UPDATED March 8, 2018 // Body mass index (BMI) and overall cognition are major contributors to the variation in brain structure and function seen in patients with psychosis, a new study suggests.
The findings highlight the importance of including these key factors in future research and have important implications for clinical practice, author Sophia Frangou, MD, PhD, professor of psychiatry, Icahn School of Medicine at Mount Sinai in New York City, told Medscape Medical News.
“There are modifiable factors that can improve brain health in patients in the same way as in the healthy population, and the two that can be implemented sort of immediately are more emphasis on BMI and physical health, and increased availability of cognitive training.”
The study was published online March 7 in JAMA Psychiatry.
Impact of BMI on the Brain
Studies suggest that schizophrenia and bipolar disorder are associated with abnormalities in cortical thickness, subcortical volumes, and white matter integrity.
In patients with these conditions, patterns of brain activation are altered. These alterations commonly lead to inefficient engagement of regulatory cortical regions.
The research also indicates that the intrinsic functional architecture of the brain is altered in patients with psychotic disorders. Hypoconnectivity within and between resting-state networks are the most commonly observed alterations.
These neuroimaging findings have been linked to clinical variables, including disease severity. However, the investigators note, the MRI signal can be influenced by lifestyle choices, such as smoking and physical activity; by general health, as indicated, for example, by BMI; and by exposure to antipsychotic medications.
In this new study, said Frangou, they wanted to address an ongoing debate within the neuroimaging community by determining whether differences between patients with psychosis and healthy control persons are due to the disease or to other variables.
“If, say, BMI had a very significant contribution to a brain imaging signal, and all your patients were fat and all your controls were thin, then perhaps your case-control differences were due to the fact that you were comparing thin and fat people rather than patients with disorder X and controls,” said Frangou.
The investigators also wanted to determine whether there is still a signal associated with disease after all the factors that could bias case-control differences were accounted for.
“It’s possible that there are brain differences that are accounted for by symptoms, but they might be completely obscured by differences that are produced by all the other variables,” said Frangou.
Another aim for the study, she said, was to learn whether different variables were differentially associated with the brain signal in patients with psychosis compared to control persons.
“For example, we wanted to know if BMI has exactly the same effect in the brain of someone with schizophrenia as opposed to someone who is completely healthy.”
The study included 92 patients with schizophrenia (25% women; mean age, 27 years), 37 patients with bipolar disorder (32.4% women; mean age, 27.5 years), and 48 healthy volunteers (41.7% women; mean age, 29.8 years).
The researchers collected multimodal MRI imaging data and derived measures of cortical thickness, subcortical volume, task-related brain activation (during working memory and emotional recognition), resting-state functional connectivity, and white matter fractional anisotropy (FA).
They also collected information on nonimaging factors known to be associated with the MRI signal. The nonimaging data set included demographic information, information on personal and family psychopathology, IQ, history of substance use, history of psychological trauma, degree of physical activity, BMI, and use of medications.
In their statistical analysis, the researchers used sparse canonical correlation analysis. This powerful multivariate method allowed them to go beyond simple correlations and to take into consideration the entire context of associations between brain signals and other factors. It also allowed them to extract the weight of the correlation.
On a scatterplot of the imaging and nonimaging data sets, the overall correlation was r = 0.63 (P < .001). Correlations were similar for patients with bipolar disorder, those with schizophrenia, and healthy volunteers.
“The positive and negative human traits really affect the brain in the way we expect them to, and this association is the same both in patients and in healthy individuals,” said Frangou.
Among the nonimaging variables, age (r = −0.53), IQ (r = 0.36), and BMI (r = −0.25) were associated with multiple imaging phenotypes.
The analysis found that BMI “had one of the highest weights in the correlation,” noted Frangou. After increased age, “it makes the most significant negative contribution to brain health,” she said.
“This suggests that weight gain in patients, in addition to all the other problems they have, contributes to a reduction in brain health and therefore should be a priority,” particularly because interventions may reduce risks, said Frangou.
Overall cognition also had large positive associations with imaging measures.
“This is a problem for patients with psychosis, but it’s a problem that can be remedied” through cognitive training programs and through physical activity regimes, said Frangou.
“One thing this analysis suggests is that maintaining general cognitive abilities should be part of the care of patients with psychosis,” she said.
The findings are in keeping with a number of initiatives, including the American Psychiatric Association’s “collaborative care” model, which focuses on integration of mental health and physical health services, said Frangou.
She’s a proponent of cognitive training programs but believes some experts have “excessive expectations” as to what such programs can accomplish. For example, training might boost working memory, but it’s not going to guarantee a job for an unemployed patient with schizophrenia.
To skeptics who say that the benefits of cognitive training aren’t maintained if the training ceases, Frangou asked whether the benefits of any exercise regime are maintained if discontinued.
The new analysis also showed that cannabis use (r = 0.23) and other substance use (r = 0.33) were associated with differences in subcortical volumes. These results “align with” reports of a modest increase in subcortical volumes in substance users, particularly in the basal ganglia, say the authors.
Alcohol use was negatively correlated with the FA variate. This, say the authors, adds to previous observations that even light recreational use of alcohol may affect white matter microstructure.
Within the multivariate models, associations with positive symptoms were retained with global neuroimaging findings (r = -0.13), cortical thickness (r = -0.22), and task-related activation variates (r = -0.18).
The neuroimaging literature has mainly supported an association between positive symptoms and cortical thinning, note the authors.
“The present study extends these findings to suggest that positive symptoms have multimodal associations with neuroimaging phenotypes that relate both to structure and function,” the authors write.
Impact of Negative Symptoms
The new study also showed that negative symptoms were mostly associated with measures of subcortical volume (r = 0.23). This positive association was unexpected because “it goes in a different direction” than other findings, said Frangou.
“Within the multivariate model, positive symptoms had a bad effect on all brain imaging — on cortical thickness, on task-related activation — so we were expecting that negative symptoms would go in the same direction, but they didn’t,” she said.
It’s not clear why this is so, she said.
The current analysis also showed that depression or anxiety was associated with measures of white matter integrity (r = 0.12).
This finding conforms to results of a large diffusion-weighted imaging study on psychosis in which a positive association was found between depressive symptoms and FA of the thalamic radiation and of the corpus callosum, the authors note.
Although antipsychotic drug dosage had a negative effect on some brain imaging measures, Frangou stressed that this should not be a main focus, because it’s not something that can be changed, like BMI and one’s level of physical activity can be changed.
“There are other things you can target to improve brain health,” she said.
“Nobody is saying we should not be looking for better treatments for the brain — absolutely we should — but I would not want the message to be, ‘Stop taking an antipsychotic drug.’ ”
The new results suggest that even after considering all the different influences on the brain, there’s still a negative association between the brain and symptoms, said Frangou.
This, she added, helps clear up the discussion of whether the associations are due to “artifacts.”
Important Take-home Message
Commenting on the findings for Medscape Medical News, Gary Donohoe, PhD, professor and chair of psychology, National University of Ireland, Galway, said that although the study was quite small, it was “well done.”
His research focuses on how brain structure and function are affected by risk factors for schizophrenia. His research group also investigates psychological therapies for psychosis, with a particular interest in remediation of cognitive deficits in schizophrenia.
“What’s interesting about this study is that the authors have managed to pool a lot of data, and in pooling it and doing a sophisticated statistical analysis, they have come up with a take-home message that’s really important, which is that mental health disorders are like every other aspect of health. They require us to make sure that people are living healthy lifestyles.”
Donohoe noted that in the study, the patients with schizophrenia had been diagnosed with that condition within the past 5 to 10 years but were already much heavier (mean BMI, 27.2) than persons in the other two groups of participants, and they smoked more and tended to be sedentary.
“Along with other good medical interventions and cognitive and psychological interventions, it really sounds like we should be investing a lot of time on good, solid public health indicators, like getting people out exercising and really watching their diet,” he said.
Patients tend to gain weight when taking a variety of antipsychotic medications, but Donohoe pointed out that the patients with psychosis in the study could not have been taking such medications for long.
“People in this study were not in their forties and fifties who have been on medication for years and years; these were people in their twenties on average, so I don’t think that the medication is the only factor that matters here. I think lifestyle is probably equally likely to matter,” he said.
Men tend to be diagnosed with schizophrenia at about age 18 years; women tend to be diagnosed at about age 20 years, he said.
Donohoe said use of IQ as a global cognition indicator was appropriate.
“The standout cognitive problem in schizophrenia is memory function, but memory itself is highly correlated with IQ,” he said.
He noted that in the study, the mean IQ for patients with schizophrenia was 93.5, which is two thirds of a standard deviation below the average. Patients with schizophrenia have a lower IQ because of their disease, he said.
Although impressed overall by the research, Donohoe stressed that it’s a correlational study that involved people at one point in time.
“This is an attempt to take all of the MRI variables that we typically look at, and take all of the lifestyle variables that we typically look at, and see how they correlate. And they correlate very well, but that doesn’t say a lot about causation,” he noted.
Melanie Wall, PhD, director of mental health data science in the Department of Psychiatry at Columbia University, New York City, also commented on the article for Medscape Medical News.
This methodologic article is a “useful and clear reminder” of the importance of trying to control for possible confounding variables, such as intelligence, BMI, substance use, and age, in nonexperimental studies such as imaging studies, said Wall.