The superiority of combination therapy for carbapenem-resistant Gram-negative bacilli (CR-GNB) infections remains controversial. In vitro models may predict the efficacy of antibiotic regimens against CR-GNB. A systematic review and meta-analysis was performed including pharmacokinetic/pharmacodynamic (PK/PD) and time-kill (TK) studies examining the in vitro efficacy of antibiotic combinations against CR-GNB [PROSPERO registration no. CRD42019128104]. The primary outcome was in vitro synergy based on the effect size (ES): high, ES ≥ 0.75, moderate, 0.35 < ES < 0.75; low, ES ≤ 0.35; and absent, ES = 0). A network meta-analysis assessed the bactericidal effect and re-growth rate (secondary outcomes). An adapted version of the ToxRTool was used for risk-of-bias assessment. Over 180 combination regimens from 136 studies were included. The most frequently analysed classes were polymyxins and carbapenems. Limited data were available for ceftazidime/avibactam, ceftolozane/tazobactam and imipenem/relebactam. High or moderate synergism was shown for polymyxin/rifampicin against Acinetobacter baumannii [ES = 0.91, 95% confidence interval (CI) 0.44-1.00], polymyxin/fosfomycin against Klebsiella pneumoniae (ES = 1.00, 95% CI 0.66-1.00) and imipenem/amikacin against Pseudomonas aeruginosa (ES = 1.00, 95% CI 0.21-1.00). Compared with monotherapy, increased bactericidal activity and lower re-growth rates were reported for colistin/fosfomycin and polymyxin/rifampicin in K. pneumoniae and for imipenem/amikacin or imipenem/tobramycin against P. aeruginosa. High quality was documented for 65% and 53% of PK/PD and TK studies, respectively. Well-designed in vitro studies should be encouraged to guide the selection of combination therapies in clinical trials and to improve the armamentarium against carbapenem-resistant bacteria.

Keywords: Antibiotic combination; Carbapenem-resistant bacteria; In vitro synergy; PK/PD; Time–kill.