New approach gives better explanation for variation in medication effect between individuals

22 July 2021• NEWSITEM

How quickly or slowly your body processes a drug is genetically determined. Therefore, your genetic information can be used to predict how you will respond to a drug, and adjust the dose accordingly. In a study published in Science Translational Medicine, LUMC researchers showed that this prediction can be made much more accurate by combining sequencing with artificial intelligence. In the future, this could lead to better dose recommendation per patient, and possibly fewer side effects.

"People react differently to drugs," says PhD student Maaike van der Lee of the Department of Clinical Pharmacy and Toxicology. "This variation in drug effect is caused by differences in our genes. As a result, a certain enzyme may work slower than average, causing a drug to remain in the body longer. In this case, the dose of the drug should be reduced to avoid side effects and improve its effect." 

Missing explanation

An enzyme involved in processing a quarter of commonly prescribed drugs is CYP2D6. The cause of the difference in activity of this enzyme between individuals is 90% genetically and 10% depends on environmental factors. "But with the current methods that predict the enzyme activity of CYP2D6 per individual, only part of the genetic cause can be explained. We call this missing heritability. That is why we started looking for a different approach," says Van der Lee.

One of the shortcomings of the current methodology is that not all variations in the gene coding for CYP2D6 are included. "In addition, enzyme activity is currently predicted using four categories: poor metabolizers, intermediate metabolizers, normal metabolizers, ultrarapid metabolizers. However, enzyme activity is not categorical, but continuous", Van der Lee explains. 

Substantial increase

The researchers have partnered with the departments of Human Genetics and Clinical Genetics to use long-read sequencing that detects all genetic variations. "We have combined this technique with artificial intelligence to map the whole gene and predict its activity on a continuous scale." 

The researchers tested this new approach on a group of over 500 breast cancer patients using tamoxifen, a drug that is converted in the body by CYP2D6. "With the current categorical method we could explain 54% of the difference in the enzyme activity of CYP2D6 in these women, with the new continuous scale this was 79%. A considerable increase," says Van der Lee. 

New standard?

It is the first time that this new approach has been used. The next step is to see if the researchers can also improve the prediction of the activity of other enzymes. "If that is the case, we can look into how this can best be applied in patients," says Van der Lee. In any case, the researcher is hopeful that this new approach will prove its worth. "I dare say that this will become the standard in the future. It would allow us to more accurately predict how an individual processes a certain drug and therefore also which dose is best. We could then prevent even more side effects and improve efficacy." 

The LUMC researchers collaborated with the University of Leuven, the Catharina Hospital Eindhoven and the Karolinska Institutet. For more information, read the article in Science Translational Medicine.

Stay informed on the latest news of the Leiden University Medical Center, and sign up for our biweekly newsletter or subscribe to our LUMC Magazine.