At present, although a small number of studies have focused on the contamination of background nucleic acid fragments in mNGS sequencing, most of them are limited to empirical speculation [15]. To our knowledge, this study is the first to apply a data-based approach to study the impact of nucleic acid fragments in the instrument environment during mNGS on the diagnosis of PJI. Although we used primers that can amplify pathogen DNA fragments in the mNGS sequencing process, human-derived sequence fragments still accounted for the vast majority, which is consistent with the sequencing results of general joint samples. The main environmental pollution DNA fragments found in this study were bacterial fragments, while viral, fungal, and parasitic fragments were extremely rare. This is consistent with previous research findings and general clinical experience that most PJIs are due to bacterial infections [15]. There were no obvious differences in the sequences of the detected fungi and viruses, which may be related to their fewer fragments. The detected residues are considered to be more likely to be mismatched bases. Furthermore, we found significant differences in the distribution of bacterial DNA fragments on the surfaces of instruments in different hospitals, which is likely to be related to bacteria from previous PJI cases handled by the device. At present, conventional disinfection methods can effectively remove all active pathogens, and no pathogens were cultured in this study. Nevertheless, DNA fragments are difficult to remove by conventional sterilization means due to their inherent stability. Therefore, DNA fragments on the device will be stored for a long time and may affect the mNGS results of subsequent PJI patients.
In a recent study on the etiology of PJI, the results of Benito et al. identified Staphylococcus as the most common cause of infection [16]. In addition, other gram-positive cocci (8-9%), gram-negative bacilli (6%), anaerobic bacteria (4%) and Candida (1%) may also cause PJI [16, 17]. Rare microorganisms that cause PJI also include, but are not limited to, Aspergillus fumigatus, Actinomyces, and Mycoplasma hominis [18, 19]. Considering the complexity of PJI infection, the application of mNGS technology to detect pathogens has positive significance for the symptomatic treatment and prognosis of PJI. The high sensitivity of mNGS leads to the possibility of false positives caused by any nucleic acid residues of pathogenic microorganisms exposed during the sampling process. Thus, it is necessary to introduce the concept of BML and monitor their possible sources. During the sampling process, the containers used are sterile and nucleic acid-free according to standard procedures. Generally, no nucleic acid residues of pathogenic microorganisms are introduced. To exclude the influence of nucleic acid residues in the detection reagents, we set up two synchronized negative controls, including an unsampled swab control (Control 1) and pure water (Control 2). The results showed that the detection reagent did not affect the distribution of microorganisms among different hospitals. Therefore, medical institutions must establish a localized mNGS BML and the rules for distinguishing it from normal pathogenic bacteria. Besides, the detection, assessment, and periodic testing of nucleic acid fragments in reagents can help minimize cross-contamination between tested samples [2, 12, 20].
The current mainstream mNGS report interpretation methods tend to be comprehensively analyzed by a team composed of clinicians, laboratory technicians, and bioinformatics personnel. Even so, the fragment contamination of environmental nucleic acid cannot be well differentiated, resulting in false positives [21]. Most of the interfering fragments found in this study are pathogens that are not common in clinical infections (eg Corynebacterium, Novosphingobium, Sphingomonas), which are relatively easier to distinguish from PJI-infected bacteria. However, it cannot be ruled out that in immunocompromised patients, the joints will be infected with these special pathogenic bacteria. On the other hand, we also encountered residues of nucleic acid fragments with strong interference (eg, Staphylococcus, Acinetobacter), which overlapped with common pathogens of PJI. These fragments make the interpretation of the results more complicated, and better methods need to be explored to judge or exclude. One study summarized common background bacterial species through experience, including Acinetobacter, Streptococcus, Propionibacterium, Bradyrhizobium, Dolosigranulum, etc. [15]. However, this is only observational and has not been statistically and verified by data. The distribution of genus was significantly different from our study. It can be seen that the background bacteria of different platforms and different regions are not completely consistent.
Through the establishment of BML of different medical institutions, we can provide a strong reference for the reporting and interpretation of clinical specimens. However, it should be noted that some contaminating bacterial groups may be indistinguishable from infectious pathogens even by establishing a library. Identifying and removing contaminant DNA requires a complex set of methodologies. Reliable bioinformatics tools and databases are needed to decide whether detected microorganisms are the cause of infection or human-derived sequence contamination as well as background contamination. In addition, we tested samples from the same source with and without the introduction of human sequences and confirmed no significant differences. This may be related to the high concentration of human DNA on the surface of the device sample itself. Analyzing bioinformatic readouts (such as coverage, abundance, etc.) will play a key role in the adoption and utility of this method. However, in low-level infections caused by low-abundance pathogens, the presence of contaminants can make the correct interpretation of mNGS readings difficult. Furthermore, we run into the problem of increasing the amount of off-camera data to increase sensitivity, but the increased sensitivity of mNGS confounds the boundaries we draw in terms of organisms or relatedness of identified organisms. In special cases, factors such as the patient's condition, the joint to be replaced, the different geographic regions, and the time from surgery to consultation should be taken into account [22]. The pathogens themselves should also be characterized, for example, coagulase-negative Staphylococci species are part of the human skin microbiome and include a large group of bacteria such as Staphylococcus epidermidis and Staphylococcus ludens [23]. Another example is Dermatobacter acnes, a common skin-colonizing bacterium that can cause joint infections. Even with bacterial load thresholds set for these specific pathogens, it is difficult to differentiate between true pathogens and contaminants. Therefore, we need a comprehensive analysis combining multiple factors, such as a higher proportion of prosthetic shoulder joint infections caused by D. acnes compared to other joints [24].
The study still has several limitations. That is, we have collected samples from different hospitals in the same period, but have not conducted continuous monitoring on the background bacteria of the samples in different periods. Similar to the regular monitoring of environmental microorganisms in medical institutions based on routine bacterial culture, periodic sequencing of residual nucleic acids from instruments for joint arthroplasty may become possible in the future. In addition, due to the low sequence of residual fragments of viruses and fungi, false-positive interference informal testing cannot be completely ruled out. Among them, it is difficult for fungi to extract gene fragments due to their thick cell wall, especially filamentous fungi. Judgment rules cannot be treated equally with gene sequence fragments of bacteria. In other words, the diagnostic thresholds of these microbial infections require the establishment of more refined rules and the accumulation of experience to distinguish the correspondence between PJI or false positives and clinical decisions.