Inhaled granulocyte/macrophage colony-stimulating factor as treatment of pneumonia-associated acute respiratory distress syndrome.

To the Editor:

Granulocyte/macrophage colony–stimulating factor (GM-CSF) is a myeloid growth factor, which induces proliferation and differentiation of macrophages and dendritic cells. Recently, several publications highlighted GM-CSF as highly protective in different preclinical models of pneumonia-associated lung injury, including influenza virus, Klebsiella pneumoniae, and Streptococcus pneumoniae infection and others (1–4). Importantly, GM-CSF protects the host in both the early phase of acute lung infection and during the stage of regeneration of the injured lung epithelium. GM-CSF applied to or expressed in the distal airways promotes bacterial or viral clearance by expanding alveolar macrophages or CD103⁺ migratory lung dendritic cells or by stimulating their host defense capacity (1–4). GM-CSF is considered an important factor for terminal differentiation and classical activation of alveolar macrophages via PU.1- and IRF5-dependent transcriptional programs, which is discussed as an underlying mechanism of improved macrophage host defense capacity in response to GM-CSF (5, 6). Furthermore, GM-CSF exerts multiple beneficial effects on the epithelial lung barrier, as it protects alveolar epithelial cells against oxidative stress–induced mitochondrial injury (7), and it promotes type II alveolar epithelial cell proliferation after LPS challenge (8). Of note, mice overexpressing GM-CSF in type II alveolar epithelial cells to high local levels were strongly protected from infection and injury (1, 3, 4, 8), suggesting that high levels of GM-CSF in the alveoli or the alveolar lining fluid should be achieved for putative acute lung injury (ALI)/acute respiratory distress syndrome (ARDS) treatment. Notably, GM-CSF levels in bronchoalveolar lavage (BAL) fluid were correlated with improved outcome of patients with ARDS (9), and early clinical studies suggested that it might improve respiratory function in patients with sepsis (10). Together, these data provide strong evidence that GM-CSF improves host defense, attenuates epithelial cell injury, promotes repair of the epithelium, and improves barrier function and gas exchange in ALI/ARDS (11).

We report on six patients with pneumonia-associated ARDS who received GM-CSF as compassionate treatment. Given that systemic GM-CSF delivery in patients with sepsis-induced lung injury failed to improve outcome (12), and considering that high local concentrations of GM-CSF in the alveolar lining fluid are a prerequisite for its beneficial effect in preclinical ARDS models, we applied GM-CSF per inhalation. The University Ethics Committee was informed about the compassionate treatment (informed consent was obtained from the legal representative of each patient) and had approved use of BAL biobank material and data derived from patients subjected to sequential bronchoscopy for diagnostic purposes. Compassionate GM-CSF treatment was provided to patients who were diagnosed with moderate to severe community-acquired pneumonia– or ventilator-associated pneumonia–associated ARDS according to the Berlin definition criteria (mean ± SEM baseline Pa_O2/Fi_O2 = 91.5 ± 13.3), received antibiotic treatment according to guidelines, and did not improve with respect to oxygenation after ≥6 days of mechanical ventilation, with or without additional extracorporeal membrane oxygenation support. One hundred twenty-five micrograms of recombinant GM-CSF (Sargramostim, Leukine, Bayer HealthCare Pharmaceuticals, Seattle, WA) were applied by Aeroneb Solo device (Covidien, Neustadt, Germany) at an interval of 48 hours (Figure 1A). To exclude any influence of the nebulization procedure on efficacy of the drug, we compared concentrations and bioactivity of the GM-CSF solution before and after aerosolization by ELISA and by quantification of GM-CSF-induced STAT5 activation in cultured A549 cells by Western blot, respectively (13), and did not reveal substantial differences.

Figure 1

(A) GM-CSF treatment regimen. [More]

We compared oxygenation of GM-CSF–treated patients with patients who displayed similar disease conditions (n = 4, mean ± SEM baseline Pa_O2/Fi_O2 = 118 ± 23.4, Table 1) but had not received GM-CSF. We recognized a significant improvement of oxygenation in response to GM-CSF inhalation during the observed time course (difference in slopes: 1.2 ± 0.4 (ΔPa_O2/Fi_O2)/d, P = 0.0035, Figure 1B). Analyses of lung compliance (C_st = Vt/[P_plat − PEEP], where C_st is static lung compliance, P_plat is plateau pressure, and PEEP is positive end-expiratory pressure) revealed a mean increase of ∼40% after 10 days post–GM-CSF inhalation, which was not observed in nontreated patients (P = 0.0787; linear mixed-effects models fit by restricted maximum likelihood [REML], not shown). With respect to morbidity scores, SAPS (simplified acute physiology score) values were 35.3 ± 6.6 and 41.8 ± 6.8 in GM-CSF–treated and nontreated patients at baseline, respectively, with significant improvement over time only in the GM-CSF–treated patients (P = 0.036; analyzed using linear mixed-effects models fit by REML; data not shown). Mean SOFA (sequential organ failure assessment) scores were 9.33 ± 1.09 in GM-CSF–treated and 10.5 ± 2.22 in untreated patients at Day −1 and improved to 6.62 ± 1.52 over time in GM-CSF–treated patients, whereas the scores slightly increased to 11.0 ± 3.03 in nontreated patients (P = 0.068; linear mixed-effects models fit by REML; data not shown).

Table 1: Characteristics and Outcome of Analyzed Patients with Acute Respiratory Distress Syndrome

GM-CSF is considered to drive M1 polarization and to improve host defense functions and survival of alveolar macrophages in vitro and in animal models in vivo (1, 4–6). To verify these beneficial effects in GM-CSF–treated patients, BAL macrophages (defined as FSC [forward scatter]^highCD3/CD19/CD14^negCD11c^highCD123^negCD68^posCD64^pos) were subjected to fluorescence-activated cell sorter (FACS)-based analyses of their polarization profiles (14). Alveolar macrophages of non–GM-CSF–treated patients displayed a progressive shift toward M2 polarization, demonstrated by enhanced CD206 expression. In contrast, GM-CSF inhalation promoted an M1 phenotype with increased CD80 and persistently low CD206 expression (Figure 1C, left and middle panels), thus demonstrating delivery of GM-CSF to the injured alveolar compartment. In addition, GM-CSF inhalation increased activation of alveolar mononuclear phagocytes (CD3/CD19/CD14^negCD11c^highCD123^neg), as determined by quantification of HLA-DR expression, most prominent at Days 3–4 after treatment, whereas HLA-DR expression remained persistently low in untreated patients (Figure 1C, right panel). This difference in polarization and activation profiles suggests that improved macrophage host defense capacity may underlie the beneficial effect of inhaled GM-CSF. Notably, GM-CSF treatment did not induce increased neutrophil influx into the alveolar compartment, as verified by FACS analyses (data not shown). Evaluation of standard blood parameters during a 10-day course after the first treatment (urea, AST, and ALT) did not reveal organ toxicity related to GM-CSF treatment (data not shown).

We recognize that these data are derived from a small number of patients and were not obtained under controlled trial conditions. Nonetheless, our findings provide the first evidence that the application of GM-CSF by the inhalation route may represent an effective strategy to drive pulmonary host defense and improve oxygenation and outcome in pneumonia-associated ARDS. A controlled clinical trial is currently in preparation.