Clinical applications and challenges of metagenomic next-generation sequencing in the diagnosis of pediatric infectious disease

Introduction

Infectious diseases pose a certain threat to the life and health of children. According to studies, 6.3 million children under 5 years of age died in 2013, of which 51.8 % (3,257,000) died of infectious diseases. Pneumonia, diarrhea, and measles were responsible for half of these deaths [1]. In China, infectious diseases were the leading cause of death in children under 5 years of age from 2010 to 2016, with pneumonia and sepsis accounting for up to 19 % of the causes of child mortality in 2016 [2]. Among newborns, preterm and very low birth weight infants are highly susceptible to infections because of their low immune function, resulting in high neonatal mortality [3].

Identification of early pathogenic microorganisms is particularly important for the diagnosis and treatment of infectious diseases, and studies have shown that mortality increases significantly with the increase in the duration of treatment administration [4]. The current traditional culture methods can easily be affected by antibiotics which result in long culture cycles. Moreover, many pathogenic microorganisms in the environment and humans are difficult to culture. Additionally, PCR detection techniques are limited to amplifying the gene sequences of known pathogenic microorganisms and cannot detect and diagnose new and rare pathogens. The detection of pathogens, whether based on bacterial culture or targeted molecular methods, relies on the clinical judgment of symptoms of infectious diseases and the empirical assessment of suspected pathogens in a clinical context. Therefore, there is a clinical need for more sensitive and accurate experimental techniques to complement traditional methods and to improve the detection rate of pathogenic pathogens for early and accurate drug administration.

Metagenomic next-generation sequencing (mNGS) is an unbiased, culture-free, high-throughput method that directly extracts deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) from clinical samples for high-throughput sequencing, and after comparison with pathogenic microbial databases and raw letter analysis, all pathogens such as bacteria, viruses, fungi, and parasites can be detected in clinical samples at one time, theoretically. It is especially suitable for the diagnosis of infectious diseases caused by complex, rare, atypical, and emerging pathogens [5] and has unique technical advantages and wide application prospects. Table 1 summarizes and compares the advantages and disadvantages of different laboratory diagnostic techniques. This review focuses on the clinical application of mNGS in different pediatric infectious diseases, emphasizing its indispensable role and value in the diagnosis and treatment of pediatric infectious diseases. Furthermore, it analyzes the current limitations of mNGS in clinical applications to provide guidelines for pediatricians to use this technology in the clinic.

Application of mNGS in pediatric respiratory infections

The etiology of respiratory infections in children is complex and diverse, encompassing viruses, bacteria, atypical pathogens, fungi, etc. [6]. In recent years, emerging and re-emerging respiratory pathogens such as novel coronaviruses and avian influenza viruses have posed serious threats to the health and lives of children. Accurate identification of the infecting pathogen is instrumental for the rational use of antimicrobial agents and for the early detection, isolation, reporting, and treatment of respiratory infectious diseases. Therefore, rapid and accurate etiological diagnosis is of significant importance for the clinical management and control of common respiratory infectious diseases.

Application of mNGS in pediatric central nervous system infections

Central nervous system infections (CNSIs) are caused by multiple pathogens that result in central nervous system damage. In pediatric populations, due to the differences among various pathogens, central nervous system infections can present a range of symptoms, making pathogen identification crucial for guiding clinical treatment and characterizing prognosis. Currently, over 40 % of encephalitis patients remain undiagnosed for specific pathogens following routine clinical testing [16]. In bacterial meningitis, pre-treatment with antibiotics often leads to negative culture results. Although PCR testing is fast and highly sensitive, it depends on clinical presumptions. mNGS enables extensive screening of pathogens from a single sample, reducing the duration of the diagnostic process.

Application of mNGS in pediatric suspected bloodstream infection

Bloodstream infection (BSI) is a process in which pathogens enter the human bloodstream and multiply to release a variety of metabolites and toxins, resulting in symptoms such as infectious toxicity and severe inflammatory metaplasia. In critically ill patients, BSI is a life-threatening complication. In children, especially neonatal patients, because of the imperfect development of the autoimmune function, it can rapidly spread to multiple organs once the disease develops, and the clinical mortality rate is high [33]. Therefore, early and rapid diagnosis and selection of sensitive antimicrobial drugs are crucial to reducing the mortality rate associated with BSIs in children. Blood culture is the gold standard method for the diagnosis of BSI in clinical laboratories; however, the process of blood culture takes a long time, and some slow-growing bacteria require a culture time of >7 days, which may delay the early treatment of children. The use of mNGS significantly shortens the treatment time because it does not require bacterial culture, reduces the volume of blood drawn from children by using plasma-free DNA as the specimen, and compensates for the disadvantage that some caustic bacteria cannot be cultured owing to their rigorous culture conditions; thus, mNGS provides a new treatment method for children with clinical BSIs.

Application of mNGS in other infectious diseases

Challenges in the application of mNGS to clinical pediatric infectious diseases

The application of mNGS has brought new ideas to the diagnosis and treatment of infectious diseases, especially in the treatment of children with unexplained infections and severe multiple infections. However, many challenges remain in the clinical application of mNGS owing to its complex operation procedure and numerous variables.

First of all, the challenge is to resolve the high host background to improve the sensitivity of mNGS. By consulting the literature, we found that the specimens applied to childhood infectious diseases are the same as those applied to adult infections, most of which are human samples such as blood, cerebrospinal fluid, nasopharyngeal swabs, and alveolar lavage fluid. The content of microorganisms and genomes in these samples is extremely low compared with that of the host cells, and the host genetic information is often >90 % by sequencing [46]. It remains debated whether it is necessary to perform de-hosting in experiments and whether the removal of human background or enrichment of microorganisms can increase the percentage of pathogens in the whole genome and thus improve the detection rate [47, 48]. However, it is possible that the pathogens of interest are removed at the same time as de-hosting, and the nucleic acid concentration of the samples is reduced, which may not meet the requirements for library construction.

The second challenge is to overcome the interference of reagents and background laboratory bacteria. Some unrelated microorganisms are widely present in clinical samples and reagents. Wilson et al. investigated the use of hybridization probes to remove redundant sequences to detect pathogenic sequences preferentially. In this approach, only 20 pg of purified RNA was added to the cerebrospinal fluid sample to suppress most of the non-human reads from water and reagents, thus reducing the number of reads from non-target microorganisms [19]. The nucleic acid extraction process is an important factor contributing to the variability of mNGS results. Currently, commercial pathogen nucleic acid extraction kits mainly employ two methods: magnetic bead-based and column-based extraction. Typically, during the development process, optimization is focused on extracting nucleic acids from specific types or categories of pathogens. Therefore, it is crucial to validate the selected pathogen nucleic acid extraction kits, including verifying their efficiency in extracting nucleic acids from Gram-positive bacteria, Gram-negative bacteria, fungi, viruses, parasites, and other pathogens, in order to avoid any potential pathogen detection failures. Meanwhile, before handling clinical specimens, it is necessary to conduct performance verification for the entire experimental process, including verifying the detection limit, inclusivity, anti-interference ability, precision, carryover and cross-contamination, stability, etc., for the “representative” species commonly associated with infectious diseases. Due to the complex, multi-step nature of mNGS testing, multiple quality control indicators are necessary, including sample controls, external controls, internal controls, library quality metrics, sequencing quality metrics, and contamination controls. As laboratory-developed tests, laboratories can adopt acceptable standards to define each parameter, ensuring sample quality and operational quality, and preventing errors during sample processing.

In addition, the establishment of laboratory standards for detecting pathogenic bacteria, laboratory-specific background pathogen libraries, and interpretation of reports have yet to be addressed and standardized. A survey at the Chinese Academy of Health and Food Inspection showed significant differences in detection capacity among laboratories, and differences in detection performance were mainly reflected in the ability to detect false positives, with large differences in the detection capacity for fungi and RNA viruses [49]. Therefore, each laboratory should confirm the performance of this experimental technology before clinical specimen testing and determine the threshold value for each pathogenic bacterium. In addition, the interpretation of mNGS results is relatively complex, with large sequence data, complex parameters, and high requirements for confidence analysts and report review and clinical interpretation personnel. Multiple factors influence the threshold for pathogen diagnosis in mNGS, including sequencing platform, sequencing process, specimen type, pathogen species, and patient condition. Currently, there is no universally recognized threshold, so it is recommended that each laboratory establish its threshold and validate it in clinical practice. Zhang et al. conducted a performance confirmation of the pathogenic macro genomic project before hospital localization from four aspects of clinical intended use, methodology establishment, performance confirmation, and standard operation workbook, which were proposed as specific implementation plans for clinical laboratories [50]. Currently, with the development of macro genomic mNGS, expert consensus on mNGS in clinical applications has been successively released in China [51], [52], [53], [54], [55], [56], [57] for clinical medical personnel and laboratory workers to learn from.

The impact of different read lengths of different pathogenic microorganisms on result interpretation is as follows: under the same conditions (same microorganism, same specimen), a higher number of reads detected for a particular microorganism suggests a higher possibility of it being a pathogenic microorganism. However, different pathogenic microorganisms have varying genome sizes, and there are differences in nucleic acid extraction efficiency. Moreover, the pathogenicity of different microorganisms also varies significantly [53]. Therefore, it is not sufficient to rely solely on the number of reads to determine infection status. It is recommended to consider both the differences in pathogenic microorganism species and their pathogenic characteristics.

In terms of antimicrobial drug resistance, when using mNGS for pathogenic microorganism identification, it is possible to consider detecting drug resistance genes. However, it is necessary to accurately locate the drug-resistance genes on the specific pathogen to determine their clinical value. This means that after assembling the genomic sequences obtained from mNGS, it is necessary to determine which pathogen the drug resistance gene is located on, whether it is on a plasmid or on a chromosome. However, it is important to note that the presence of resistance/virulence genes does not necessarily mean they are expressed, and there may be discrepancies between genotype and phenotype. Additionally, some mechanisms leading to resistance/pathogenicity cannot be definitively identified through gene testing alone. Current technologies for species attribution analysis of resistance/virulence genes are not yet fully developed. For normal bacterial samples, accurately matching drug-resistance genes with specific pathogenic microorganisms poses significant challenges [58].

Interpreting reports poses a significant challenge in experimental contexts. When mNGS identifies pathogens with low sequence reads, caution is necessary in determining their diagnostic value. It is imprudent to solely rely on mNGS results to guide anti-infection therapy. Instead, diagnostic decisions should be informed by analyzing the type and collection reliability of the specimen, excluding interference from technical aspects of the detection process, and considering the type of pathogens detected in conjunction with the patient’s clinical condition for a comprehensive assessment, ultimately validating its clinical significance and value. For rare pathogens, such as Cryptococcus neoformans, further validation is necessary through other laboratory tests when detected at low sequence reads by mNGS. In cases where mNGS results are negative, other auxiliary test results should be integrated for diagnosis. It has been observed in clinical practice that children with negative mNGS results can be diagnosed with infections through alternative testing methods, highlighting the influence of sample collection and experimental procedures. Qian et al. [59] suggested that compared to traditional culture and smear testing, the negative predictive value (NPV) of mNGS is 95.79 % [95 % CI, 93.8–97.8 %]. M. tuberculosis and Aspergillus are noted to be pathogens that may evade detection, potentially due to their thick, complex cell walls which hinder nucleic acid release. Wilson et al. showed [60] that mNGS could accurately diagnose 18 cases of meningitis or encephalitis with negative cerebrospinal fluid results through serological experiments.

For positive mNGS results, if such findings align with clinical manifestations but lack other laboratory evidence, PCR verification is advised, alongside a recommendation for further conventional laboratory tests. If positive sequencing results are not supported by clinical manifestations or laboratory findings, diagnosis should not solely rely on sequencing results but rather on traditional laboratory findings as the primary clinical reference.

The heavy cost is another major challenge owing to which mNGS is currently unavailable in pediatric clinics. Viruses in pediatric respiratory infections are often caused by RNA viruses, such as respiratory syncytial virus, human rhinovirus, human metapneumovirus, influenza virus, parainfluenza virus, and coronavirus [61]; therefore, DNA+RNA sequencing is often required to identify the causative pathogen in the same patient but is expensive and cannot be universally applied. Moreover, there is limited literature in our inclusion that summarizes the sensitivity of mNGS alone for virus detection. Therefore, when using mNGS, the existing consensus should be referred to, combined with clinical manifestations, to select the appropriate sequencing technology and to avoid overuse. Clinical staff should combine the mNGS results with the clinical symptoms of the child and other tests to make a joint judgment of the patient’s condition.

Conclusions

In conclusion, although there are various difficulties in the clinical use of mNGS in pediatric infectious diseases, studies have shown that mNGS can improve the sensitivity of pathogenic microbial detection compared with traditional detection methods and compensate for the shortcomings associated with traditional detection techniques. Simultaneously, doctors and scientific researchers should improve various standards for the clinical application of this technology, reflecting its advantages of high efficiency as well as fast and accurate clinical diagnosis and treatment and changing the detection and diagnosis mode of infectious diseases in the past. With the development and clinical application of high-throughput mNGS, the cost of detection will be reduced in the future, and the problems of background bacteria, host ratio, and other aspects of differential diagnosis of childhood infectious diseases will be solved.