77)	>0.99
γ-Interferon	5 (9)	2 (3)	0.13
Intensive care unit (ICU) stay	23 (43)	34 (45)	0.83
Duration of ICU stay (days), median (range)	10 (2–38)	11 (3–58)	0.55
Mechanical ventilation	18 (34)	22 (29)	0.58
Duration of mechanical ventilation (days), median (range)	9 (1–35)	11 (2–58)	0.91
Antifungal prophylaxis before diagnosis of invasive aspergillosis infection	44 (83)	48 (64)	0.018
Antifungal regimen used (primary or salvage therapy)
Lipid formulation of amphotericin B-containing regimen	42 (79)	60 (80)	0.92
Echinocandin-containing regimen	23 (43)	34 (45)	0.83
Azole-containing regimen	23 (43)	38 (51)	0.42
Total therapy duration (days), median (range)	34 (8–401)	43 (4–335)	0.14
Favorable response to antifungal therapy	8 (15)	23 (31)	0.06
Aspergillosis-attributable death within 12 weeks of initiating antifungal therapy	32 (60)	30 (40)	0.023
Death from any cause within 12 weeks of initiating antifungal therapy	37 (70)	41 (55)	0.08

A chi-square test showed that 15% of patients in the GTX group and 31% of those in the non-GTX group had a favorable response at the end of therapy (P = 0.06) (Table 2). In the subset analysis of patients with invasive pulmonary aspergillosis, 9% of patients in the GTX group compared with 29% of those in the non-GTX group had a favorable response (P = 0.03). Multivariate logistic regression analyses showed no significant association between GTX and response. Three prognostic factors were found to be independently associated with patients' final response: persistent neutropenia (P = 0.002), infection with A. terreus species (P = 0.014), and treatment with an azole-containing regimen (P = 0.003). On the other hand, patients on the azole-containing regimen were more likely to have a favorable response (odds ratio 6.1, 95% CI: 1.9 to 19.8). After adjusting for these three factors, a multiple logistic regression analysis showed that GTX had no significant effect on response (P = 0.52).

There was no significant difference in the overall mortality rate between the GTX and non-GTX groups within 12 weeks of initiating therapy (70% versus 55%, P = 0.08) (Table 2). However, the GTX group have a significantly higher IA-related mortality rate than the non-GTX group (60% versus 40%, P = 0.023). In patients with invasive pulmonary infections, the GTX group was also found to have a significantly higher IA-related mortality rate than the non-GTX group (69% versus 40%, P = 0.009). Competing risk analyses was carried out to assess the effect of GTX on IA-related mortality. The cumulative incidence curves of IA-related deaths for the two groups were estimated and compared by a univariate competing risk analysis (Figure 1), using death due to other causes as a competing event. It showed that patients in the GTX group had significantly higher probability of IA-related death within 12 weeks of initiating therapy than patients in the non-GTX group (P = 0.018). Then the independent effect of GTX was evaluated by a multivariate competing risk analysis. Five factors were identified to be significantly associated with aspergillosis-related death: persistent neutropenia, shortness of breath at baseline, ICU stay, treatment with azole-containing regimen, and treatment with GTX (Table 3). Patients treated with the azole-containing regimen were less likely to have an aspergillosis-related death (P < 0.0001). On the other hand, patients with persistent neutropenia (P = 0.001), shortness of breath at baseline (P = 0.003), ICU stay (P < 0.0001), or GTX therapy (P = 0.011) were more likely to have an aspergillosis-related death. Specifically, patients who received GTX were two times (95% CI: 1.2–3.3) more likely to have an aspergillosis-related death than those who did not receive GTX.

Figure 1.

Cumulative incidence curves of IA-related deaths for prolonged neutropenic hematologic malignancy (HM) patients with aspergillosis who received GTXs and those who did not (non-GTX) by a competing risk analysis, using death due to other causes as a competing event (P = 0.018) (n = 128).

Table 3.

Multivariate competing risk model for aspergillosis-related mortality, using death due to other causes as a competing event (n = 128)

Variables	N	Hazard ratio	95% confidence interval	P
Persistent neutropenia				0.001
Yes	71	2.6	1.5, 4.7
No	57	1.0
Shortness of breath at baseline				0.003
Yes	55	2.2	1.3, 3.8
No	73	1.0
Intensive care unit (ICU) stay				<0.0001
Yes	57	3.0	1.8, 5.1
No	71	1.0
Azole-containing regimen in primary or salvage antifungal therapy				<0.0001
Yes	61	0.3	0.2, 0.5
No	67	1.0
Received GTX				0.011
Yes	53	2.0	1.2, 3.3
No	75	1.0

Of the 53 patients who received GTX, 24 (45%) developed fever, one (2%) developed skin rash, and 28 (53%) developed pulmonary reactions characterized by worsening shortness of breath and/or pulmonary infiltrates within 48 h of receiving GTX. GTX recipients who developed pulmonary reactions tended to have a higher rate of IA-related deaths within 12 weeks of follow-up than the GTX recipients who did not develop pulmonary reactions (71% versus 48%, P = 0.08). The majority of patients who received GTX had also received G-CSF or GM-CSF. We found no association between pulmonary complications and G-CSF/GM-CSF treatment in patients receiving GTXs (P = 0.61).

Fifty-three patients received one or more GTX after antifungal therapy initiation and four of them started to receive GTX within 1 week before therapy. The median number of GTX received was 7 (range: 1–44) during the course of infection. The median duration between therapy initiation and the first GTX received was 7 days (range: 0–69 days). Of the 53 patients, 37 (70%) died within 84 days of antifungal therapy initiation and the median duration between their last GTX and death was 5 days (range: 0–52 days). This duration was significantly shorter in patients with IA-related deaths than in those who died of other causes (median: 4 days versus 22 days, P = 0.022). The subset analysis of patients with invasive pulmonary infections showed that the patients with IA-related deaths had a shorter survival time after their last GTX than those who died of other causes (median duration: 3 days versus 31 days, P = 0.024). We also found that patients who received GTX within 1 week after antifungal therapy initiation were more likely to die of IA-related causes than those who received GTX later (81% versus 38%, P = 0.001). This was also true in patients with invasive pulmonary infections where 85% of those who received GTX within 1 week after antifungal therapy initiation died of IA-related causes, whereas 42% of those who received GTX later died of IA (P = 0.018). When looking at overall mortality, we found that patients who received GTX within 1 week after antifungal therapy initiation were more likely to die than those who received GTX later (89% versus 50%, P = 0.002) and in those with invasive pulmonary infections as well (95% versus 58%, P = 0.019). Finally, of all the 128 patients in the study, the patients with IA-related deaths were found to have received more GTX than those without IA-related death (median number of GTX: 1 versus 0, P = 0.018). There was a trend suggesting a possible association between the number of GTX received and overall mortality (P = 0.07).

discussion

Our study showed that GTX did not improve the response of IA to antifungal therapy and was associated with an increased IA-related mortality in HM patients with prolonged neutropenia. Surprisingly, our data showed that GTX were associated with increased IA-related mortality despite the fact that the non-GTX group had similar or worse prognostic factors. For example, patients in the non-GTX group were less likely to have received antifungal prophylaxis and have a significantly higher rate of invasive pulmonary aspergillosis or allogeneic HSCT before the IA diagnosis. These factors are considered to be associated with worse outcomes of IA [19–21].

Furthermore, the subgroup of patients with invasive pulmonary aspergillosis who received GTX had a worse response to antifungal therapy (P = 0.03) and had a significantly higher rate of IA-related mortality when compared with those who did not receive GTX (P = 0.009).The earlier the patient received GTX, the greater the risk of death, particularly IA-related death. In addition, the number of GTX treatments received was significantly associated with IA-related mortality. In 53% of patients, severe pulmonary reactions characterized by worsening shortness of breath or pulmonary infiltrates were the most frequent adverse effects of GTX.

GTX have been used in patients with prolonged neutropenia or neutrophil dysfunction to minimize this risk. In a meta-analysis of eight randomized, controlled trials in which GTX therapy was used, it was shown that the risk of mortality from infection was reduced when used as prophylaxis [11]. The prophylactic use of GTX therapy in neutropenic allogeneic HSCT recipients has also significantly but modestly helped in reducing days of fever and intravenous antimicrobial usage. However, this modest effect raised the question of the cost-effectiveness of this intervention [22].

In HM patients with IA, prolonged and persistent neutropenia is associated with a poor outcome. As shown by a recent study of 449 neutropenic HM patients with IA, persistent neutropenia was independently associated with failure to respond to antifungal therapy and increased risk of death from IA [2].

Unlike its benefit as prophylaxis, GTX as an adjunct treatment of bacterial or fungal infections has been based on case reports and cohort studies that suggest potential benefits with no clear evidence of a positive benefit–risk ratio [12–16, 23]. Our findings regarding the lack of efficacy of GTX therapy are in agreement with those of Seidel et al., who randomly assigned 74 febrile neutropenic malignancy patients (49 with IA) to receive either antimicrobial therapy alone or plus GTX and found similar clinical outcomes [24].

In our study, the unfavorable outcomes associated with GTX could be related to lung injury sustained during the transfusion. Our data showed that 53% of the patients who received GTX had a pulmonary reaction within 48 h of transfusion. This reaction was the most frequent and serious adverse effect, often resulting in transfer to the ICU. A study by McCullough et al. using indium-111-labeled neutrophils transfused to patients showed that granulocytes were abnormally sequestered in the lungs after exposure to granulocyte alloantibodies [25]. Dutcher et al. also demonstrated that leukocyte alloantibodies led to the pulmonary sequestration of transfused granulocytes and a subsequent significant decrease in circulating granulocyte survival, resulting in failure of the granulocytes to localize at the sites of infection [26, 27]. This transfusion-related acute lung injury (TRALI) occurs as the recipient develops leukocyte antibodies against the transfused granulocytes from the donor [28].

TRALI has been well described over the past four decades in patients who receive GTX. The pulmonary fungal infection plays a role in activating the pulmonary endothelium and often adds to the sequestration of transfused granulocytes in the pulmonary circulation [28–31]. This observation is supported by the fact that the pulmonary aspergillosis in particular was significantly associated with higher IA-related mortality (P = 0.009) and worse response (P = 0.03) to antifungal therapy in patients who received GTX. Hence, it is most likely that TRALI might be the cause of the pulmonary reaction occurring in >50% of the GTX recipients in our current study leading to the increased IA-related mortality.

Since an anti-human leukocyte antigen and anti-granulocyte antibodies have been implicated in the failure of GTX [25, 26, 32], several investigators studied the efficacy of GTX using compatible matched granulocytes and demonstrated reduced mortality [33]. Nevertheless, the optimal method for granulocyte compatibility testing is not well established [34]. Future trials may help answer the question of whether the efficacy and safety of GTX are related to alloimmunization.

Although some studies found that the lethal pulmonary reactions noted in patients receiving GTX were associated with the use of amphotericin B [4], other studies did not find such any association which was also the case in our study [5, 35].

Although the occurrence of pulmonary reactions within 48 h following GTX suggests that they might be mostly related to TRALI and a pulmonary leukoagglutinin reaction [36], other factors such as infections known to cause pulmonary disease (e.g. cytomegalovirus and human herpesvirus 6) that could be introduced with the transfused granulocytes should also be considered [37, 38]. Although the risk of cytomegalovirus transmission through GTX has been reported to be <5%, the disease is devastating and can lead to serious respiratory complications in patients with HM or have undergone HSCT [38]. Furthermore, in a recent study of leukemia patients at our institution, GTX was also significantly associated with human herpesvirus 6 viremia [39].

The present study was subject to several potential limitations. Because of the retrospective design of the study, the patients were treated and received GTX at the discretion of the primary physician. The patients were not on a defined prospective clinical protocol with specific inclusion and exclusion criteria. Therefore, we could not assess whether patients who received GTX were perceived by their physician to be sicker or deteriorating rapidly. Patients receiving GTX may have been more severely ill than non-GTX recipients, indicated also by the fact that more patients in the GTX group had been given systemic antifungal prophylaxis. GTX could have been prescribed as a last resort to salvage the patients. This might have caused a bias disfavoring GTXs. However, our results are consistent with what was suggested by Stanworth who found no clear evidence from the published randomized, controlled trials to support or refute the use of GTX in patients with neutropenia [13]. Further well-designed prospective trials are required to evaluate the efficacy of this intervention in these patient populations.

In conclusion, we found that GTX did not improve the response of IA infection to antifungal therapy and was associated with an increased IA-related mortality rate in HM patients with prolonged neutropenia. GTX was also associated with an acute pulmonary reaction and respiratory deterioration in more than half of the GTX recipients, which might have contributed to the worse outcomes associated with this intervention. Considering the risks and costs of GTX treatment, as well as the relatively high logistical requirements of daily GTX and its lack of efficacy in helping treat IA infection, its use in HM patients with an IA should be avoided or considered with caution.

disclosure

The authors have declared no competing financial interests.