B63-19 Determinants of BAL PCR Performance in Children: Pathogen Frequency and Neutrophilia

Background

Quantitative multiplex PCR offers rapid pathogen identification, though clinical utility is limited by detection of non-viable organisms or residual nucleic acids. This study aimed to evaluate PCR performance against quantitative BAL culture and to determine how pathogen frequency and neutrophilic inflammation influence diagnostic fidelity.

Methods

We analyzed 6,000 pathogen-specific observations from 400 pediatric bronchoalveolar lavage (BAL) samples at a single quaternary-care urban children’s hospital. PCR performance was evaluated against gold standard quantitative culture ≥ 104 CFU/mL. Pathogens were stratified into two statistically distinct groups based on their pooled baseline prevalence (PBP) (p < 0.001): high-frequency (HF) pathogens (n = 1,600; M. catarrhalis, S. aureus, H. influenzae, S. pneumoniae, PBP 14%) and low-frequency (LF) pathogens (n = 4,400; P. aeruginosa, E. coli, S. pyogenes, and 8 others, PBP 2%). PCR performance was evaluated across thresholds 104-107 copies/mL. Sub-analyses targeted normal neutrophils (≤10%, n = 1,020) and severe neutrophilia (≥43%, n = 3,525).

Results

For PCR-culture concordance analyses, complete paired data were available for 5,730 pathogen-sample combinations. PCR at 104 copies/mL threshold yielded the highest Youden index (J = 0.91), with sensitivity 97%, specificity 94%, PPV 48%, and NPV 100%. For HF pathogens, PPV was 50% at 104 copies/mL and 62% at 105 copies/mL and rose to 71% at 106 copies/mL (p < 0.001 vs. 104 copies/mL). For LF pathogens, PPV was 81% at 105 copies/mL with a normal neutrophil count, NPV was 99% at all PCR thresholds. Severe neutrophilia was associated with an increase in PCR and culture positivity but improvement of PPV was not significant (75% at 107 copies/mL).

Conclusion

Negative PCR rules out infection with 100% NPV in a normal neutrophil count. PCR performs with excellent sensitivity and specificity at 104 copies/mL PCR threshold; however, PPV is variable based on pathogen frequency. For HF pathogens, extrapolation from PCR data is limited by detection of non-viable organisms at low PCR threshold of 104 copies/mL and improves at a higher threshold 106 copies/mL. High bacterial load associated with severe neutrophilia does not significantly improve confirmation of infection by PCR because a positive PCR reflects the total microbial burden rather than selecting for infection-causing organisms alone. For For LF pathogens, PPV was 81% at 10⁵ copies/mL. In samples with a normal neutrophil count, NPV was 99% at all PCR thresholds.

This abstract is funded by: None