Severe pneumonia stands as a pressing health concern, marked by heightened short and long-term mortality rates (Bein et al., 2018; Sangla et al., 2020). This illness can manifest as community-acquired pneumonia (CAP), hospital-acquired pneumonia (HAP), or ventilator-associated pneumonia (VAP). Identifying the etiology of severe pneumonia can be particularly challenging in patients with prior antibiotic exposure or mechanical ventilation. Timely and accurate administration of antimicrobial therapy plays a pivotal role in optimizing outcomes among critically ill patients (Kalil et al., 2016; Torres et al., 2017; Metlay et al., 2019).

Veno-venous extracorporeal membrane oxygenation (vv-ECMO) has emerged as a life-support technique gaining traction in critically ill patients grappling with reversible refractory respiratory failure. The timely and precise implementation of ECMO has been proven to be a significant enhancement in survival rates and overall prognosis (Oh et al., 2021; Kochanek et al., 2022). However, when considering invasive procedures like continuous kidney replacement therapy (CRRT), deep vein puncture, tracheal intubation, and even ECMO intubation, the potential for infection should not be underestimated, these procedures can increase patients’ vulnerability to infections(Kühn et al., 2020). The decision to initiate vv-ECMO is primarily made in patients who have already undergone antibiotic therapy and mechanical ventilation, and the occurrence of HAP and VAP can further complicated antimicrobial therapy. Additionally, nosocomial infections occurring in vv-ECMO patients are associated with elevated mortality risk (Grasselli et al., 2017).

Presently, various methods for pneumonia etiological detection, encompassing morphological examination, isolation culture, and immunological techniques, exhibit distinct merits (Mazloomirad et al., 2021). However, conventional approaches have limitations concerning the time required for detection, positive identification rates, and specificity, particularly in severe infection scenarios. Metagenomic Next Generation Sequencing (mNGS), also known as high-throughput sequencing, presents several advantages, including shorter turnaround times and enhanced accuracy in pathogen identification (Lu et al., 2020). This technique can be applied to an array of specimen types, including bronchoalveolar lavage fluid (BALF), sputum, blood, pleural effusion, and tissue, and requires only a modest quantity of DNA/RNA extraction for efficient pathogen detection and identification (Han et al., 2019; Miller and Chiu, 2021). Moreover, mNGS allows for quantitative or semi-quantitative assessment of organism concentration in a sample through sequencing reads, which proves particularly advantageous in detecting multiple microbial strains and rare pathogens (Salipante et al., 2014). Importantly, mNGS demonstrates relatively lower susceptibility to the effects of prior antibiotic exposure compared to conventional methods (Chiu and Miller, 2019; Huang et al., 2020; Tang et al., 2021). Furthermore, the treatment adjustments guided by mNGS has shown rapid symptom alleviation, improved lung imaging, reduced hospitalization durations, and mitigation of antibiotic-related adverse reactions (Wang et al., 2020; Sun et al., 2021; Zhou et al., 2021).

In this study, we conducted a comprehensive review of non-immunocompromised patients with severe pneumonia who underwent vv-ECMO at the First Affiliated Hospital of Zhengzhou University. Our investigation assessed the value of mNGS technology in both identifying the etiology of pneumonia and guiding clinical strategy.

Severe pneumonia presents a critical challenge with high mortality rates, primarily attributed to the limitations of current diagnostic techniques in pathogen identification and the escalating issue of antibiotic resistance (Costelloe et al., 2010; Currie et al., 2014; Jain et al., 2015). Consequently, timely and precise pathogen identification has emerged as a paramount necessity. In our retrospective study, we undertook an assessment of mNGS’s efficacy in pathogen detection and personalized antibiotic therapy for non-immunocompromised severe pneumonia patients receiving vv-ECMO support. Our findings substantiated that mNGS boasted a superior pathogen detection rate (85.9%) compared to traditional culture methods, identifying a broader array of pathogenic types (65 strains vs. 23 strains) – a remarkable achievement even in patients who had undergone prolonged antibiotic treatment before mNGS sample collection. Importantly, our study unveiled that mNGS results could contribute crucial insights for clinical decision-making regarding antibiotic therapy.

Non-immunocompromised severe pneumonia patients, despite initially normal or near-normal immune status, could develop acquired infections due to associated conditions and ECMO-induced immune dysfunction. This phenomenon underscored the significance of acquired infections during ECMO support, affecting up to 55% of patients (Al-Fares et al., 2019; Frerou et al., 2021). The pivotal role of infections in the prognosis of ECMO patients has been highlighted in earlier research (Biffi et al., 2017; Grasselli et al., 2017). While traditional diagnostic methods such as G/GM testing, bacterial culture, and quantitative polymerase chain reaction (qPCR) are commonly used, they often possess inherent limitations (Haneke et al., 2016) and can struggle to accurately detect mixed pathogens (Crotty et al., 2015). This context calls for a revolutionary diagnostic approach to enable timely and precise pathogen detection, crucial for averting uncontrolled infections that significantly influence the prognosis of critically ill patients undergoing ECMO treatment.

In addition to its heightened sensitivity, rapid detection cycles, and versatile applications, mNGS technology circumvents the need for bacterial isolation and remains unaffected by prior antibiotic usage, effectively reducing the false-negative rate (Brown et al., 2018; Carpenter et al., 2019). Recent studies even demonstrate mNGS’s capability to detect potential pathogens in samples that routine tests classify as negative, thereby enhancing diagnostic accuracy for infections (Zinter et al., 2019; Chen et al., 2021). Theoretically, mNGS has the potential to detect a comprehensive range of nucleotide sequences and pathogens, expanding the scope of infection diagnostics. Consequently, mNGS is particularly advantageous in detecting mixed pathogen infections (Wang et al., 2019). Given that patients treated with vv-ECMO have often undergone prior antibiotic treatment and mechanical ventilation, mNGS holds special value in this context. Our findings in this study echoed previous research (Cheng et al., 2019), indicating that the positive rate of traditional culture (64.9%) was lower than that of mNGS detection (85.9%). Notably, our study reinforced the robustness of mNGS in detecting a wide variety of viruses, as well as specific pathogens like Mycoplasma pneumoniae, Mycobacterium spp., and Neisseria neocaledoniensis, reaffirming its ability to identify special pathogen infections and mixed infections.

Among the organisms frequently detected through mNGS, Klebsiella pneumoniae, Cytomegalovirus, and Acinetobacter baumannii emerged as predominant pathogens in non-immunocompromised severe pneumonia patients supported by vv-ECMO. Interestingly, various Gram-positive bacteria, fungi, viruses, and specific pathogens were exclusively detected in Group D, while Acinetobacter baumannii, Pseudomonas aeruginosa, and Klebsiella pneumoniae exhibited higher prevalence in Group D compared to Group I. These organisms are commonly associated with Hospital-Acquired Pneumonia (HAP) and Ventilator-Associated Pneumonia (VAP). Research suggests that initial viral or bacterial infections can trigger a dysregulation of cytokine and chemokine production, setting off an overactivated signaling cascade that leads to excessive immune cell recruitment and inflammatory processes. Consequently, this cascade may inflict severe lung damage, providing a gateway for pathogens to penetrate deep tissues and perpetuate viral and bacterial replication (Wilden et al., 2021). Furthermore, compromised immune responses in later stages may be attributed to pathogen-mediated inhibition of immune cells or signaling pathways, resulting in diminished pathogen clearance. These mechanisms provide insights into the increased morbidity and mortality associated with viral and bacterial co-infections (Aguilera and Lenz, 2020; Wilden et al., 2021).

The management of pneumonia remains complex due to shifting epidemiological patterns. Owing to the lack of point-of-care diagnostic tests that can accurately pinpoint microbial etiology (excluding viral pneumonia), initial antibiotic treatment often relies on empirical methods. While empirical antibiotic treatment plays a crucial role in controlling disease progression, inherent limitations persist, potentially leading to inadequate microbial antibiotic coverage and inappropriate use of broad-spectrum antibiotics. Such limitations can impact prognosis, drive drug resistance, heighten drug toxicity, and escalate treatment-related expenses (Maertens et al., 2005; Polgreen et al., 2007; Tumbarello et al., 2015). Additionally, a considerable proportion of patients (ranging from 38.0% to 60.8%) require adjustments to their antibiotic regimen due to inappropriate empirical antibiotic therapy (Alvarez-Lerma, 1996; Kollef and Ward, 1998). Amid the realm of empirical antibiotic use, early incorporation of mNGS holds the promise of targeted therapy initiation, thus enhancing clinical antibiotic management to counteract resistance. Moreover, by reducing hospital stay durations, ICU admission lengths, ventilation periods, and minimizing the occurrence of antibiotic-related complications while significantly curtailing ICU costs, mNGS can have a favorable impact on clinical outcomes (Wang et al., 2020; Zhang et al., 2020; Shen et al., 2021; Sun et al., 2021; Xie et al., 2021). In our study, 26 out of 50 patients underwent antibiotic adjustments based on mNGS results, even though there was no statistically significant difference in clinical outcomes. This could be attributed to the higher number of pathogenic microorganisms detected in Group D, where the number of antibiotic adjustment cases was higher. Moreover, mNGS testing occurred after ECMO initiation, relatively late in detection timing, possibly having less influence on patients’ conditions. Despite these limitations, the guiding role of mNGS in antibiotic treatment remains noteworthy. Further prospective studies are needed to explore whether mNGS can guide the optimization of antibiotic therapy.

While mNGS detection presents numerous benefits, traditional culture remains indispensable. In our single-center study, 34.8% of pathogens identified through traditional culture went undetected by mNGS. Therefore, a combination of traditional culture and mNGS detection for pathogen identification and drug sensitivity testing offers a more comprehensive and effective approach for guiding clinical treatment in real-world scenarios.

This study is not without limitations. The retrospective nature of our study might constrain the generalizability of the outcomes, emphasizing the necessity for broader cohort studies to validate our findings. Furthermore, not all samples underwent RNA and DNA virus detection, and the influence of mechanical ventilation on pathogen results cannot be ignored. Finally, ECMO, being an invasive procedure, could potentially affect post-ECMO treatment outcomes.

In conclusion, our results indicated that mNGS has a higher positive rate than the conventional culture and can be used to effectively identify mixed pathogens in non-immunocompromised patients with severe pneumonia supported by vv-ECMO, especially in those patients with negative cultures due to infections with special pathogens or prior antibiotic exposure. However, it is important to note that the interpretation of mNGS data should be complemented by clinical data and conventional culture results due to the lack of uniform diagnostic criteria.

The data that support the findings of this study have been deposited into the EMBL database with accession number PRJEB65300.

The studies involving humans were approved by the ethics committee of the First Affiliated Hospital of Zhengzhou University. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin because this study is retrospective and no additional intervention should be performed on the patients.

X-XL: Data curation, Formal Analysis, Software, Writing – original draft. C-ZN: Data curation, Formal Analysis, Software, Writing – original draft. Y-CZ: Conceptualization, Methodology, Supervision, Writing – review & editing. G-WF: Methodology, Validation, Writing – review & editing. HZ: Methodology, Supervision, Writing – review & editing. M-JH: Resources, Writing – review & editing. JL: Resources, Supervision, Writing – review & editing.