Association between normalized Mycoplasma pneumoniae sequence reads and severity of pediatric pneumonia: a...

Mycoplasma pneumoniae (MP) is a leading cause of pediatric community-acquired pneumonia (CAP), with an increasing proportion of severe cases in the post-coronavirus disease 2019 (COVID-19) era. High-throughput targeted next-generation sequencing (tNGS) enables semi-quantitative detection of MP; however, the clinical significance of normalized MP sequence reads derived from oropharyngeal swabs remains unclear. This study aimed to explore whether normalized MP sequence reads from oropharyngeal swabs can serve as a biomarker for early assessment of disease severity in pediatric MP pneumonia (MPP).

In this cross-sectional study, 262 consecutive children with MPP were enrolled from January to December 2023 at Fujian Children's Hospital. Clinical and laboratory data were retrospectively reviewed. Normalized MP sequence reads were derived from pre-antibiotic oropharyngeal swabs and processed using a standardized fastp/Bowtie2 bioinformatics pipeline. Severe MPP (SMPP) was defined according to the 2019 Chinese Guidelines for the Management of CAP in Children. Associations between MP reads and disease severity were assessed using multivariable logistic regression, adjusting for age, sex, fever duration, coinfection, and A2063G mutation status. Model discrimination, calibration, and incremental value were evaluated using area under the curve (AUC), DeLong test, Hosmer-Lemeshow test, bootstrap validation, integrated discrimination improvement (IDI), continuous net reclassification improvement (NRI), and decision curve analysis (DCA).

Among 262 children (median age 6.3 years; 55.3% male), 131 were classified as SMPP and 131 as general MPP (GMPP). Compared with GMPP, SMPP patients had longer hospital stays and fever duration, higher rates of oxygen therapy, glucocorticoids, and second-line antibiotics (all P<0.05), as well as elevated inflammatory markers [C-reactive protein (CRP), D-dimer (DDi), ferritin; all P<0.05]. Normalized MP sequence reads were significantly higher in children with SMPP. After multivariable adjustment, children in the highest quartile (Q4) had a 2.69-fold increased risk of SMPP compared with those in the lowest quartile (Q1). As a standalone predictor, MP reads demonstrated modest discrimination (AUC =0.592). The clinical model achieved an AUC of 0.710, which increased to 0.723 after incorporation of MP reads (DeLong, P=0.40). Despite the modest improvement in AUC, reclassification metrics showed significant improvement (IDI =0.015, P=0.01; continuous NRI did not reach statistical significance), and DCA demonstrated greater net clinical benefit for the full model across a broad range of threshold probabilities. The full model showed good calibration and stability in bootstrap validation.

Normalized MP sequence reads are independently associated with disease severity and systemic inflammatory burden in pediatric MPP. Although their standalone discriminatory capacity is limited, incorporation into a clinical model provides meaningful incremental predictive value and potential utility for early SMPP risk stratification.