Detection of periprosthetic joint infection pathogens by metagenomic next-generation sequencin

Abstract

Objective To investigate and compare the capability of metagenomic next-generation sequencing (mNGS) in detecting pathogens and diagnosing of periprosthetic joint infection (PJI) from synovial fluid and sonicate fluid of patients who underwent revision arthroplasty. Methods Thirty-five consecutive patients who underwent revision arthroplasty from May 2018 to November 2018 were included prospectively. There were 22 males and 13 females, 11 hip revisions and 24 knee revisions. All the patients were divided into the PJI group and aseptic loosening (AL) group. Synovial fluid and sonicate fluid of the explanted prostheses were obtained for microbiological culture and mNGS tests. Periprosthetic tissues were only collected for culture. Synovial fluid of three patients undergoing primary arthroplasty been treated by sonication as the negative control group concurrently. Comparisons of microbiological results and diagnostic value from mNGS and culture tests were performed. Results In the 13 culture-positive PJI patients, mNGS results of synovial fluid were positive in 12 cases, while culture and mNGS results were completely consistent at species level in 7 cases, consistent at the genus level in 1 case. mNGS results of sonicate fluid were positive in 13 cases, while culture and mNGS results were completely consistent at species level in 9 cases, consistent at the genus level in 1 case. In 7 culture-negative PJI patients, 6 cases had consistent mNGS results at species level both from synovial fluid and sonicate fluid, however, one case had positive mNGS result only from sonicate fluid. All culture results and mNGS results of synovial fluid were negative in all 15 AL patients, however, mNGS results of sonicate fluid was positive in 1 AL case. Cultures and mNGS results were negative in all three pairs of negative-control samples. In all 70 samples, mNGS detected 24 pathogens in sonicate fluids and 22 pathogens in synovial fluids. There was no significant difference in number of raw reads and human reads ratio between mNGS of sonicate fluid and synovial fluid. mNGS of sonicate fluid generated significantly higher number of microbial reads and of stringently mapped reads of pathogen in species-level than that of synovial fluids. There was no significant difference in diagnostic sensitivity of PJI between mNGS of sonicate fluids and synovial fluids (90.0% vs 100.0%). Both of them were significantly higher than that of culture of synovial fluid, periprosthetic tissues. Diagnostic sensitivity of sonicate fluid mNGS was not significantly higher than that in culture of sonicate fluid (65%). The specificities were similar among various microbiological testing methods. Conclusion mNGS of either synovial fluid or sonicate fluid from patients who underwent revision arthroplasty can be used to detect the presence of pathogens effectively and diagnose PJI accurately. mNGS can identify more pathogens and generate a higher number of pathogenic reads from sonicate fluids than synovial fluid. mNGS of synovial fluids has met the clinical diagnostic demands for most PJI patients. mNGS of sonicate fluid could be applied in some cases. Key words: Metagenomics; Prosthesis-related infections; Molecular diagnostic techniques; Sonication