Cefepime Concentration-Toxicity Relationship in Hematological Patients Treated With Continuous Infusion: A Prospective Study.

BACKGROUND

Cefepime-induced neurotoxicity (CIN) is commonly described but neurotoxic concentration threshold remains unclear.

METHOD: This single-center prospective study included patients with febrile neutropenia treated by continuous infusion of cefepime in the hematology department. Both cefepime steady-state concentrations measurement (Css) and neurological assessment were performed at defined timepoints during the first 10 days after the onset of cefepime. Cefepime imputability was determined for each neurological event. The main objective was to assess the relationship between plasma cefepime exposure and the occurrence of neurotoxicity. Secondary objectives were to establish a cefepime Css threshold discriminating CIN and control group and identify influencing covariates.

RESULTS

One hundred and thirty-four patients were analyzed (346Css). CIN occurred in 9.7% of the cases (14/145 courses). Maximal Css were not significantly different between the 2 groups (53.2 ± 31.3 vs 39.5 ± 13.3 mg/L). Median neurotoxicity onset was 5 days. After multivariable analysis, both concomitant chimeric antigen receptor T-cell administration and each mg/mL increase in cefepime Css were independently associated with CIN onset (odds ratio 19.7 [5.6-69.1] and 1.030 [1.005-1.056], P < .001 and P = .020), respectively. Receiver operating characteristic curve analysis identified an adjusted threshold concentration of 36.7 mg/L (sensitivity 89%, specificity 90%) to differentiate the 2 groups.

CONCLUSIONS

This study provides real-life data on CIN using a multidisciplinary approach. Cefepime exposure was found to be independently associated with the occurrence of neurotoxicity together with chimeric antigen receptor T-cell therapy. The neurotoxicity threshold of 36.7 mg/L confirms a lower toxicity threshold than for other β lactams, offering an additional tool for the management of high-dose cefepime therapy.