Prediction of C‐reactive protein dynamics during meropenem treatment in neonates and infants

Aims C‐reactive protein (CRP) is used to determine the effect of antibiotic treatment on sepsis in neonates/infants. We aimed to develop pharmacokinetic–pharmacodynamic (PKPD) model of meropenem and CRP in neonates/infants and evaluate its predictive performance of CRP dynamics. Methods Data from neonates/infants treated with meropenem in 3 previous studies were analysed. To the previously developed meropenem PK models, the addition of turnover, transit or effect compartment, delay differential equation PD models of CRP as a function of meropenem concentration or its cumulative area under the curve (AUC) were evaluated. The percentage of neonates/infants (P 0.1 , P 0.2 ) in whom the ratio of the fifth day CRP to its peak value was predicted with an error of <0.1 (<0.2) was calculated. Results A total of 60 meropenem treatment episodes (median [range] gestational age 27.6 [22.6–40.9] weeks, postnatal age 13 [2–89] days) with a total of 351 CRP concentrations (maximum value 65.5 [13–358.4] mg/L) were included. Turnover model of CRP as a function of meropenem cumulative AUC provided the best fit and included CRP at the start of treatment, use of prior antibiotics, study and causative agent Staphylococcus aureus or enterococci as covariates. Using meropenem population predictions and data available at 0, 24, 48, 72 h after the start of treatment, P 0.1 (P 0.2 ) was 36.4, 36.4, 60.6 and 66.7% (70.0, 66.7, 72.7 and 78.7%), respectively. Conclusion The developed PKPD model of meropenem and CRP as a function of meropenem cumulative AUC incorporating several patient characteristics predicts CRP dynamics with an error of <0.2 in most neonates/infants.