We are bombarded with advertisements from drug companies wherein they warn us to see our doctor immediately if we experience any of the enumerated serious side effects. Predicting, and hopefully trying to avoid, such serious problems is a scientific challenge involving risk not only to test animals, but to human beings. A team of scientists is creating a computer simulation model that could reduce the threats posed by pharmaceutical drugs to living beings.
Taking 2005 data, scientists created a network linking 809 medications to 852 known side effects. They added in chemical properties of the drug, such as its molecular weight and melting point, plus what part of the body the drug acts upon. Then, they programmed the computer to predict new connections between drugs and side effects that were likely to occur post market – after days and even months of an individual taking the drug.
The need to use technology instead of human guinea pigs was highlighted by a Stanford University biomedical informatics specialist from outside the study. Russ Altman noted that prerelease testing proves that a drug is effective, but doesn?t prove that it is not dangerous. Many side effects don?t show up until the drug is in your medicine cabinet. Headline-catching injuries and deaths reach the thousands each year from undiscovered, adverse side effects. The program developed by the research team picked up on the potential for suicidal thoughts related to a seizure drug marketed as Zonogran (zonisamide) and heart attacks linked to the diabetes drug Avandia (Rosiglitazone Maleate).
The team consisting of Aurel Cami, Alana Arnold, Shannon Manzi and Ben Reis from Harvard Medical School and Children?s Hospital Boston applied network math in their project. Typically used for assessing social relationships or how a disease spreads, the method was able to reveal areas of concern in drug / side effect relationships. Their trained, logistic regression model predicted adverse drug related events that were not known when the 2005 drug safety database was compiled. The computer?s results compared favorably to the same drug safety database that had been updated with real-life experiences in the five years leading up to 2010.
Does a lovely icon lull you into believing a drug is safe?
The side effects in the data base included common reactions such as nausea and vomiting to more serious reactions such as circulatory collapse and shock. Drugs in the study included such tongue twisters as pantoprazole and eszopiclone, better known as Lunestra which is marketed with the pretty icon that soothes us into lowering our defenses.
The scientists concluded in a Science Translational Medicine article that the computer-based predictive network methods might be useful for predicting previously unknown adverse drug events. That would be good news for lab rats and people alike. The team?s next target is drug-to-drug interactions which are rarely analyzed in drug trials.