A Comparison of Pathogen Culture and Metagenomic Next-Generation Sequencing in Periprosthetic Joint Infections and Implications for Empirical Antibiotic Selection: An 9-Year Retrospective Study

Background Periprosthetic joint infection (PJI) is a severe complication of total joint arthroplasty. Traditional microbial culture has notable limitations, frequently resulting in culture-negative cases. Although mNGS improves pathogen detection, comparative studies between the two methods are still insufficient. This study aimed to build a PJI pathogen database via culture and mNGS, compare the two methods, and predict the efficacy of antibiotic combination regimens. Method This study was a single-center retrospective analysis conducted at our hospital, enrolling 407 patients diagnosed with PJI between 2016 and 2024. All patients underwent microbial culture and mNGS testing, and the efficacy of empirical antimicrobial therapy was evaluated based on antibiotic susceptibility test results. Based on antimicrobial susceptibility results, the predicted efficacy of combination regimens was calculated as 100% minus the probability of simultaneous resistance to all included agents. Result Compared with culture, mNGS had a lower negative rate (6.85% vs. 42.40%) and identified more mixed infections. Vancomycin-based regimens achieved 100% efficacy against Gram-positive pathogens, while rifampicin combinations were effective for MSSA and Streptococcus (> 98%) but less so for MRSA (12.8%–81.4%) and Enterococcus (53.3%–75.1%). For Gram-negative pathogens, most combinations showed high efficacy (> 80%), with ceftriaxone or meropenem plus amikacin achieving the highest rates for E. coli (99.7%–99.9%) and K. pneumoniae (93.7%–98.1%), whereas efficacy against P. aeruginosa was lower, particularly for ceftriaxone-based regimens (75.0%). Conclusion mNGS excelled in detecting rare and fastidious bacteria and mixed infections as a valuable supplement. Clinicians should refine diagnostic approaches by customizing microbial culture based on patients’ clinical status and integrating mNGS as appropriate. Integration of mNGS results with predictive model allows for optimal clinical antibiotic selection, providing more accurate and effective guidance for antimicrobial therapy.