An enhanced understanding of shoulder periprosthetic joint infection using next-generation sequencing: findings at...

Management of shoulder periprosthetic joint infection remains a clinical challenge with patients failing both medical and operative treatment despite adherence to best practice guidelines. An important question is whether these failures reflect reinfection of the causative organism or the persistence of infection due to inadequate treatment at the index surgery. Utilizing next-generation sequencing (NGS) and longitudinal clinical follow-up, this study hypothesized that some of these failures were the result of an organism that was present at the time of initial surgery but not isolated by traditional culture.

In this prospective study, samples were collected from 41 patients undergoing revision shoulder surgeries between June 2016 and May 2017. Synovial fluid and deep tissue samples (anterior capsule, inferior capsule, glenoid, humeral canal, and underneath prosthetic humeral head) were obtained intraoperatively in a standardized fashion and analyzed by NGS (MicroGenDx, TX). Concordant samples from these sites were also sent for institutional culture and held for 13 days. Demographics, comorbidities, and inflammatory markers were collected per standard-of-care. Patients were classified per the 2018 International Consensus Meeting Consensus definition of shoulder periprosthetic joint infection. Treatment failure was defined as any unplanned procedure for infection that yielded positive cultures, during the 3-year follow-up. Concordance of the infecting pathogen cultured at failure with the NGS analysis at the index revision procedure was determined.

There were 19 NGS-positive and 32 NGS-negative cases. During the study follow-up period, 6 unexpected reoperations (failures) occurred that yielded positive cultures, as well as 3 unexpected procedures for instability (n = 2) or cuff insufficiency (n = 1). NGS-positive results at index revision were significantly associated with subsequent failure by infection (P = .022), whereas positive cultures at index revision were not (P = .671). Furthermore, the organism cultured at subsequent failure was present on NGS testing at the index revision procedure in the majority of cases (4 of 6 patients; 66.7%). Five of the 6 cases (83.3%) that underwent unexpected reoperation did not receive prolonged antibiotics beyond standard perioperative prophylaxis due to negative cultures at the time of revision.

Our findings suggest that most of the unexpected failures (66.7%) by infection could be attributed to an organism previously detected by NGS and not identified by culture. Further multicenter work with larger numbers and longer-term follow-up is needed to corroborate any association between NGS analysis and subsequent treatment failure.