Improving Microbiological Diagnostic Yield of Bronchoalveolar Lavage (BAL) Specimens in Pediatric Allogeneic Hematopoietic Stem Cell Transplant (AlloHSCT) Recipients

Abstract

Background Pulmonary complications, including both infectious and noninfectious etiologies, are a frequent cause of morbidity and mortality in pediatric alloHSCT patients from the immediate post-transplant period to 5 years post-transplant. Early detection and initiation of appropriate antimicrobial therapy improve survival. However, present techniques for identifying infectious organisms are limited by pretreatment with antimicrobial therapies and the need for a priori knowledge of likely viral pathogens, greatly decreasing the yield of BAL. Our group has published the results of 193 pediatric HSCT patients and found a 43% probability of need for BAL in the year following stem cell transplant with only a 40% yield of identifying an infectious organism. Massively parallel sequencing (MPS) has the potential to mitigate these limitations, by identifying the genetic sequence of all DNA or RNA in a sample without the need for viable organisms or microbe specificity. Methods Pediatric alloHSCT patients with new pulmonary findings on imaging underwent BAL by the pediatric pulmonary service at MFCH as a pilot study (NYMC-181). BAL was performed per standard WMC clinical practice methods and lavage fluid sent for bacterial, fungal, and viral cultures, aspergillus galactomannan, βD-glucan, and PCRs for CMV, EBV, and adenovirus, and other tests as indicated. Excess BAL fluid was preserved in Trizol prior to RNA isolation. cDNA libraries were generated from each sample and sequenced on the Illumina NextSeq platform. Sequence reads were de novo assembled using the Trinity assembler and classified using Kraken and a custom-built database. Results from MPS and clinical testing were compared. Results 9 patients underwent 10 BALs (1 patient underwent 2 BALs) from July 2016 through December 2017 and had sufficient BAL fluid for inclusion in this study (Figure 1). Clinical microbiological testing identified infections in 6 patients (60%), 3 with bacterial infections (2 Staphylococcus haemolyticus & 1 Streptococcus viridans), 3 with viral pathogens (1 CMV, 1 EBV, and 1 RSV), and 1 fungal infection (Candida). MPS identified infections in 5 patients, one fewer than the clinical testing. Results were wholly concordant in 2 patients (Staph haemolyticus and RSV) and partly concordant in 1 patient (Candida detected, but different viruses found on clinical and sequencing testing). CMV, EBV, S. haemolyticus, and S. viridans (1 patient each) were detected on clinical testing, but not using sequencing. Haemophilus influenza, rhinovirus, and enterovirus (1 patient each) were detected using sequencing, but not clinical testing. Conclusions RNA based massively parallel sequencing techniques can serve as an adjunct to standard clinical testing of BAL fluid for improved detection of infections (especially polymicrobial) in pediatric recipients of alloHSCT.