Jinju Guk, Charles Burdet, Mélanie Magnan, Xavier Duval, Nathalie Grall, Olivier Clermont, Erick Denamur, France Mentré, Jérémie Guedj.
Universite de Paris, IAME, INSERM, F-75018 Paris, France
Objectives: Antibiotic residues profoundly disturb the ecological equilibrium which may lead to modified bacterial dynamics and dysbiosis. Yet, little is known on how these bacterial dynamics occur, and to what extent it is modulated by the drug concentration. To study this question, we conducted the CEREMI trial [1], that compared the impact of two antibiotics, ceftriaxone (CRO) and cefotaxime (CTX) on gut microbiota in healthy volunteers. CRO and CTX share antibacterial spectrum and clinical indications; they differ in their pharmacokinetic (PK) properties [1]. Here, we aimed to perform a model-based analysis to characterize the impact of these drugs on the bacterial populations of gut microbiota.
Methods: The CEREMI trial (PI X. Duval, sponsor AP-HP) enrolled 22 healthy volunteers (HV) randomly assigned to receive short infusion either CRO (1g/d, n=11) or CTX (1g/8h, n=11) for 3 days. Fecal samples collected three times before treatment, every day during treatment, seven times in the first month after the end of treatment and up to 180 days after treatment cessation, for determination of fecal drug concentrations, total Gram-negative bacilli counts (GNB, 13 samples/HV), Escherichia spp. counts by qPCR (7 samples/HV) and total bacterial counts by flux cytometry. The bacterial growth of each population was modeled by a logistic growth model [2]. Fecal concentrations were undetectable in the vast majority of the patients, probably due to the action of beta-lactamases, and could not be used as a marker of drug exposure in the gut. Therefore, we assumed that ‘pseudo’ drug concentrations in the gut obeyed a simple pharmacokinetic model, with a 0 or 1st order absorption during treatment period administration, followed by a 1st order elimination after the 3 days of the treatment period. Without loss of generality, the maximal drug concentrations were fixed to 1. We modeled the increase of the rate of bacterial elimination related to the drug assuming either linear, log-linear or Emax effect of pseudo-concentration. The best PK model was selected using the GNB counts data, for which richer sampling design was available during and after treatment. We then modeled separately Escherichia spp. counts and total bacterial counts, using individual PK parameters obtained from the model of GNB. For each model selection was done using the BIC and plausibility of parameter estimates. Parameter estimation was performed using nonlinear mixed effect models and the SAEM algorithm in Monolix 2018R2 (Lixoft, France).
Results: The best model describing drug PK was a first-order absorption and elimination in feces, with a maximal concentration achieved at the end of the antibiotic treatment. Interestingly, CTX had a higher elimination rate than CRO (0.25/day (RSE 24.6 %) vs 1.24/day (RSE 44.2 %), p=0.003), consistent with results found in blood (3.84/day for CRO, 40.6/day for CTX). Bacterial initial populations were equal to 7.46 log10 CFU/g, 5.72 log10 CFU/g and 11.3 log10 CFU/g for GNB, Escherichia spp., and total bacterial counts, respectively, and bacterial maximal doubling time were equal to 1.04/day, 0.84/day and 1.46/day for GNB, Escherichia spp., and total bacterial counts. In all bacterial populations, linear drug effect best described the impact of pseudo drug concentration on bacterial elimination. The drug sensitivity did not differ between CRO and CRX in any bacterial populations. Likewise, the drug sensitivity was largely similar for GNB and Escherichia spp. with values of 0.69 and 0.57, respectively. Our model predicted that GNB and Escherichia spp. were rapidly reduced after treatment, leading to a median drop of 3.62 log10 CFU/g for CRO and 3.26 log10 CFU/g for CTX in GNB, and of 2.68 log10 CFU/g for CRO and 2.34 log10 CFU/g for CTX in Escherichia spp. at the end of treatment duration. Further, the model predicted that 14 days (CRO) and 6 days (CTX) were needed to come back to 95% of steady-state values. Of note, total bacterial counts were mostly not affected by the treatment except for two individuals.
Conclusions: Our model could quantify the impact of CRO and CTX on two gut bacterial populations. Although the action of beta-lactamase prevented direct observation of drug in feces, we could predict that CRO had a higher elimination rate in the gut than CTX, consistent with what was observed in the blood. Both GNB and Escherichia spp. were highly sensitive to drug concentrations.
References:
[1] Burdet, C. et al. Ceftriaxone and cefotaxime have similar effects on the intestinal microbiota in human volunteers treated by standard-dose regimens. Antimicrob. Agents Chemother. 63, e02244-18 (2019).
[2] T. B. Robertson, The Chemical Basis of Growth and Senescence. Lippincott, Philadelphia,1923.
Reference: PAGE () Abstr 9476 [www.page-meeting.org/?abstract=9476]
Poster: Oral: Drug/Disease Modelling