Background: Klebsiella pneumoniae carbapenemase-producing K. pneumoniae (KPC-Kp) central nervous system (CNS) infections represent a major therapeutic challenge in neurosurgical patients. Intraventricular or intrathecal polymyxin B sulfate (PMB) is commonly used as salvage therapy but is limited by substantial neurotoxicity. Ceftazidime–avibactam (CZA) exhibits potent in vitro activity against KPC-Kp; however, prospective clinical and pharmacokinetic evidence supporting its use in CNS infections remains limited. Methods: In this prospective, single-centre observational study, adult neurosurgical patients with culture-confirmed KPC-Kp CNS infections admitted to the neurointensive care unit of Huashan Hospital were enrolled. Patients received either intravenous CZA (CZA group, n = 15) or intrathecal/intraventricular PMB-based therapy (PMB group, n = 10). Primary outcomes included clinical cure, microbiological eradication, 28-day mortality, and safety. Therapeutic drug monitoring was performed to determine steady-state plasma and cerebrospinal fluid (CSF) concentrations of ceftazidime, avibactam, and polymyxin B, enabling assessment of CSF penetration and exposure–toxicity relationships. Results: Overall clinical cure and microbiological eradication rates were 68.0% and 84.0%, respectively, with a 28-day mortality of 20.0%. Compared with PMB, CZA was associated with a significantly higher clinical cure rate (86.7% vs. 40.0%, p = 0.024) and a numerically higher eradication rate (93.3% vs. 70.0%). No neurological adverse events occurred in the CZA group, whereas neurological toxicity was observed in 60.0% of PMB-treated patients (p < 0.001). Functional outcomes favoured the CZA group, with lower modified Rankin Scale scores at discharge and at 6 months. Pharmacokinetic analyses demonstrated that steady-state CSF concentrations of ceftazidime and avibactam exceeded commonly accepted pharmacodynamic targets, while markedly elevated PMB CSF concentrations were observed in patients with neurological toxicity. Conclusions: While intravenous CZA showed potentially favourable efficacy and safety compared with local PMB in this cohort, these preliminary findings should be interpreted as hypothesis-generating given the small sample size and non-randomised design. These results provide a rationale for further validation in larger multicentre, randomised controlled trials.
Wang et al. (Wed,) studied this question.