Enhanced DNA and RNA pathogen detection via metagenomic sequencing in patients with pneumonia

Abstract

Background: Metagenomic next-generation sequencing (mNGS) is an important supplement to conventional tests for pathogen detections of pneumonia. However, mNGS pipelines were limited by irregularities, high proportion of host nucleic acids, and lack of RNA virus detection. Thus, a regulated pipeline based on mNGS for DNA and RNA pathogen detection of pneumonia is essential.

Methods: We performed a retrospective study of 151 patients with pneumonia. Three conventional tests, culture, loop-mediated isothermal amplification (LAMP) and viral quantitative real-time polymerase chain reaction (qPCR) were conducted according to clinical needs, and all samples were detected using our optimized pipeline based on the mNGS (DNA and RNA) method. The performances of mNGS and three other tests were compared. Human DNA depletion was achieved respectively by MolYsis kit and pre-treatment using saponin and Turbo DNase. Three RNA library preparation methods were used to compare the detection performance of RNA viruses.

Results: An optimized mNGS workflow was built, which had only 1-working-day turnaround time. The proportion of host DNA in the pre-treated samples decreased from 99 to 90% and microbiome reads achieved an approximately 20-fold enrichment compared with those without host removal. Meanwhile, saponin and Turbo DNase pre-treatment exhibited an advantage for DNA virus detection compared with MolYsis. Besides, our in-house RNA library preparation procedure showed a more robust RNA virus detection ability. Combining three conventional methods, 76 (76/151, 50.3%) cases had no clear causative pathogen, but 24 probable pathogens were successfully detected in 31 (31/76 = 40.8%) unclear cases using mNGS. The agreement of the mNGS with the culture, LAMP, and viral qPCR was 60%, 82%, and 80%, respectively. Compared with all conventional tests, mNGS had a sensitivity of 70.4%, a specificity of 72.7%, and an overall agreement of 71.5%.

Conclusions: A complete and effective mNGS workflow was built to provide timely DNA and RNA pathogen detection for pneumonia, which could effectively remove the host sequence, had a higher microbial detection rate and a broader spectrum of pathogens (especially for viruses and some pathogens that are difficult to culture). Despite the advantages, there are many challenges in the clinical application of mNGS, and the mNGS report should be interpreted with caution.

Keywords: Early pathogen detection; Metagenomic next-generation sequencing; Pneumonia.

Fig. 1

Complete mNGS assay workflow. A complete workflow for simultaneous DNA and RNA pathogen detection in different kinds of samples for pneumonia based on mNGS on one working day was developed. The pipeline includes sample processing, library preparation, sequencing, data processing, threshold criteria for pathogen detection and final results reporting

Fig. 2

mNGS assay optimization on host DNA depletion. a The ratio of unique reads mapped to the human genome before and after human DNA depletion (mean with SD). b Relative enrichment of sequencing reads mapped to microorganisms by the host DNA depletion approach. c Pathogen detection in 29 samples without host DNA depletion (below) and after host DNA depletion (upper), shown by species RPM normalized by min–max normalization. d Relative enrichment of pathogen species reads before and after human DNA depletion in positive BALF specimens. Viruses (n = 3) are EBVs; G+ bacteria (n = 11) include S. pneumoniae and Tropheryma whipplei; G− bacteria (n = 7) include Pseudomonas aeruginosa, Klebsiella pneumoniae and Haemophilus influenza; fungi (n = 7) include Aspergillus fumigatus, Candida albicans and Candida tropicalis; and Chlamydia (n = 1) is Chlamydia psittaci. e Comparison of pathogen detection with two host DNA depletion methods. Three different BALF specimens spiked with HSV1, VZV, EBV, S. pneumoniae and Aspergillus Niger were undergo host DNA depletion with the saponin method and the MolYsis kit. After sequencing with 15 M for each library, species reads were calculated respectively

Fig. 3

Comparison of culture, LAMP, and mNGS identification in terms of pathogen detection spectrum. Pathogen classification categories of culture and mNGS identification were displayed in a, b. Pathogen species and the corresponding number of cases identified by culture, LAMP and mNGS identification are shown in c–e, respectively. Different colours indicate different pathogen categories

Fig. 4

Concordance between metagenomic next-generation sequencing (mNGS) and conventional tests. a Pie chart demonstrating the positivity distribution for the detection of pathogens by mNGS and conventional testing in 151 cases. b Positive and negative agreement of mNGS versus culture, LAMP assay, qPCR and all conventional tests