Multidrug (MDR) and extensively drug resistant (XDR) Gram negative prosthetic joint infections (PJI): Role of surgery and impact of colistin administration

Abstract

Factors influencing treatment outcome of patients with Gram-negative bacterial (GNB) multidrug-resistant (MDR) and extensively drug-resistant (XDR) prosthetic joint infection (PJIs) were analysed. Data were collected (2000–2015) by 18 centres. Treatment success was analysed by surgery type for PJI, resistance (MDR/XDR) and antimicrobials (colistin/non-colistin) using logistic regression and survival analyses. A total of 131 patients (mean age 73.0 years, 35.9% male, 58.8% with co-morbidities) with MDR (n = 108) or XDR (n = 23) GNB PJI were assessed. The most common pathogens were Escherichia coli (33.6%), Pseudomonas aeruginosa (25.2%), Klebsiella pneumoniae (21.4%) and Enterobacter cloacae (17.6%). Pseudomonas aeruginosa predominated in XDR cases. Isolates were carbapenem-resistant (n = 12), fluoroquinolone-resistant (n = 63) and ESBL-producers (n = 94). Treatment outcome was worse in XDR versus MDR cases (P = 0.018). Success rates did not differ for colistin versus non-colistin in XDR cases (P = 0.657), but colistin was less successful in MDR cases (P = 0.018). Debridement, antibiotics and implant retention (DAIR) (n = 67) was associated with higher failure rates versus non-DAIR (n = 64) (OR = 3.57, 95% CI 1.68–7.58; P < 0.001). Superiority of non-DAIR was confirmed by Kaplan–Meir analysis (HR = 0.36, 95% CI 0.20–0.67) and remained unchangeable by time of infection (early/late), antimicrobial resistance (MDR/XDR) and antimicrobials (colistin/non-colistin) (Breslow–Day, P = 0.737). DAIR is associated with higher failure rates even in early MDR/XDR GNB PJIs versus implant removal. Colistin should be preserved for XDR cases as it is detrimental in MDR infections.