by Paul Govern
With the aid of a four-year, $3.4 million grant from the National Institute of Mental Health, researchers at Vanderbilt University Medical Center and Massachusetts General Hospital (MGH) will apply new techniques to investigate treatment resistance of two devastating mental disorders — major depressive disorder, which befalls 15 percent of people at some point in their lives, and schizophrenia, which affects approximately 1 percent.
The award is technically two grants, with funds split evenly between VUMC and MGH.
“There are effective drugs available for these disorders, but across both patient groups about one-third don’t respond and many more suffer for years before finding an effective treatment plan,” said a co-principal investigator for the study, geneticist Douglas Ruderfer, PhD, assistant professor of Medicine, Psychiatry and Behavioral Sciences, and Biomedical Informatics. “Previous work has provided evidence that this nonresponse is determined to some degree by genetic variation, but we lack specific genetic markers that could inform and improve treatment.”
Results of the study could potentially give health care teams a window into individual risk for nonresponse to initial treatments for these devastating illnesses, said co-principal investigator Colin Walsh, MD, MA, assistant professor of Biomedical Informatics, Medicine, and Psychiatry and Behavioral Sciences.
“We might be able to offer more advanced therapies to those who need them earlier in the course of treatment to avoid ineffective first-line medications, their side effects, worsening symptoms, unneeded hospitalization, and more.”
The two VUMC investigators have pioneered a new approach to power the search for genetic causes of conditions and illnesses. Using machine learning techniques and retrospective de-identified electronic health record (EHR) data, they’ll first build separate predictive models of treatment resistance. In depression, they will predict likelihood of receiving electroconvulsive therapy, which is only given after nonresponse to initial therapies. In schizophrenia, they will predict those who will end up receiving clozapine, an antipsychotic medication given only after nonresponse to other therapies.
Next, they’ll use these models to assign a probability of being treatment resistant to thousands of genotyped patients with the respective disorders represented in biobanks maintained by VUMC and MGH. Finally, they’ll look for genetic patterns associated with these treatment resistance probabilities.
Walsh and Ruderfer first demonstrated the approach in a suicide attempt risk study published in January.
“We’re aiming to build predictive EHR risk models for nonresponse that might themselves prove beneficial in clinical practice. Further, we’re hoping the substantially increased power from our approach will result in finding treatment-relevant genetic variants, which to this point has been a challenge for the field,” Ruderfer said.
The study is funded by the National Institutes of Health (MH116269).
