Poster 11 - Adapting a low-cost continuous culture bioreactor for pharmacokinetic/pharmacodynamic modeling of antibiotic treatment adherence

Saturday, October 26, 2024

10:00 AM - 12:00 PM EST

Location: Exhibit Hall E, F & G

Presenting Author(s)

John Joyce (he/him/his)

Highschool Student
Hawken Highschool
Shaker Heights, Ohio, United States

Last Author(s)

RB

Rowan Barker-Clarke

Postdoctoral Fellow
Cleveland Clinic Foundation
Cleveland, Ohio, United States

Co-Author(s)

Anna Stacy (she/her/hers)

Undergraduate Student
Department of Physics, Case Western Reserve University, United States

Primary Investigator(s)

JS

Jacob Scott

Radiation Oncologist & Principal Investigator
Cleveland Clinic Foundation, United States

Co-Author(s)

VG

Vishhvaan Gopalakrishnan

Postdoctoral Fellow
Cleveland Clinic Foundation, United States

Introduction: With the WHO identifying antimicrobial resistance as one of the top ten global public health threats, efficient measures to understand antibiotic resistance are crucial [1]. Resistance evolution in vivo can be better understood using pharmacokinetic/pharmacodynamic (PK/PD) models, which simulate drug absorption, distribution, metabolism, and excretion, and the relationship between drug concentrations and their effects. PK/PD in silico and in vitro studies help simulate in vivo conditions, helping to optimize dosing regimens, predict resistance, minimize toxicity, and reduce uncertainties ahead of in vivo testing [2]. Continuous culture devices and bioreactors, such as morbidostats and chemostats, also simulate in vivo dynamics by permitting continuous longer-term evolutionary experiments.

Unfortunately, many PK/PD in vitro modeling methods suffer from high costs or limited dynamic control over dosing concentrations and schedules. Additionally, many commercially available bioreactors are expensive or inflexible to adaptation for PK/PD experiments. Existing low-cost continuous culture bioreactors have primarily been designed to analyze antibiotic resistance under sustained selection pressure without incorporating PK/PD parameters, and attempts to use hollow-fiber bioreactors for in vitro PK/PD modeling have been limited due to complex and cost-prohibitive sterilization [3].

To address the need for low-cost dynamic PK/PD models we extend the design of our ultra-low-cost EVolutionary biorEactor (EVE) to enable PK/PD modeling. We utilized in silico models to design algorithms for in-vitro PK/PD studies and demonstrated the EVE’s potential to produce these dynamic antibiotic environments in continuous culture over long durations.

Materials and

Methods: The EVE (Fig. 1A), a bioreactor initially designed as a morbidostat, permits precise and automated control over the exchange of the growth medium, cell density measurement rates, and drug addition [4]. Each of the four 24ml culture units is connected to two input flasks and one output flask via peristaltic pumps averaging a flow rate of 0.8ml/s, with all relevant mechanisms controlled by a Raspberry Pi. To enable this low-cost bioreactor to conduct PK/PD experiments, we implemented new arbitrary dosing and dilution algorithms with user control over inter-dose timing and dilution rates. We then experimentally verified the PK/PD algorithms by simulating different dosing schedules with solutions of different optical densities, validating changes in concentration by measured changes in optical density (OD). Water was used to simulate the permissive (i.e., drug-free) media, and blue-dyed water represented the selective media.

Our algorithms were designed for experiments testing the PK/PD impact of antibiotic treatment adherence, specifically analyzing the impact of missed doses, on the evolution of antibiotic resistance. To ensure relevance to the design of experimental protocols, we used the FEArS (Fast Evolution on Arbitrary Seascapes) python package to simulate the evolution of E. coli with beta-lactamase point mutations (Weinreich strains) under various Cefotaxime dosing regimens [5]. We simulated adherent and non-adherent multi-dose scheduling in silico model, using absorption and passaging parameters that adequately simulate the EVE in vitro.

Results, Conclusions, and Discussions: The plots of observed OD vs. predicted concentrations dynamics for both the simulated adherent and non-adherent schedules demonstrate consistency (Fig. 1B,C), indicating that our EVE algorithms are capable of simulating PK/PD experiments in vitro. Our FEArS models showed potential relevance of Cefotaxime schedule adherence to the treatment of E. coli Weinreich strains, and highlight that dropped doses had the potential to allow evolutionary escape and resistant to treatment to develop. Together, these approaches can be used in future experiments to study and verify our in silico predictions with FEArS in vitro with the EVE, as well as evaluate both previously tested and untested hypothetical dosing regimens.

Though using OD to measure the concentration of bacteria in a suspension is effective, OD detection signals can be subject to noise as shown in our results. Improving signal quality can be addressed in future work both by evaluating physical sources of noise and also using additional post-experiment signal processing methods to remove noise. Further, hardware additions could be made to allow the automated addition of multiple drugs into culture units, creating the ability to analyze multi-drug dosing regimens and their impacts on the evolution of antibiotic resistance.
Ultimately, the updated EVE system for in vitro PK/PD modeling, particularly when combined with in silico mathematical models and relevant clinical information, can be used to design improved dosing regimens in a low-cost, automated, and effective manner.

Acknowledgements (Optional):

References (Optional): [1] World Health Organization. “Antimicrobial resistance.” (2023). Available at:https://www.who.int/news-room/fact-sheets/detail/antimicrobial-resistance; [2] Vaddady, Pavan K et al. “In vitro pharmacokinetic/pharmacodynamic models in anti-infective drug development: focus on TB.” Future medicinal chemistry vol. 2,8 (2010): 1355-69. doi:10.4155/fmc.10.224; [3] Budha, Nageshwar R et al. “A simple in vitro PK/PD model system to determine time-kill curves of drugs against Mycobacteria.” Tuberculosis (Edinburgh, Scotland) vol. 89,5 (2009): 378-85. doi:10.1016/j.tube.2009.08.002; [4] Gopalakrishnan, Vishhvaan et al. “A low-cost, open-source evolutionary bioreactor and its educational use.” eLife vol. 11 e83067 (2022). doi:10.7554/eLife.83067; [5] King, Eshan S et al. “Fitness seascapes are necessary for realistic modeling of the evolutionary response to drug therapy” bioRxiv (2022). doi:10.1101/2022.06.10.495696