Effect of Neutropenic Critical Illness on Development and Prognosis of Acute Respiratory Distress Syndrome

To the Editor:

Previous neutrophil recovery investigations (1) established the link between neutrophil recovery, organ injury, such as acute respiratory distress syndrome (ARDS), and death, supporting the essential role of the neutrophil in the pathogenesis of ARDS. However, such a link between neutropenia itself and ARDS development and/or prognosis is not yet established.

In this study, we had two separate but related goals. First, by using patient-level data from three well-phenotyped ICU cohorts, we sought to determine whether neutropenia is independently associated with the development of ARDS. Subsequently, by using patient-level data from randomized controlled trials performed by the ARDS Network (ARDSNet), we sought to determine if neutropenia is associated with the prognosis of ARDS.

Some of the results of these studies have been previously reported in the form of an abstract (2).

Methods

We performed an analysis of prospectively collected data from critically ill subjects enrolled in three ICU cohorts concurrently with a secondary analysis of patient-level data from subjects with ARDS enrolled in six ARDSNet trials.

ICU cohorts

Subjects were enrolled in three ongoing ICU cohorts of critically ill patients at New York-Presbyterian–Weill Cornell Medicine (WCM), Brigham and Women’s Hospital (BWH), and Asan Medical Center (AMC). Details have been given previously (3). The study protocol was approved by the institutional review boards of WCM (1405015116), BWH (2008-P-000495), and AMC (2011–0001).

For the ICU cohorts, ARDS was adjudicated independently by at least two physicians within the first 5 (WCM) or 7 (BWH and AMC) days of ICU admission. ARDS was defined according to the Berlin definition.

Neutropenia was determined on ICU Day 0 and defined as total white blood cell (WBC) count <1.0 × 10³ cells/μl (severe neutropenia: 0.5 × 10³ cell/μl).

Sepsis was defined according to Sepsis-3 criteria, and severity of illness was defined according to the sequential organ failure assessment score modified to remove the platelet component (modified sequential organ failure assessment score). The primary outcome was the association between neutropenia and the development of ARDS after adjusting for variables known to be associated with the development of ARDS, namely, pneumonia and nonpulmonary sepsis.

ARDSNet trials

Subjects from ARDSNet trials were enrolled in the ARMA (Low-Tidal-Volume Trial), ALVEOLI (Assessment of Low Tidal Volume and Elevated End-Expiratory Volume to Obviate Lung Injury), FACTT (Fluid and Catheter Treatment Trial), ALTA (Albuterol for the Treatment of Acute Lung Injury), EDEN (Early vs. Delayed Enteral Nutrition), and SAILS (Statins for Acutely Injured Lungs from Sepsis) trials. As previously (10), access to the ARDSNet trial data was received after submission of our protocol (available on request), including a power analysis and prespecified analyses, to the NIH/NHLBI Biologic Specimen and Data Repository Information Coordination Center. The WCM institutional review board granted a nonhuman subjects research waiver for this study (1708018504).

Neutropenia was determined on enrollment Day 0 and defined as total WBC count <1.0 × 10³ cells/μl.

The primary outcome was the association between neutropenia and 60-day mortality after adjusting for variables known to be associated with mortality in ARDS, namely age, liver failure, malignancy, and Acute Physiology and Chronic Health Evaluation III (adjusted to remove the total WBC count component).

Results

ICU cohorts

Of the 1,229 subjects enrolled in the three ICU cohorts, 87 (7.1%) had neutropenia. Baseline characteristics are shown in Figure 1. Subjects with neutropenia were more likely to have nonpulmonary sepsis (62.1% vs. 34.6%; P < 0.001) than subjects without neutropenia.

Figure 1.

Baseline characteristics of subjects enrolled in (A) the ICU cohorts (blue) and (B) the ARDSNet trials (gray). *Pneumonia and nonpulmonary sepsis include subjects with either sepsis or septic shock. ^†Biomarker data from Day 0 of the ARMA (Low-Tidal Volume Trial) and ALVEOLI (Assessment of Low Tidal Volume and Elevated End-Expiratory Volume to Obviate Lung Injury) trials. Blood was collected in ethylenediaminetetraacetic acid tubes, plasma was isolated, and aliquots were made and frozen at −56.7°C. Data for IL-6, TNF-R1, ICAM-1, and SP-D were available for 1,262, 539, 1,261, and 1,036 subjects with nonneutropenic ARDS and 32, 17, 32, and 28 subjects with neutropenic ARDS, respectively. Continuous variables are presented as mean (SD) if data were normally distributed and as median (IQR) if data were not normally distributed. ARDS = acute respiratory distress syndrome; ARDSNet = ARDS Network; ICAM-1 = intercellular adhesion molecule 1; IQR = interquartile range; mAPACHE III = Acute Physiology and Chronic Health Evaluation modified to remove the total white blood cell count component; mSOFA = sequential organ failure assessment score modified to remove the platelet component (SOFA_coagulation); SOFA_coagulation = SOFA score component corresponding to platelet count; SP-D = surfactant protein D; TNF-R1 = tumor necrosis factor receptor 1.

ARDS developed in 28.2% of subjects with neutropenia versus 17.3% of subjects without neutropenia. Among subjects with neutropenia and ARDS, the median time of neutropenia (including days before ICU admission) was 9 days (interquartile range, 4–18 d) and the median WBC count at ARDS diagnosis was 0.3 × 10³ cells/μl (interquartile range, 0.1–0.6 × 10³ cells/μl). Three subjects with neutropenia and ARDS had WBC count recovery at ARDS diagnosis. Neutropenia was associated with ARDS development in the adjusted analysis (odds ratio, 2.48; 95% confidence interval [CI], 1.43–4.22; P < 0.001) (Figure 2).

Figure 2.

Outcomes of subjects enrolled in the ICU cohorts (blue) and the ARDSNet trials (gray). (A) Association between neutropenia and development of ARDS and in-hospital mortality among critically ill subjects enrolled in the three ICU cohorts. (B) Association between neutropenia and 60-day mortality among subjects with ARDS enrolled in the six ARDSNet trials. *Among subjects without neutropenia and those with neutropenia, 194 and 24 subjects in the ICU cohorts developed ARDS, respectively. ^†Adjusted for pneumonia and nonpulmonary sepsis. ^‡Restricted to critically ill subjects with ARDS risk factor of pneumonia or nonpulmonary sepsis. ^§Among ICU cohort subjects who develop ARDS, the impact of neutropenia on odds of more severe ARDS (mild vs. moderate or severe, or moderate vs. severe). ^‖Adjusted for age, liver failure, malignancy, and Acute Physiology and Chronic Health Evaluation III minus component for total white blood cell count. ^¶Among 1,294 patients with complete data for all covariates. Hazard ratio for adjusted association between neutropenia and 60-day mortality for 1,294 patients included in this model before the inclusion of IL-6: hazard ratio, 1.78; 95% CI, 1.10–2.88; P = 0.018. ARDS = acute respiratory distress syndrome; ARDSNet = ARDS Network; CI = confidence interval; mSOFA = sequential organ failure assessment score modified to remove the platelet component.

ARDSNet trials

Of the 4,333 subjects with ARDS enrolled in the ARDSNet trials (3–8), 204 (4.7%) had neutropenia. Subjects with neutropenia and ARDS were more likely to have nonpulmonary sepsis (27.5% vs. 21.1%; P = 0.039) than subjects with ARDS without neutropenia (Figure 1). Baseline severity of illness (modified Acute Physiology and Chronic Health Evaluation) was higher (P < 0.001) in neutropenic than in nonneutropenic ARDS.

Compared with subjects with ARDS without neutropenia, subjects with ARDS and neutropenia had higher concentrations of IL-6 (4,632 vs. 253 pg/ml; P < 0.001) and TNF-R1 (tumor necrosis factor receptor 1) (5,461 vs. 3,066 pg/ml; P = 0.014) (Figure 1).

Neutropenia was associated with 60-day mortality in patients with ARDS in the adjusted analysis (hazard ratio, 1.38; 95% CI, 1.10–1.73; P = 0.005) but not in a post hoc adjusted analysis that included IL-6 concentrations (hazard ratio, 1.41; 95% CI, 0.86–2.32; P = 0.171) (Figure 2).

Discussion

Our finding of an association between neutropenia and the development of ARDS in ICU cohorts may highlight the heterogeneity of ARDS pathobiology. Although circulating neutrophils are believed to be essential to ARDS development, the vast majority of our subjects with neutropenia and ARDS (21 of 24) developed ARDS in the absence of circulating neutrophils, suggesting alternative, neutrophil-independent mechanisms. Others have hypothesized that indirect lung injury mediated by endothelial injury may be essential to this paradox (11, 12). Consistent with this, the subjects with neutropenia in our ICU cohorts had more exposure to nonpulmonary ARDS risk factors (nonpulmonary sepsis) (62.1%) compared with subjects without neutropenia (34.6%).

We also saw higher rates of nonpulmonary sepsis as a risk factor for ARDS among subjects with neutropenia enrolled in the ARDSNet trials. This was mirrored in subjects with neutropenia and ARDS having a baseline clinical and biomarker profile that was hyperinflammatory, including higher baseline vasopressor use and proinflammatory mediators (IL-6 and TNF-R1). The etiology of this proinflammatory signature is unknown. Neutropenia was associated with 60-day mortality in the ARDSNet trials and, interestingly, had a similar 60-day mortality rate (43% vs. 44%) as the recently identified hyperinflammatory subphenotype in the same ARDSNet dataset (13).

Our study has strengths and limitations. Our combined datasets are large (N = 5,562) and include patient-level data. Yet, in the ICU cohorts, we lacked data on relevant confounders (transfusion and net fluid balance) for the development of ARDS. Also, the high concurrence between neutropenia and thrombocytopenia precluded an assessment of whether thrombocytopenia independently contributed (e.g., through pulmonary hemorrhage) to ARDS development among subjects with neutropenia. In the ARDSNet trials, limited data on the etiology of neutropenia (23% malignancy) suggest more diverse causes of neutropenia compared with the ICU cohorts (93% malignancy), limiting direct comparison of these two datasets. Similarly, our observation of differential biomarker profiles in neutropenic versus nonneutropenic ARDS is novel but should be tempered by differences in severity of illness and the possibility that inflammation was causative (i.e., neutrophil margination into relevant tissues) of neutropenia in some cases. In both datasets, high hematologic malignancy rates may have influenced the observed mortality. Finally, in neither dataset, like other relevant reports (14), did we have data on the usage of granulocyte colony-stimulating factor, which may improve mortality in ICU patients with neutropenia and cancer.

In conclusion, we showed that neutropenia is associated with the development of ARDS among critically ill patients and with higher mortality among patients with ARDS.