Brain Abnormalities May Be Biomarkers for Bipolar Disorder

Researchers from the University of Texas in Houston and the Federal University of Rio Grande do Sul in Brazil have used brain abnormalities to detect bipolar disorder, according to research published in Biological Psychiatry: Cognitive Neuroscience and Neuroimaging.

Extensive neuroimaging research in recent decades has found structural brain abnormalities in patients with bipolar disorder (BD), including reduced gray matter density in certain areas and decreased white matter density in other areas. However, such observations “have not been translated into objective and clinically useful biomarkers,” wrote Benson Mwangi, PhD, from the Department of Psychiatry and Behavioral Sciences at the University of Texas Health Science Center and colleagues.

The researchers explored the use of neuroimaging combined with a machine learning algorithm to potentially distinguish between individuals with BD and healthy controls, identify clinically relevant brain pathways, and stratify patients based on disease burden.

 

Using structural magnetic resonance imaging (MRI) in 128 patients with BD and 128 demographically matched healthy controls, the researchers obtained density maps of gray and white matter that were used to “train” the algorithm that would distinguish individuals with BD from controls.

Patients were further divided by clinical stage: the early-stage bipolar disorder I (BDI) subgroup included patients who reported less than 3 lifetime manic episodes without hospitalization; patients with 10 lifetime manic episodes with hospitalizations comprised the late-stage BDI subgroup; and the late-stage BDI subgroup included all remaining BDI patients. Because bipolar disorder II (BDII) is characterized by hypomanic rather than manic episodes, patients with this type of BD were not stratified to a clinical stage and instead were assigned to a BDII subgroup.

Results show that the algorithm distinguished patients with BD from controls with 70.3% and 64.9% accuracy using white matter density data and gray matter density data, respectively. The algorithm trained with combined white and gray matter data distinguished patients with BD from controls with 64% accuracy.

“Multiple brain regions, largely covering the fronto-limbic system, were identified as ‘most relevant’ in distinguishing both groups,” the authors reported.

The most relevant white matter brain regions that distinguished patients with BD from controls were the cerebellum, brainstem, cingulated gyrus, and corpus callosum. The most relevant gray matter regions in differentiating patients with BD from controls were the orbitofrontal cortex, superior frontal gyrus, temporal lobes, and midbrain. These observed reductions in gray matter and white matter are in line with previous research findings. 

While there was not a significant difference in predicted probability scores between controls and patients with early-stage BDI, scores did differ significantly between controls and patients with intermediate-stage and late-stage BDI; patients with high probability scores according to the algorithm belonged to the late-stage BDI subgroup.

These findings suggest that the algorithm was able to distinguish between healthy controls and the latter 2 patient groups, but not between controls and patients with early-stage BDI. Additionally, a significant difference was observed in cerebellar white matter density between controls and intermediate- and late-stage BDI patients. 

These results point to neuroprogression in BD, possibly due to “increased microglia activation that leads to the release of inflammatory markers potentially translating into tissue loss,” the authors wrote.

The current “findings on neuroprogression in patients with BD need to be confirmed using a longitudinal design, which may give more insights compared with a cross-section design,” they concluded.

Reference

Mwangi B, Wu MJ, Cao B, et al. Individualized prediction and clinical staging of bipolar disorders using neuroanatomical biomarkers. Biol Psychiatry Cogn Neurosci Neuroimaging. 2016;1(2):186-194.