Researchers Develop Model to Predict Side Effects of Drugs

UCSD researchers recently designed a model that uses red blood cells to predict the side effects individual patients will experience in response to specific drugs. The Systems Biology Research Group published its study on Oct. 28 in the journal Cell Systems.

Galletti Professor of Bioengineering at the Jacobs School of Engineering, Adjunct Professor of Medicine and principal investigator Bernhard Palsson explained how side effects are unique to the individual.

“We’re not just interested in predicting the efficacy of a drug, but its side effects as well,” Palsson said to the Jacobs School of Engineering. “Side effects are very personalized. Two different people can take the same drug, but one person might experience side effects while the other doesn’t.”

Lead author of the study Aarash Bordbar described how the side effects of drugs can have an impact on national death rates and the economy.

“Over 100,000 Americans die each year because of drug side effects,” Bordbar told the UCSD Guardian. “It’s an annual $140 billion healthcare issue in the United States.”

The kinetic model determines how different people will respond to a drug treatment through analysis of over 100 metabolite measurements, including sugars and amino acids. This data is then integrated with a network model, allowing researchers to simulate different conditions.

Personalized whole-cell kinetic models of red blood cell, or erythrocyte, metabolisms were constructed using 24 healthy individuals based on fasting-state plasma, erythrocyte metabolomics and whole-genome genotyping, according to the article in Cell Systems. The model simulations identified inosine triphosphatase deficiency as a genetic variation that may protect against ribavirin-induced anemia. With this information, researchers better understand why approximately 8 to 10 percent of patients taking ribavirin, a drug treatment for Hepatitis C, experience anemia.

Bordbar explained that using red blood cells was logistically convenient for this research as well as in the clinical field.

“We started with the red blood cell because it’s easy to get blood from people,” Bordbar told the Guardian. “If future diagnostics were made, you would want to do a blood test rather than something like a biopsy for cell tissue.”

In addition to creating personalized kinetic models of erythrocytes using metabolomics data, the research concluded that kinetic parameters, rather than metabolite levels, better represent genotypes, and individual differences in dynamics occur on physiologically relevant timescales, as stated in the Cell Systems article.

Visiting scholar Neema Jamshidi discussed how the kinetic models may impact personalized medicine.

“An application of these models, with which we can screen for potential susceptibility for side effects of drugs, is significant because the cost of side effects is [also] significant,” Jamshidi told the Guardian. “If you can predict that a patient will experience a side effect, it becomes a risk-benefit question.”

While previous studies used larger sample sizes, this research integrates a greater number of measurements from individual subjects. Jamshidi explained that future studies involving focused samples will provide better information for this type of research.

“Studies where the focus is on a 100,000 or a million people are not as valuable [for our purposes] as identifying a smaller-but-targeted cohort with specific measurements and specific genes of interest and then testing those in an objective manner,” Jamshidi said.

The researchers expect to conduct further experiments on commonly-used drugs that are known to cause hemolytic anemia. They are also looking to develop predictive models for platelet cells and ultimately a liver-cell model.