Periprosthetic joint infection (PJI) is one of the serious complications after total joint replacement (McNally et al., 2021). Timely and accurate microbiological diagnosis is crucial for effective infection control and clinical decision-making (Huang et al., 2020). False-negative results due to low activity or low concentrations of bacteria, and difficult-to-cultivate pathogens in vitro are common limitations of this traditional method (Berbari et al., 2007; Font-Vizcarra et al., 2010; Parvizi et al., 2014; Tan et al., 2018; Tarabichi et al., 2018a). The metagenomic next-generation sequencing (mNGS) is a molecular detection method for pathogens that have attracted great attention in recent years. This method detects all microbial nucleic acid fragments in a specimen by unbiased shotgun sequencing and prompts its correlation with clinical infection. The mNGS technology has demonstrated broad application in pathogen identification of PJI cases (Goswami and Parvizi, 2020; Huang et al., 2020; Huang et al., 2020; He et al., 2021). Through improving the sensitivity of pathogen detection, mNGS has a positive impact on improving the prognosis of PJI patients, rational use of antibiotics, and reducing the economic burden of patients (Wang et al., 2020).

Because of its extremely high analytical sensitivity, even trace amounts of exogenous nucleic acids introduced during sampling, reagent handling, or laboratory procedures can generate false-positive signals, which is one of the biggest challenges in interpreting mNGS testing reports. Street et al. concluded that high levels of DNA contamination may be introduced in the process of mNGS, which is difficult to remove and can only be minimized during the operation (Street et al., 2017). Huang et al. also reported the interference of exogenous DNA contamination in the detection of bone and joint infections using mNGS (Huang et al., 2019). Detecting contaminating nucleic acid sequences during mNGS and mistaking them for pathogenic microorganisms may be a common problem for all inspectors (Tarabichi et al., 2018b). Such misclassification directly affects clinical decisions, including unnecessary antimicrobial therapy, prolonged hospitalization, inappropriate revision surgery, and increased healthcare costs.

Given that contamination patterns vary by institution, workflow, and laboratory environment, the lack of standardized background microbial references increases the risk of false-positive interpretation in PJI diagnostics. Therefore, establishing institution-specific background microbial libraries (BMLs) is essential to correctly distinguish true pathogens from residual environmental or reagent-derived signals.

In this study, for the first time, we experimentally detected and established PJI-related interfering nucleic acid BML in different medical institutions. We obtained residual nucleic acid on the surface of instruments for joint arthroplastyin different medical institutions through mNGS, and applied bioinformatics methods to analyze and compare the differences in microbial distribution. To more accurately apply mNGS to diagnose PJI diseases, each medical institution must establish its BML database.

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 (Thoendel et al., 2018). In this study, we demonstrated that residual nucleic acid fragments persist on acetabular reamers used for joint arthroplasty across different medical institutions, despite complete inactivation of viable microorganisms by routine sterilization. These background DNA signals, predominantly bacterial in origin, varied markedly between hospitals and formed institution-specific microbial patterns. Such residual fragments are not removed by standard disinfection because DNA is more stable than viable organisms, and their presence can lead to false-positive pathogen detection in mNGS. Given that mNGS reports often guide antimicrobial selection and surgical planning in PJI, misinterpreting background DNA as true infection poses a risk of unnecessary antibiotic exposure or inappropriate operative intervention. Our findings therefore highlight the clinical importance of recognizing and accounting for institution-specific background microbial DNA when interpreting mNGS results for PJI diagnosis.

Recent studies on PJI etiology have shown that Staphylococcus remains the most common pathogen, with gram-positive cocci, gram-negative bacilli, anaerobes, and Candida also contributing to infection, while rare organisms such as Aspergillus fumigatus, Actinomyces, and Mycoplasma hominis have been occasionally reported (Del Pozo and Patel, 2009; Cobo and Del Pozo, 2011; Benito et al., 2016; Bassetti et al., 2019). Given this microbial complexity, mNGS provides important diagnostic value, but its extremely high sensitivity makes it susceptible to false-positive signals generated by trace nucleic acid residues introduced during sampling or laboratory handling. Although sterile, nucleic-acid–free consumables are used, residual DNA in reagents or the instrument environment can still interfere with interpretation. In this study, two synchronized negative controls confirmed that detection reagents did not influence inter-hospital microbial differences. However, the presence of stable background DNA fragments on surgical instruments remains a critical challenge, particularly because many of the interfering taxa detected here (e.g., Corynebacterium, Novosphingobium, Sphingomonas) are uncommon in PJI yet may occasionally cause infection in immunocompromised patients (Street et al., 2017; Palan et al., 2019; Huang et al., 2020; Indelli et al., 2021). More importantly, some residues overlapped with common PJI pathogens such as Staphylococcus and Acinetobacter, complicating clinical interpretation.

Previous work by Thoendel et al. demonstrated that contaminating microbial DNA in metagenomic workflows, especially from reagents, complicates interpretation of PJI sonicate fluid sequencing (Thoendel et al., 2018). Their study identified characteristic reagent-derived contaminants such as Acinetobacter, Streptococcus, Cutibacterium, and Staphylococcus and highlighted the challenge of distinguishing these organisms from true pathogens. Our findings expand upon this by showing that contamination is not limited to reagents: even after complete sterilization, surgical instruments retain institution-specific microbial DNA signatures that vary substantially across hospitals. This suggests that, in addition to reagent contamination described by Thoendel et al., localized environmental or instrument-derived nucleic acids represent an additional and previously underrecognized source of background noise. By characterizing these hospital-level background profiles, our dataset complements and extends prior work, providing practical evidence supporting the need for institution-specific BMLs to improve the accuracy of mNGS-based PJI diagnosis.

Establishing institution-specific BMLs therefore provides an essential reference for interpreting mNGS results, although some contaminant taxa may still be difficult to distinguish from true pathogens. Determining the clinical significance of detected organisms requires integrated assessment using bioinformatic features (such as coverage and abundance), reagent background monitoring, and clinical context. For pathogens that are part of the human microbiome—such as coagulase-negative Staphylococci or Dermatobacter acnes—even quantitative thresholds may not reliably separate contamination from infection (Tande and Patel, 2014; Davidson et al., 2019; Rao et al., 2020). Consequently, mNGS interpretation must incorporate multiple factors, including patient condition, joint type, geographic variability, and timing relative to surgery, particularly in low-abundance infections where contamination can obscure true pathogen signals.

Furthermore, although bacteria constituted the majority of background microbial reads, low-level fungal, viral, and parasitic sequences were also detected across institutions. These non-bacterial signals are clinically relevant because mNGS reports frequently include rare fungal or viral reads that may be misinterpreted as true infections, often originating from bioinformatic artifacts (such as the misclassification of host DNA as Sporozoa) (Pain et al., 2008; Gu et al., 2019) or trace reagent contamination (the “kitome”) (Salter et al, 2014). Establishing hospital-specific background profiles for these microbial groups helps distinguish sporadic environmental or reagent-derived nucleic acids from clinically meaningful pathogens, thereby reducing unnecessary antifungal or antiviral treatments and improving the specificity of mNGS-based PJI diagnosis.

In addition to characterizing background microbial profiles, several practical strategies may help mitigate the influence of environmental or reagent-derived contamination in clinical mNGS workflows. Routine inclusion of reagent blanks, periodic environmental monitoring, and the application of statistical filtering approaches (such as abundance-based thresholds or contaminant-identification algorithms) can assist in identifying low-level background signals. Institution-specific BMLs can be further integrated into this framework by periodically sequencing sterilized instruments, negative controls, or operating-room environmental samples and storing the resulting background profiles in a local reference repository. These BMLs may be updated on a scheduled basis—such as every 3–6 months—or whenever laboratory workflows, reagent batches, or disinfection procedures undergo major changes. During routine mNGS interpretation, organisms detected in clinical specimens can then be compared against the institution’s BML to determine whether they reflect likely contaminants or true pathogens. Such combined mitigation and implementation strategies may substantially improve the accuracy and clinical utility of mNGS-based PJI diagnostics.

In interpreting our findings, it is essential to recognize that metagenomic sequencing data from low-biomass samples must be evaluated with caution and within the appropriate clinical context. The extremely high analytical sensitivity of mNGS allows trace environmental or reagent-derived nucleic acids to be detected alongside true pathogens, and such signals do not inherently imply clinical infection. In this study, low-abundance protozoan and viral reads were identified, but these organisms are biologically implausible causes of PJI and were therefore regarded as background metagenomic noise rather than meaningful microbiological findings. These observations underscore that mNGS should not be used as a stand-alone diagnostic tool. Accurate interpretation requires integration of sequencing results with clinical presentation, imaging, inflammatory markers, intraoperative findings, and conventional microbiology. Effective dialogue among clinicians, microbiologists, and bioinformaticians is critical to avoid overinterpretation of sequencing artefacts and to determine whether detected organisms have true pathogenic relevance. Our findings highlight that institution-specific background microbial libraries serve not as independent diagnostic criteria but as contextual references that support multidisciplinary evaluation and help distinguish genuine pathogens from environmental or reagent-derived signals during clinical decision-making.

The study still has several limitations. First, the sample size per institution was relatively small, which limits the statistical power of this exploratory analysis. Although clear inter-hospital clustering patterns and significant diversity differences were observed, larger and longitudinal sampling will be needed to further validate the robustness of these findings. Second, although we collected samples from multiple hospitals during the same period, we did not perform continuous monitoring of background microbial profiles across different time intervals. Similar to routine environmental microbiological surveillance in medical institutions based on culture-based assessments, periodic sequencing of residual nucleic acids from instruments used in joint arthroplasty may be required in the future. Third, because the sequences of residual viral and fungal fragments were very low, false-positive interference during formal diagnostic testing cannot be completely excluded. Fungal nucleic acids, especially from filamentous fungi, are difficult to extract due to their thick cell walls; therefore, interpretation rules for fungi cannot be directly applied using bacterial thresholds.

The surface of the instrument for joint replacement sterilized by the routine process does not contain any active microorganisms, but there are residual microbial nucleic acid fragments, which will interfere with the mNGS test results. The residual microbial nucleic acid fragments on the surface of joint replacement devices are mainly bacterial DNA, and there are differences in the distribution of bacteria between different medical institutions, which may be related to the historical cases handled by the medical treatment. In the current form, it is necessary to establish independent BML in different medical institutions to improve the accuracy of mNGS on the diagnosis of PJI.