Skip to main content
Premium Trial:

Request an Annual Quote

UH Researchers Use Predictive Modeling Approach to Expedite Antibiotic Development


By Uduak Grace Thomas

Researchers at the University of Houston have developed a computational method that includes dosing regimens in the modeling process to explore the most effective treatment courses for bacterial infections.

The investigators claim that their method could potentially shave some years off development time for new antibiotics — which, according to current estimates, takes more than a decade and costs on average $800 million — because it lets researchers computationally screen thousands of drug candidates, dosing frequencies, and treatment durations to select the most lethal combination for the bugs.

The scientists, Vincent Tam, an associate professor of clinical sciences, and Michael Nikolaou, a professor of chemical and biomolecular engineering, published their method in a recent issue of PLoS Computational Biology.

"Our work proposes a new computational method that will provide quantitative insight to the interaction between certain antibiotics and pathogens," Tam said in a statement. He added that the approach models the interaction between an antimicrobial agent and a pathogen in order to "both help develop new antibiotics and optimize existing medications to curb the prevalence of drug-resistant bacteria."

The three-year research project was initially funded by a $400,000 grant from the National Science Foundation.

In the grant abstract, the researchers wrote that they aimed to "develop and validate in vitro mathematical modeling tools that can guide the testing of dosing regimens" noting that the framework utilizes concepts that haven't yet been employed in drug research.

A possible rationale, as Tam explained to BioInform, stems "from the fact that mathematical modeling is not well integrated in mainstream biological science," and, as such, it isn't "routinely done in drug development [studies]." He noted, however, that modeling has gained some "attention" in recent years as a tool that could help researchers drill down to a short list of drugs out of thousands of potential candidates. Nevertheless, to date most modeling approaches have not taken dosing information into account.

In the paper, the authors wrote that their computerized approach is a step beyond randomly selecting variables such as treatment frequencies and drug concentrations, which hitherto has been the norm in these types of studies. The method uses a "dynamic mathematical model ... to derive the dosing intensity," thus providing a more "objective comparison of various dosing regimens."

"Our approach gives us the ability to take extra variables into consideration in an attempt to develop a more robust computational tool that covers a wider spectrum of relevant scenarios in new drug development," Nikolaou said in a statement. "Some pharmaceutical companies are following our developments closely, and we are in the process of refining a model prototype in the form of a computer program to ultimately be used in a clinical setting."

Tam declined to provide further details on the identities of the pharma companies but he did say that they are "major international pharmaceutical companies with established programs to develop new antibiotics."

Furthermore, while he couldn't give a specific timeline, Tam said the team hopes to have a clinical prototype in a few years.

Optimized Dosing

The model predicts how bacterial populations might respond to multiple drug exposures to get a better sense of how much medicine a patient should take, how often medication should be taken, and for how long.

"Ideally you want to have all the [drug] combinations to find the best one, but you don’t have enough time or resources," Tam said. "The system essentially ... captures key features of the interaction between a drug candidate and the effect that it exerts on target bacteria."

The process begins with experiments to determine how target bacteria respond to different concentration of specific antibiotics, Tam explained. "We expose [the bacteria] at fixed but escalating [drug] concentrations, and then we would track the bacterial response like how fast are they being killed, or how fast are they growing ... typically [over] 24 hours."

In the paper, the authors created a daily dosage regimen using 6000 mg of a hypothetical drug administered all at once; in two doses of 3000 mg each; or four doses of 1500 mg.

Once they had the experimental results of the bacterial response to the antibiotics, "we put this information into the system and [used] computer simulations [to suggest] different scenarios," Tam said. "So if we were to give this dose, this is the predicted outcome and if we were to give that dose, this is the predicted outcome."

To create the predictions, Tam said that the researchers used several software packages including Mathematica, Matlab, and ADAPT II, along with customized code that incorporated parameters such as normal bacterial growth rates, bacterial mortality in the presence of varying antibiotic concentration, and dosing interval. It also takes into account bacterial adaptation in response to drugs and the host's response to the antibiotic.

Some results from the simulation showed that in one scenario, the dosing frequency — administering a higher concentration at once or in smaller quantities — did not have much of an impact on the bacteria, as long as the dose remained constant. In another scenario, bacteria died when they were exposed to drug concentrations above a specific threshold for a prolonged period of time.

Based on their findings, the researchers concluded that "an antimicrobial agent can be shown to exhibit both concentration and time-dependent killing, but the proposed model was flexible enough to describe different distinct pharmacodynamic profiles."

Once they complete the simulations, Tam said the team validates its findings in further lab tests.

Although the researchers used a hypothetical treatment schedule to illustrate how the system works, in the grant, they wrote that they planned to "test the effect of three representative antibiotics" on bacteria.

The drugs listed in the grant include meropenem; which is marketed by AstraZeneca under the brand names Merrem and Meronem; levofloxacin, which is marketed by Sanofi-Aventis and Ortho-McNeil-Janssen Pharmaceuticals under the brand names Tavnic and Levaquin respectively; and tobramycin.

Tam said the method will be made available to drug developers either as a software package or as web-based tool and that the university holds a provisional patent on the technology.

While the method was developed to study antibiotics, the researchers noted that it can be applied to investigate effects of antifungals and antivirals as well as the effect of chemotherapy on cancerous cell populations.

The scientists concede that their method, while it encompasses several important variables, doesn't take into account information such as the immune system's role in interactions with resistant bacteria, though they are working on enabling the models to incorporate this data.

Meanwhile, Tam's team plans to create models to predict the interactions of multiple drugs on pathogens, which could be useful in selecting combination treatments for diseases like tuberculosis and HIV. He also said that they hope to create a system that clinicians can use to make therapy decisions at the bedside, although he said, that system is still some years away.

Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.

The Scan

Study Finds Few FDA Post-Market Regulatory Actions Backed by Research, Public Assessments

A Yale University-led team examines in The BMJ safety signals from the US FDA Adverse Event Reporting System and whether they led to regulatory action.

Duke University Team Develops Programmable RNA Tool for Cell Editing

Researchers have developed an RNA-based editing tool that can target specific cells, as they describe in Nature.

Novel Gene Editing Approach for Treating Cystic Fibrosis

Researchers in Science Advances report on their development of a non-nuclease-based gene editing approach they hope to apply to treat cystic fibrosis.

Study Tracks Responses in Patients Pursuing Polygenic Risk Score Profiling

Using interviews, researchers in the European Journal of Human Genetics qualitatively assess individuals' motivations for, and experiences with, direct-to-consumer polygenic risk score testing.