Butterfly Effect May Foil Genomic Medicine
The so-called butterfly effect, in which a small creature can cause something on the scale of an earthquake merely by flapping its wings, is fodder for debate on whether the digital revolution in medicine can deliver on its promise for precision medicine.
Larry Norton, MD
The so-called butterfly effect, in which a small creature can cause something on the scale of an earthquake merely by flapping its wings, is fodder for debate on whether the digital revolution in medicine can deliver on its promise for precision medicine, Larry Norton, MD, said in a presentation on genomic medicine and the doctor-patient relationship at the Miami Breast Cancer Conference®.
Norton, of Memorial Sloan Kettering Cancer Center, said that growing public faith in the power of software and precision medicine to solve the problem of cancer should be tempered by a respect for randomness in the way genes and environmental factors create unexpected outcomes. More and more human activity involves digital information, and doctors are missing important signals that can only be revealed by direct interaction with patients, he added.
“I’m not a nihilist, but public expectation for precision medicine is extraordinarily high,” Norton said. “Precision medicine is essentially the belief that molecular biology and a person’s molecular profile can be used to determine the optimal treatment.” He pondered whether that is a truly rational expectation given what great mathematical theorists have postulated about cause and effect and what we know about chance and inherited characteristics.
Isaac Newton, he said, believed strongly in the power of predictive science, if variables such as planetary motion and position could be determined. Henri Poincaré stated that prediction could be impossible if too many variables were introduced, and Edward Norton Lorenz introduced the concept of minor forces being able to wield tremendous influence on events—the precursor to the notion of the butterfly effect. Similarly, he said, “since we’re dealing with gene networks that are more than 2 genes, it’s very unlikely that these gene networks will be completely predictable.”
He explained how 2 genomically equal mice, 1 force-fed and obese and the other raised on a normal diet, could have offspring that differ in their ability to maintain normal weight. “The offspring of the force-fed mouse, who’s then fed a healthy diet and does not remain obese, will still have a greater likelihood of having a metabolic syndrome,” Norton said.
Similarly, “you can have a genetic abnormality in a parent that is not passed down, but the trait is passed down,” he said, giving examples of variations of oncogenes and modifying genes in fruit flies and their progeny. “Without knowledge of the parental generation, you would not be able to predict whether the fruit fly would have a tumor or not, or the size of the tumor, because you wouldn’t know what the gene expression pattern was like in the parental line.”
Returning to his thesis point, Norton said, “If it’s not logical mathematically and if the laboratory data is not in support of it, why is it in the zeitgeist that precision medicine is going to work?” He said that the digital age has deeply affected expectations in the business and scientific community. For example, IBM’s Watson for Oncology computerized treatment aid is heavily dependent on this type of thinking, he said.
In addition, doctors now look at their electronic health records 30% of the time they’re with patients, which means that one-third of the time patients’ visual cues are not being noticed, he said.
“If indeed we don’t take into account the parental environment—the parental genes—if we don’t take into account the human interactions, we’re going to make mistakes, and reliance on molecular data only can lead us astray. I think this explains a lot of the issues we’re coming into as we try to apply realistically molecular data to precision medicine,” he said.
<<<
