Severe pneumonia is a common critical illness with an increasing morbidity and mortality, patients with this disease often require admission to intensive care unit (ICU) (Kalil et al., 2016; Metlay et al., 2019; Zhou et al., 2021). Despite significant advances in its etiological investigation and antimicrobial therapy, severe pneumonia remains the leading cause of death among infectious diseases worldwide (De Pascale et al., 2011). There is no denying that early, rapid, and accurate pathogenic diagnosis is crucial in guiding promt antibiotic treatment, thus improving prognosis and reducing fatality (Cillóniz et al., 2021).

Traditionally, clinicians select antibiotics empirically and then adjust treatment based on the results of conventional methods. However, microbiological culture-based tests do not meet clinical needs due to their time-consuming nature, low sensitivity, lack of diagnostic tests for rare pathogens and vulnerability to external influences (Torres et al., 2021). Inspiringly, metagenomic next-generation sequencing (mNGS), a new pathogen detection technique with high efficiency, broad pathogen spectrum and increased sensitivity, has been widely used in clinic gradually (Gu et al., 2019). mNGS theoretically performs unbiased and detailed high-throughput sequencing of the total DNA or RNA content of almost all known pathogens, including bacteria, fungi, viruses, mycobacterium tuberculosis, parasites and atypical pathogens, and then compares the obtained sequence information with databases (Chiu and Miller, 2019). Several case reports and clinical trials have demonstrated the great value of mNGS in the pathogenic diagnosis of severe and complex cases, including rare pathogens, mixed infections (Wang et al., 2019) and infectious diseases presenting with atypical symptoms, such as Chlamydia psittaci (Li et al., 2021), Chlamydia abortus (Zhu et al., 2022), leptospirosis presenting as severe alveolar hemorrhage (Chen M. et al., 2021) and so on. Several studies (Huang et al., 2020; Chen S. et al., 2022; Wu et al., 2022) have revealed that the sensitivity and specificity of mNGS detection are markedly superior than that of conventional methods.

However, the results of studies on the diagnostic and prognositic values of mNGS on severe pneumonia are controversial (Xie et al., 2019; Zhang et al., 2020; Xi et al., 2022). In addition, to our knowledge, there is no relevant systematic review and meta-analysis to provide a higher level of evidence on the clinical values of mNGS results on severe pneumonia. Therefore, we performed this meta-analysis to evaluate and compare mNGS and conventional methods on the diagnostic performance and prognostic impact of severe pneumonia.

We strictly followed the standards of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (http://www.prisma-statement.org/) in reporting the findings of this review. The protocol for this study was registered in PROSPERO (CRD42022325564).

Severe pneumonia is one of the most critical diseases with leading mortality among infectious diseases in ICU (Nair and Niederman, 2021). In the clinic, conventional methods sometimes fail to find the pathogen in patients with severe pneumonia. A reliable study (Chalmers et al., 2011) shows that the fatality rate is as high as 50% in patients who receive only empiric therapy without reliable evidence of pathogenic microorganisms. Prompt targeted antimicrobial therapy treatment is an effective means to improve the survival rate of patients with severe pneumonia.

The detection of DNA or RNA by mNGS provides rapid, efficient, and accurate access to the entire pathogen genome information within the whole test sample, which has been used widely for pathogen diagnosis in clinic in recent years (Miller et al., 2019). However, to our knowledge, there is no relevant systematic review and meta-analysis to provide more reliable evidence on the clinical values of mNGS on severe pneumonia. This study is performed to evaluate the diagnostic value and prognostic impact of mNGS compared with conventional methods in patients with severe pneumonia.

In order to accurately assess the pathogenic diagnostic value of mNGS in severe pneumonia and its impact on prognosis, we comprehensively collected the studies published on the application of mNGS to the etiological diagnostic of severe pneumonia compared with conventional methods, and finally included 23 studies for meta-analysis. The pathogen detection positive rate of the mNGS group was 80.48% (1233/1532), much higher than that of the conventional methods group (45.78%, 705/1540), which indicated that mNGS has a higher diagnostic value than conventional methods and can be used as an effective method for rapid etiological diagnosis of severe pneumonia. Bacteria can be detected by both mNGS and traditional detection methods. mNGS is superior for detecting fungi, viruses, and rare pathogens, which can be used as an adjunct or complementary tool to conventional microbial testing methods to provide an etiologic basis for the accurate anti-infection treatment of severe pneumonia. It is well known that the traditional microbial detection method is susceptible to the influence of normal bacteria in the body, especially after using antibiotics. Patients with severe pneumonia are often initially treated with an anti-microbial therapy, which reduces the positive rate of detection by conventional methods. mNGS can improve the pathogen detection positive rate and is less affected by external influences (Gu et al., 2019). Moreover, some reports have confirmed that mNGS achieve a faster diagnosis of pathogens and can detect unknown pathogens and even drug-resistance genes (Crofts et al., 2017). Another clinical application of mNGS is to identify microbial colonization or infection by monitoring the immune response of patients, ultimately achieving rational application of antibiotics, suppressing bacterial resistance, and reducing the economic and social burden of infectious diseases (Besser et al., 2018).

Meanwhile, compared with the conventional methods group, adjusting the treatment based on the mNGS results significantly decreased 28-day and 90-day mortality and shortened the length of hospital and ICU stay of patients with severe pneumonia. Patients with serious conditions have a short window of time for clinicians to save their lives and precise treatment is crucial to their prognosis. mNGS was faster, taking an average of two days, whereas conventional methods required at least three to five days (Xie et al., 2019). In addition, the higher pathogen-positive rate of mNGS detection gives clinicians the opportunity to select accurate anti-microbial drugs as early as possible, greatly increasing the proportion of target treatments. mNGS lead to more rapid required, accurate diagnosis than conventional methods in severe pneumonia. As a result, mNGS was associated with a better clinical prognosis of severe pneumonia.

mNGS facilitates accurate diagnosis and treatment. Still, there are many challenges, such as the lack of common standards for outcome analysis and common guidelines for report interpretation. The quality and stability of mNGS analysis results may not be stable to some extent, which is closely related to the operator’s skill and the detection efficiency of different laboratories may not be compared. A key challenge inherent to mNGS is that microbial nucleic acids from most patients’ samples are dominated by human host backgrounds (Hasan et al., 2016). Alternative methods exist for the depletion of human background DNA during the preanalytical phase, one approach is to selectively lyse human white blood cells using saponin or other chemical reagents (Marotz et al., 2018), and a different approach is to target low-molecular-weight cell-free DNA or RNA and remove high-molecular-weight genetic content that is often associated with human genomic material (Bukowska-Ośko et al., 2016). Another potential disadvantage of mNGS is the contamination of the sample. We need strictly adhere to quality control procedures for reagents and workflows to maintain as sterile and nucleic acid-free a testing environment as possible. The use of negative controls, reagent evaluation, and regular brushing is needed to ensure that laboratory and sample cross-contamination does not produce false positive results (Schlaberg et al., 2017). In addition, mNGS detection is highly sensitive and requires clinicians with rich experience to comprehensively consider the patient’s condition to make a judgment.

The included studies have limitations, such as small sample sizes, single-centered and retrospective analysis (Pan, 2020; Sun et al., 2020; Zhang et al., 2022), and confounding bias that could not be completely excluded. Besides, all 24 studies included were from China, making the inclusion of a homogeneous population and thus limiting the extensibility of the results. This limits the external generalizability of the results to some degree. Finally, we could not conduct a SORC curve analysis to assess the diagnostic efficacy of mNGS and conventional methods for severe pneumonia due to the scarcity of studies reporting sensitivity and specificity.

The pathogen detection positive rate of mNGS was much higher than that of conventional methods, thereby indicating that mNGS has an extremely good diagnostic performance for severe pneumonia. Besides, adjusting treatment based on mNGS results can reduce the 28-day and 90-day mortality of patients with severe pneumonia, and shorten the length of hospital and ICU stay. Therefore, we suggest that patients with severe pneumonia should be tested for mNGS in addition to traditional culture as early as possible to improve the prognosis and reduce the length of hospital stay.

We describe contributions to the paper as follows: conceptualization - ML, XD and GF. methodology - ML and XD. validation - ML and JY. formal analysis - ML and XD. data curation – ML, ChaZ and JY. writing original draft – ML. writing, review and editing - all listed. visualization - ML, XD and GF. supervision - XD and GF. All authors contributed to the article and approved the submitted version.