Yield of clinical metagenomics: insights from real-world practice for tissue infections

Background While metagenomic next-generation sequencing (mNGS) has been acknowledged as a valuable diagnostic tool for infections, its clinical validity and impact on patient management when using fresh tissue samples remains uncertain. Methods We conducted a retrospective cross-sectional study involving patients who underwent tissue mNGS at a tertiary hospital in China from February 2021 to February 2024, aiming to assess its ability to detect plausible pathogens and its clinical validity and impact. Findings A total of 520 mNGS results from 508 patients were analysed, detecting plausible pathogens in 302 (58.1%) tests, including 260 single-pathogen and 42 (13.9%) multi-pathogen results. Rare pathogens, such as Balamuthia mandrillaris, Bartonella henselae, and Sporothrix globosa, were identified. Of the multi-pathogen results, 22 were with predominance of anaerobes. mNGS showed higher positivity in cases with high initial suspicion of infection than those used for ruling out infection (PR 1.961, 95% CI: 1.604-2.394) and in patients living with HIV (PR 1.312, 95% CI: 1.047-1.643) or solid organ transplant recipients (PR 1.346, 95% CI: 1.103-1.643) compared to immunocompetent individuals. Sensitivity and specificity for diagnosing confirmed/ probable infections were 85.0% (95% CI: 76.7%-93.3%) and 93.7% (95% CI: 86.8%-100.0%), respectively. mNGS influenced clinical management in 258 (49.6%) cases by identifying new infections and in 112 (21.5%) by excluding infections. It prompted initiation (20.2%), modification (23.1%), or discontinuation (6.3%) of antimicrobial therapy. Interpretation mNGS demonstrates high diagnostic accuracy for tissue infections. Its impact on clinical management highlights the need to integrate it into current diagnostic practices. Funding National Natural Science Foundation of China (No. 82472371), " Leading Geese" Research and Development Plan of Zhejiang Province (No. 2024C03218), and Pudong New Area Joint Project (PW2021D-09). Copyright (c) 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).