Utility of Metagenomic Next-Generation Sequencing for Etiological Diagnosis of Patients with Sepsis in Intensive Care Units

ABSTRACT The performance of metagenomic next-generation sequencing (mNGS) was evaluated and compared with that of conventional culture testing in patients with sepsis. Prospective blood and bronchoalveolar lavage fluid (BALF) samples from 50 patients with sepsis were tested using cultures (bacterial, fungal, and viral) and mNGS of microbial DNA (blood and BALF) and RNA (BALF). mNGS had higher detection rates than blood culture (88.0% versus 26.0%, P < 0.001) and BALF culture (92.0% versus 76.0%, P = 0.054). RNA-based mNGS has increased the detection rate of several bacteria, fungi, and viruses, but not mycobacteria and Toxoplasma gondii. The number of multiple detections per specimen was higher in BALF (92.0%) than in blood (78.0%) samples, and the highest number of pathogens detected in a single specimen was 32. Among blood samples, compared to cultures, mNGS detected significantly more bacteria (P < 0.001), fungi (P = 0.012), and viruses (P < 0.001), whereas BALF mNGS had a higher detection rate for bacteria (P < 0.001) and viruses (P < 0.001). The percentage of mNGS-positive samples was significantly higher than that of culture-positive samples for several Gram-negative bacteria, some Gram-positive bacteria, and viruses, but not fungi. Mycobacteria had a higher detection rate by culture than by mNGS, but the difference was not significant due to the small sample size. The positive and negative agreements with 95% confidence intervals of mNGS and culture were 62.0% (50.4 to 72.7) and 96.8% (96.5 to 97.1), respectively. mNGS offers a sensitive diagnostic method for patients with sepsis and is promising for the detection of multipathogen infections. Clinical correlation is advised to interpret mNGS data due to the lack of unified diagnostic criteria. IMPORTANCE Delays in effective antimicrobial therapy have resulted in decreased survival rates among patients with sepsis. However, current culture-based diagnostic methods have low sensitivity because of concurrent antibiotic exposure and fastidious and atypical causative organisms. Among patients with sepsis, we showed that mNGS methods had higher positive rates than culture methods, especially for bacteria, viruses, and multipathogen infections, which are difficult to culture and detect in patients treated with antibiotics. RNA-based mNGS has increased the detection rate of several bacteria, fungi, and viruses, but not mycobacteria and Toxoplasma gondii. mNGS also showed a high negative percent agreement with cultures. However, the interpretation of mNGS data should be combined with clinical data and conventional methods considering the lack of unified diagnostic criteria.