The severity of symptoms in posttraumatic stress disorder (PTSD) varies greatly across individuals in the first year after trauma and it remains difficult to predict whether someone might worsen, improve, or recover.
In a new study, Yale researchers used brain imaging and machine learning to build models that were able to predict the severity of PTSD symptoms, in both the short and long term, in people who experienced traumatic events. The findings, researchers say, provide a deeper understanding of PTSD and the brain and could yield new targets for treatment in the future.
The study was published March 10 in the journal JAMA Network Open.
While much of what’s known about the neurological underpinnings of PTSD has come from research on specific areas of the brain, researchers have increasingly discovered that they will have to widen their lens.
“When it comes to PTSD, often researchers will study the amygdala, known as the brain’s fear center, and the hippocampus, which is involved in memory processes,” said Ziv Ben-Zion, a postdoctoral fellow at Yale School of Medicine (YSM), who is co-lead author of the new study along with Alexander Simon, a doctoral student at YSM. “But as research has advanced, we’ve come to understand that in psychiatric disorders, alterations happen in large-scale brain networks, not only in isolated regions.”
To capture this broader perspective, Ben-Zion and his colleagues used functional magnetic resonance imaging (fMRI) to examine whole-brain functional connectivity patterns.
For the study, they looked specifically at adults who had been admitted to a general hospital’s emergency department after experiencing a traumatic event, such as a car accident, assault, or robbery. Each of the 162 participants underwent fMRI scans one month after their experience; these scans were administered while individuals were at rest and as they completed tasks that assessed emotional reactivity and sensitivity to risk and reward. Participants also underwent clinical assessments for PTSD severity at one month, six months, and 14 months after their traumatic event.
After training a machine learning model on the patients’ functional brain images and clinical assessments, the researchers found that the model was able to predict an individual’s PTSD symptom severity at one month and 14 months post-trauma. But the model couldn’t make strong predictions about symptom severity at the six-month time point, which, the researchers say, was likely due to the condition not being stable yet and there still being significant variability in symptoms among individuals.
The model also identified which brain networks were most important for the predictions at each time point and which types of symptoms were most strongly predicted. For example, at one month after trauma, the model accurately predicted symptoms of avoidance and negative changes in mood and cognition, while at 14 months post-trauma it better predicted intrusion and hyperarousal symptoms.
“The motivation behind the study was to see if we could identify something in the brain early on that could help us predict who will develop more severe symptoms over time and who will recover,” said Ben-Zion. “Our findings show that early brain connectivity can predict symptom trajectory, which will be useful for diagnosis and prediction of PTSD, and hopefully even treatment.”
Ben-Zion works in the labs of Ifat Levy, the Elizabeth Mears and House Jameson Professor of Comparative Medicine, and Ilan Harpaz-Rotem, professor of psychiatry, both of YSM. Harpaz-Rotem served as co-senior author of the study along with Dustin Scheinost, associate professor of radiology and biomedical imaging at YSM.
While some of the brain networks identified by the model aligned with previous research on PTSD, two — the visual and motor sensory networks — were a bit of a surprise, the researchers say. They suspect they may be related to flashbacks, a known symptom of PTSD.
The findings also highlight how much change occurs over time with PTSD. The brain networks and symptom types strongest at one month post-trauma were different than those most pronounced at 14 months.
Overall, the findings could bring more clarity to how to best manage PTSD.
“If early brain connectivity at one month after trauma can give us some idea about who will develop more severe symptoms, it could help us establish more objective measurements for diagnosis,” said Ben-Zion. “And while more research needs to be done, maybe in the future there will be a way to target the specific brain networks implicated in PTSD.”
