Respiratory and blood samples metagenomic sequencing in diagnosing pulmonary infections in hematologic patients

BACKGROUND: Bacterial and fungal pulmonary infections (BFPI) are common in hematological patients and pose significant diagnostic challenges. Metagenomic next-generation sequencing (mNGS) is valuable for diagnosing BFPI. However, for hematological patients with limited access to lower respiratory tract samples (LRTS), the clinical value of blood-mNGS compared to LRTS-mNGS requires further investigation. METHODS: A retrospective analysis was conducted on 160 cases with suspected pneumonia who underwent both blood-mNGS and LRTS-mNGS within one week. Diagnostic performance and impacts on antimicrobial adjustments were evaluated using clinical composite diagnosis (CCD) as the reference. RESULTS: Compared to CCD, LRTS-mNGS showed significantly higher positive percent agreement (PPA) than blood-mNGS [93.7% (119/127) vs. 34.6% (44/127), P < 0.001], with negative percent agreements (NPA) of 87.5% (21/24) and 91.7% (22/24), respectively. Blood-mNGS showed higher PPA in neutropenic than non-neutropenic patients [56.8% (21/37) vs. 25.6% (23/90), P = 0.001], with unique fungal detection advantages, identifying additional fungi in 7 cases: Mucorales (3), Aspergillus spp. (2), both Mucorales and Aspergillus spp. (1), and Pneumocystis spp. (1). Resistance genes were detected only by LRTS-mNGS. mNGS positively influenced antimicrobial adjustments in 54.3% (69/127) of cases, particularly for pathogens with low empirical coverage, such as Legionella spp. (0.0%, 0/7), Pneumocystis spp. (52.4%, 11/21), and Mucorales (55.6%, 5/9). Blood-mNGS detected these pathogens at rates of 71.4% (5/7), 23.8% (5/21), and 77.8% (7/9), respectively. CONCLUSION: LRTS-mNGS outperformed blood-mNGS in diagnostic performance and resistance detection. Blood-mNGS identified pathogens in one-third of BFPI cases, with value for certain fungal infections, especially in hematologic patients with limited LRTS access. TRIAL REGISTRATION: Not applicable. This study is a retrospective analysis and does not require clinical trial registration.