Evaluation of ceftriaxone and other antibiotics against Escherichia coli, Pseudomonas aeruginosa, and Streptococcus pneumoniae under in vitro conditions simulating those of serious infections

Predicting Antimicrobial Activity at the Target Site: Pharmacokinetic/Pharmacodynamic Indices versus Time-Kill Approaches

Antibiotic dosing strategies are generally based on systemic drug concentrations. However, drug concentrations at the infection site drive antimicrobial effect, and efficacy predictions and dosing strategies should be based on these concentrations. We set out to review different translational pharmacokinetic-pharmacodynamic (PK/PD) approaches from a target site perspective. The most common approach involves calculating the probability of attaining animal-derived PK/PD index targets, which link PK parameters to antimicrobial susceptibility measures. This approach is time efficient but ignores some aspects of the shape of the PK profile and inter-species differences in drug clearance and distribution, and provides no information on the PD time-course. Time–kill curves, in contrast, depict bacterial response over time. In vitro dynamic time–kill setups allow for the evaluation of bacterial response to clinical PK profiles, but are not representative of the infection site environment. The translational value of in vivo time–kill experiments, conversely, is limited from a PK perspective. Computational PK/PD models, especially when developed using both in vitro and in vivo data and coupled to target site PK models, can bridge translational gaps in both PK and PD. Ultimately, clinical PK and experimental and computational tools should be combined to tailor antibiotic treatment strategies to the site of infection.

Experimental Validation of a Mathematical Framework to Simulate Antibiotics with Distinct Half-Lives Concurrently in an In Vitro Model

Antimicrobial resistance has been steadily increasing in prevalence, and combination therapy is commonly used to treat infections due to multidrug resistant bacteria. Under certain circumstances, combination therapy of three or more drugs may be necessary, which makes it necessary to simulate the pharmacokinetic profiles of more than two drugs concurrently in vitro. Recently, a general theoretical framework was developed to simulate three drugs with distinctly different half-lives. The objective of the study was to experimentally validate the theoretical model. Clinically relevant exposures of meropenem, ceftazidime, and ceftriaxone were simulated concurrently in a hollow-fiber infection model, with the corresponding half-lives of 1, 2.5, and 8 h, respectively. Serial samples were obtained over 24 h and drug concentrations were assayed using validated LC-MS/MS methods. A one-compartment model with zero-order input was used to characterize the observed concentration-time profiles. The experimentally observed half-lives corresponding to exponential decline of all three drugs were in good agreement with the respective values anticipated at the experiment design stage. These results were reproducible when the experiment was repeated on a different day. The validated benchtop setup can be used as a more flexible preclinical tool to explore the effectiveness of various drug combinations against multidrug resistant bacteria.

Antimicrobial pharmacokinetics and preclinical in vitro models to support optimized treatment approaches for uncomplicated lower urinary tract infections

INTRODUCTION

Urinary tract infections (UTIs) are extremely common. Millions of people, particularly healthy women, are affected worldwide every year. One-in-two women will have a recurrence within 12-months of an initial UTI. Inadequate treatment risks worsening infection leading to acute pyelonephritis, bacteremia and sepsis. In an era of increasing antimicrobial resistance, it is critical to provide optimized antimicrobial treatment.

AREAS COVERED

Literature was searched using PubMed and Google Scholar (up to 06/2020), examining the etiology, diagnosis and oral antimicrobial therapy for uncomplicated UTIs, with emphasis on urinary antimicrobial pharmacokinetics (PK) and the application of dynamic in vitro models for the pharmacodynamic (PD) profiling of pathogen response.

EXPERT OPINION

The majority of antimicrobial agents included in international guidelines were developed decades ago without well-described dose-response relationships. Microbiology laboratories still apply standard diagnostic methodology that has essentially remained unchanged for decades. Furthermore, it is uncertain how relevant standard in vitro susceptibility is for predicting antimicrobial efficacy in urine. In order to optimize UTI treatments, clinicians must exploit the urine-specific PK of antimicrobial agents. Dynamic in vitro models are valuable tools to examine the PK/PD and urodynamic variables associated with UTIs, while informing uropathogen susceptibility reporting, optimized dosing schedules, clinical trials and treatment guidelines.

Dynamic Laser Speckle Imaging Meets Machine Learning to Enable Rapid Antibacterial Susceptibility Testing (DyRAST)

Rapid antibacterial susceptibility testing (RAST) methods are of significant importance in healthcare, as they can assist caregivers in timely administration of the correct treatments. Various RAST techniques have been reported for tracking bacterial phenotypes, including size, shape, motion, and redox state. However, they still require bulky and expensive instruments-which hinder their application in resource-limited environments-and/or utilize labeling reagents which can interfere with antibiotics and add to the total cost. Furthermore, the existing RAST methods do not address the potential gradual adaptation of bacteria to antibiotics, which can lead to a false diagnosis. In this work, we present a RAST approach by leveraging machine learning to analyze time-resolved dynamic laser speckle imaging (DLSI) results. DLSI captures the change in bacterial motion in response to antibiotic treatments. Our method accurately predicts the minimum inhibitory concentration (MIC) of ampicillin and gentamicin for a model strain of Escherichia coli (E. coli K-12) in 60 min, compared to 6 h using the currently FDA-approved phenotype-based RAST technique. In addition to ampicillin (a β-lactam) and gentamicin (an aminoglycoside), we studied the effect of ceftriaxone (a third-generation cephalosporin) on E. coli K-12. The machine learning algorithm was trained and validated using the overnight results of a gold standard antibacterial susceptibility testing method enabling prediction of MIC with a similarly high accuracy yet substantially faster.

Receptor Blockade: A Novel Approach to Protect the Brain From Pneumococcal Invasion

Background

Pneumococci are the major cause of bacterial meningitis globally. To cause meningitis pneumococci interact with the 2 endothelial receptors, polymeric immunoglobulin receptor (pIgR) and platelet endothelial cell adhesion molecule (PECAM-1), to penetrate the blood-brain barrier (BBB) and invade the brain.

Methods

C57BL/6 mice were infected intravenously with bioluminescent pneumococci, and treated with ceftriaxone (1 hour postinfection) and anti-pIgR and PECAM-1 antibodies (1 or 5 hours postinfection), then monitored for 5 and 10 days. Bacterial brain invasion was analyzed using IVIS imaging and bacterial counts.

Results

Ceftriaxone, given early after pneumococcal challenge, cleared pneumococci from the blood but not from the brain. After combining ceftriaxone with receptor blockade, using anti-pIgR and PECAM-1 antibodies, we found 100% survival after 5 and 10 days of infection, in contrast to 60% for ceftriaxone alone. Combined antibiotic and antibody treatment resulted in no or few viable bacteria in the brain and no microglia activation. Antibodies remained bound to the receptors during the study period. Receptor blockade did not interfere with antibiotic permeability through the BBB.

Conclusions

We suggest that adjunct treatment with pIgR and PECAM-1 antibodies to antibiotics may prevent pneumococcal meningitis development and associated brain damages. However, further evaluations are required.

Fosfomycin efficacy and emergence of resistance among Enterobacteriaceae in an in vitro dynamic bladder infection model

Background

Urinary tract infections (UTIs) are among the most common bacterial infections and a frequent indication for antibiotic use. Fosfomycin, an important oral antibiotic for outpatient UTIs, remains a viable option for MDR uropathogens. We aimed to perform pharmacodynamic profiling simulating urinary concentrations to assess the adequacy of the current dosing regimen.

Methods

A dynamic in vitro bladder infection model was developed, replicating urinary fosfomycin concentrations after gastrointestinal absorption, systemic distribution and urinary elimination. Concentrations were measured by LC-MS/MS. Twenty-four Enterobacteriaceae strains (Escherichia coli, Klebsiella pneumoniae and Enterobacter cloacae; MIC range 0.25-64 mg/L) were examined. Pathogen kill and emergence of resistance was assessed over 72 h.

Results

Observed in vitro fosfomycin concentrations accurately simulated urinary fosfomycin exposures (Tmax 3.8 ± 0.5 h; Cmax 2630.1 ± 245.7 mg/L; AUC0-24 33 932.5 ± 1964.2 mg·h/L). Fifteen of 24 isolates regrew, with significant rises in fosfomycin MIC (total population MIC50 4 to 64 mg/L, MIC90 64 to > 1024 mg/L, P = 0.0039; resistant subpopulation MIC50 128 to > 1024 mg/L, MIC90 >1024 mg/L, P = 0.0020). E. coli and E. cloacae isolates were killed with pharmacokinetic/pharmacodynamic EI50 of fAUC0-24/MIC = 1922, fCmax/MIC = 149 and fTime>4×MIC = 44 h. In contrast, K. pneumoniae isolates were not reliably killed.

Conclusions

Using dynamic in vitro simulations of urinary fosfomycin exposures, E. coli and E. cloacae isolates with MIC >16 mg/L, and all K. pneumoniae isolates, were not reliably killed. Emergence of resistance was significant. This challenges fosfomycin dosing and clinical breakpoints, and questions the utility of fosfomycin against K. pneumoniae. Further work on in vitro dose optimization is required.

In vitro pharmacodynamic models to determine the effect of antibacterial drugs

In vitro pharmacodynamic (PD) models are used to obtain useful quantitative information on the effect of either single drugs or drug combinations against bacteria. This review provides an overview of in vitro PD models and their experimental implementation. Models are categorized on the basis of whether the drug concentration remains constant or changes and whether there is a loss of bacteria from the system. Further subdifferentiation is based on whether bacterial loss involves dilution of the medium or is associated with dialysis or diffusion. For comprehension of the underlying principles, experimental settings are simplified and schematically illustrated, including the simulations of various in vivo routes of administration. The different model types are categorized and their (dis)advantages discussed. The application of in vitro models to special organs, infections and pathogens is comprehensively presented. Finally, the relevance and perspectives of in vitro investigations in drug discovery and clinical research are elucidated and discussed.

What in vitro models of infection can and cannot do

The science of pharmacodynamics analyzes the relationship between an antimicrobial's bactericidal effects and its pharmacokinetics. Ideally, randomized and well-controlled clinical trials are the best way to determine pharmacodynamic properties. However, in vitro models that recapitulate in vivo drug clearance profiles represent an increasingly important technology for carrying out pharmacodynamic studies in a more cost-effective, timely, and easily controlled fashion. Although in vitro pharmacodynamic models cannot incorporate all variables seen in vivo, they do provide valuable information for the drug development process and the determination of optimal dosing regimens.

Use of Modeling Techniques to Aid in Antibiotic Selection

Over the past decade, the use of modeling techniques in the development of novel antibiotics has been primarily associated with in vitro dynamic models. These models allow comparisons among different antibiotics by simulating human pharmacokinetics. Although dynamic models have been used extensively, their full potential has not been achieved because of inadequate experimental design and/or suboptimal quantitation of bacterial killing/regrowth curves inherent in many studies. These issues are discussed in this review, which is based on recent pharmacodynamic findings with novel fluoroquinolones.

Prediction of the effects of inoculum size on the antimicrobial action of trovafloxacin and ciprofloxacin against Staphylococcus aureus and Escherichia coli in an in vitro dynamic model

The effect of inoculum size (N0) on antimicrobial action has not been extensively studied in in vitro dynamic models. To investigate this effect and its predictability, killing and regrowth kinetics of Staphylococcus aureus and Escherichia coli exposed to monoexponentially decreasing concentrations of trovafloxacin (as a single dose) and ciprofloxacin (two doses at a 12-h interval) were compared at N0 = 10(6) and 10(9) CFU/ml (S. aureus) and at N0 = 10(6), 10(7), and 10(9) CFU/ml (E. coli). A series of pharmacokinetic profiles of trovafloxacin and ciprofloxacin with respective half-lives of 9.2 and 4 h were simulated at different ratios of area under the concentration-time curve (AUC) to MIC (in [micrograms x hours/milliliter]/[micrograms/milliliter]): 58 to 466 with trovafloxacin and 116 to 932 with ciprofloxacin for S. aureus and 58 to 233 and 116 to 466 for E. coli, respectively. Although the effect of N0 was more pronounced for E. coli than for S. aureus, only a minor increase in minimum numbers of surviving bacteria and an almost negligible delay in their regrowth were associated with an increase of the N0 for both organisms. The N0-induced reductions of the intensity of the antimicrobial effect (IE, area between control growth and the killing-regrowth curves) were also relatively small. However, the N0 effect could not be eliminated either by simple shifting of the time-kill curves obtained at higher N0s by the difference between the higher and lowest N0 or by operating with IEs determined within the N0-adopted upper limits of bacterial numbers (IE's). By using multivariate correlation and regression analyses, linear relationships between IE and log AUC/MIC and log N0 related to the respective mean values [(log AUC/MIC)average and (log N0)average] were established for both trovafloxacin and ciprofloxacin against each of the strains (r2 = 0.97 to 0.99). The antimicrobial effect may be accurately predicted at a given AUC/MIC of trovafloxacin or ciprofloxacin and at a given N0 based on the relationship IE = a + b [(log AUC/MIC)/(log AUC/MIC)average] - c [(log N0)/(log N0)average]. Moreover, the relative impacts of AUC/MIC and N0 on IE may be evaluated. Since the c/b ratios for trovafloxacin and ciprofloxacin against E. coli were much lower (0.3 to 0.4) than that for ampicillin-sulbactam as examined previously (1.9), the inoculum effect with the quinolones may be much less pronounced than with the beta-lactams. The described approach to the analysis of the inoculum effect in in vitro dynamic models might be useful in studies with other antibiotic classes.

Parameters of bacterial killing and regrowth kinetics and antimicrobial effect examined in terms of area under the concentration-time curve relationships: action of ciprofloxacin against Escherichia coli in an in vitro dynamic model

Although many parameters have been described to quantitate the killing and regrowth of bacteria, substantial shortcomings are inherent in most of them, such as low sensitivity to pharmacokinetic determinants of the antimicrobial effect, an inability to predict a total effect, insufficient robustness, and uncertain interrelations between the parameters that prevent an ultimate determination of the effect. To examine different parameters, the kinetics of killing and regrowth of Escherichia coli (MIC, 0.013 microg/ml) were studied in vitro by simulating a series of ciprofloxacin monoexponential pharmacokinetic profiles. Initial ciprofloxacin concentrations varied from 0.02 to 19.2 microg/ml, whereas the half-life of 4 h was the same in all experiments. The following parameters were calculated and estimated: the time to reduce the initial inoculum (N0) 10-, 100-, and 1,000-fold (T90%, T99%, and T99.9%, respectively), the rate constant of bacterial elimination (k(elb)), the nadir level (Nmin) in the viable count (N)-versus-time (t) curve, the time to reach Nmin (t(min)), the numbers of bacteria that survived (Ntau) by the end of the observation period (tau), the area under the bacterial killing and regrowth curve (log N(A)-t curve) from the zero point (time zero) to tau (AUBC), the area above this curve (AAC), the area between the control growth curve (log N(C)-t curve) and the bacterial killing and regrowth curve (log N(A)-t curve) from the zero point to tau (ABBC) or to the time point when log N(A) reaches the maximal values observed in the log N(C)-t curve (I(E); intensity of the effect), and the time shift between the control growth and regrowth curves (T(E); duration of the effect). Being highly sensitive to the AUC, I(E), and T(E) showed the most regular AUC relationships: the effect expressed by I(E) or T(E) increased systematically when the AUC or initial concentration of ciprofloxacin rose. Other parameters, especially T90%, T99%, T99.9%, t(min), and log N0 - log Nmin = delta log Nmin, related to the AUC less regularly and were poorly sensitive to the AUC. T(E) proved to be the best predictor and t(min) proved to be the worst predictor of the total antimicrobial effect reflected by I(E). Distinct feedback relationships between the effect determination and the experimental design were demonstrated. It was shown that unjustified shortening of the observation period, i.e., cutting off the log N(A)-t curves, may lead to the degeneration of the AUC-response relationships, as expressed by log N0 - log Ntau = delta log Ntau, AUBC, AAC, or ABBC, to a point where it gives rise to the false idea of an AUC- or concentration-independent effect. Thus, use of I(E) and T(E) provides the most unbiased, robust, and comprehensive means of determining the antimicrobial effect.

Comparison of plasma and free tissue levels of ceftriaxone in rats by microdialysis

Ceftriaxone has a very high plasma protein binding (up to 98%) that is saturable and decreases with higher concentrations. This high protein binding results in high concentrations in plasma that are frequently related to the anti-infective activity. However, because only the free fraction of the drug is pharmacologically active and most of the infections are located in the tissues, it is more relevant to evaluate unbound concentrations in the interstitial space. Plasma and tissue pharmacokinetics of ceftriaxone in rats after single intravenous administration were investigated at two different concentrations (50 and 100 mg/kg). Both plasma and tissue samples were taken simultaneously from the same animal and analyzed by reversed-phase high-performance liquid chromatography. Free tissue levels in the thigh muscle were measured by microdialysis. The concentration in plasma is much higher than the free concentration in tissue. After determination of nonlinear protein binding by microdialysis and including these parameters in the pharmacokinetic model, it is possible to predict free concentrations in the interstitial space from plasma levels for any given dose.

Net effect of inoculum size on antimicrobial action of ampicillin-sulbactam: studies using an in vitro dynamic model

To examine the predictable effect of inoculum size on the kinetics of the antimicrobial action of ampicillin-sulbactam, five TEM-1 beta-lactamase-producing Escherichia coli strains were studied in an in vitro dynamic model at two different initial inocula (N0S). All bacteria were exposed to ampicillin-sulbactam in a simulated system reflecting the pharmacokinetic profiles in human tissue after the administration of a single intravenous dose of ampicillin (2 g) plus sulbactam (1 g). Each strain was studied at low (4.0 to 5.2 log CFU/ml) and high (5.0 to 7.1 log CFU/ml) N0S. Despite pronounced differences in susceptibilities, the patterns of the killing curves observed with a given strain at different N0S were similar. As expected, viable bacterial counts increased with inoculum size. Striking visual contrasts in the respective curves for each organism were reflected by the area under the bacterial count-time curve (AUBC) but not by the difference between the N0 and the lowest bacterial counts (Nmin) at the nadir of the killing curve: the N0-associated changes in the AUBC on average were 75%, versus 2.5% for log N0--logNmin. To examine qualitative differences in antimicrobial effects at different N0S (i.e., the net effect of the inoculum), the difference in the high and low N0S was subtracted from each point on the killing curve obtained at the higher N0 for each strain. These adjusted curves were virtually superimposable on the observed killing curves obtained at the lower N0. Moreover, by using adjusted data, the AUBC values were similar at the two inocula, although slight (average, 11%) but systematic increases in the AUBC occurred at high N0S. Thus, there was only a weak net effect of inoculum size on the antibacterial effect of ampicillin-sulbactam. Due to similar slopes of the AUBC-log N0 plots, the antibacterial action at different N0S may be easily predicted by an approximate equation; the predicted AUBCs were unbiased and well correlated with the observed AUBCs (r = 0.997). Compiled data obtained with normalized AUBCs for different strains at different N0S yielded a positive correlation (r = 0.963) between the N0-normalized AUBC and the MIC of ampicillin-sulbactam. The adjustment and normalization procedure described might be a useful tool for revealing the net effect of the inoculum and to predict the inoculum effect if there are no qualitative differences in antimicrobial action at different inocula.

In vitro synergism of ceftriaxone combined with aminoglycosides against Pseudomonas aeruginosa

The antipseudomonal activities of ceftriaxone (CEF) or ceftazidime (CAZ), each combined with tobramycin (TOB) or netilmicin (NET), against 90 clinically significant Pseudomonas aeruginosa isolates were examined both by checkerboard and time-kill assays. As expected, susceptibility testing of single antibiotics by agar dilution demonstrated good activity for CAZ (89% susceptible), TOB (94%), and NET (58%), but poor activity for CEF (15%). Checkerboard studies revealed striking synergy (FIC indices < or = 0.5) for CEF, however, in combination with either TOB (72%) or NET (81%), compared with CAZ-TOB (44%) or CAZ-NET (60%) (P < 0.01, respectively). No antagonism (FIC indices > or = 4) was found in any of these combinations. The MIC90s of CEF, CAZ, or aminoglycosides in the combinations were reduced at least fourfold: CEF, from > 128 to 32 mg/liter; CAZ, from 16 to 4 mg/liter; TOB, from 4 to 0.5 mg/liter; and NET, from 32 to 4 mg/liter. With CEF and NET, the percentage of strains sensitive to < or = 8 mg/liter of both drugs alone and in combination increased from 9% to 69%. Results of the time-kill assay for CEF-NET agreed reasonably well with the checkerboard method (Spearman rank correlation coefficient, 0.40, P < or = 0.01), and generated a bactericidal outcome in 60% (24 of the 40 isolates), when tested with combinations at 1/4 MBC of either antibiotic alone. Importantly, concentrations of CEF and aminoglycoside combinations that demonstrated synergy by either checkerboard or time-kill assays were achievable in serum clinically. These data suggest a unique interaction of CEF-aminoglycoside combinations against P. aeruginosa.(ABSTRACT TRUNCATED AT 250 WORDS)