Neutrophil percentages in bronchoalveolar lavage fluid: Implications for diagnosing bacterial pneumonia in patients with immunocompromise and neutropenia

Abstract

Rationale: Pneumonia is the most common infection in ICU patients and a leading cause for death. Assessment of bronchoalveolar lavage fluid (BALF) cellularity can aid in pneumonia diagnosis. Low percentages (<50%) of BALF neutrophils have a high negative predictive value for bacterial pneumonia in a general medical ICU population. The operating characteristics in patients with immunocompromise and neutropenia are less clear.

Objective: To compare BALF % neutrophils operating characteristics in patients with and without immunocompromise or neutropenia.

Methods: This was a single center observational cohort study. Patients were categorized into three groups: (1) patients with neutropenia, (2) patients with underlying immunocompromise, and (3) patients with neither. BAL-level analysis reflected neutropenia and immunocompromise state on day of BAL sampling. Operating characteristics of BALF % neutrophils were calculated using varying thresholds of alveolar neutrophilia. Median [Q1,Q3] are reported for nonparametric data and compared using Mann-Whitney U tests.

Results: 688 mechanically ventilated patients had 1736 BAL samples. Among bacterial pneumonia episodes, no difference was found in BALF % neutrophils between patients with underlying immunocompromise and patients with neither neutropenia nor immunocompromise on day of sampling: 82.0% [61.0, 91.0] vs 81.0% [66.0, 91.0], p=0.859. However, when neutropenic on day of sampling, the median BALF % neutrophils was 35.0% [8.8, 67.5] (p<0.001 compared with other categories). In patients with neutropenia, a BALF % neutrophil threshold of 7% had a sensitivity of 90% for excluding bacterial pneumonia.

Conclusions: We found that among patients with bacterial pneumonia, BALF % neutrophil was not significantly lower in patients with a broad spectrum of immunocompromised states but was significantly lower when measured during acute neutropenia. We found varying thresholds of BALF % neutrophils across the three groups. Patients with neutropenia who mount even a low percent of alveolar neutrophils should raise concern for bacterial pneumonia.

Keywords: alveolar neutrophilia; bronchoalveolar lavage fluid analysis; chest infections; host response; immunocompromise; neutropenia; pneumonia.

Figure 1.

(A) Patient course example. An illustrative hypothetical course of a patient highlighting changes in their blood neutrophil counts over the ICU admission. A patient with neutropenia at any point during their ICU admission is categorized as ‘neutropenic’. For BAL level analysis, blood neutrophil count at the time of an episode-defining BAL, which is the BAL collected that the committee marked as the start of pneumonia episode, determined their categorization for analysis. A patient’s BAL data included in analysis could therefore span the different categories based on their blood neutrophil count at time of sampling and start of a new pneumonia episode. Follow-up BALs during a pneumonia episode were not included in the analysis. (B) Overview of study design. Level one: total patients. Level two: total number of patients split by whether they were neutropenic during admission, immunocompromised as defined by research team on study entry, or neither. (C) Overview of BAL level data. Level one: total number of BALs. Level two: total number of BALs obtained by group discerning if sample was collected on a day of neutropenia, collected from patients with underlying immunocompromise, or patients with neither. Level three: number of episode-defining samples by group

Figure 2.

(A) Box plots showing BALF % neutrophils and interquartile range by pneumonia episode etiology. The data is split by patients with neutropenia, patients with underlying immunocompromise, or patients with neither on day of BAL sampling. Data were compared by pairwise Mann-Whitney U tests. (B) Box plots showing peripheral % neutrophils and interquartile range by pneumonia episode etiology on the same day as the BAL samples presented in Figure 1A. The data is split by patients with neutropenia, patients with underlying immunocompromise, or patients with neither based on their peripheral neutrophil count on day of BAL sampling. Data were compared by pairwise Mann-Whitney U tests.

Figure 3.

(A) Sensitivity and specificity plotted at varying BALF % neutrophils thresholds to identify the presence of bacterial pneumonia as outlined by multiplex PCR and culture results in patients with neutropenia (B) Sensitivity and specificity plotted at varying BALF % neutrophils thresholds to identify the presence of bacterial pneumonia as outlined by multiplex PCR and culture results in patients with underlying immunocompromise (C) Sensitivity and specificity plotted at varying BALF % neutrophils thresholds to identify the presence of bacterial pneumonia as outlined by multiplex PCR and culture results in patients with neither