Introduction

Dynamic contrast-enhanced MRI (DCE-MRI) has been suggested as an effective tool for assessing blood-brain barrier (BBB) dysfunction in many brain disorders including epilepsy, tumors, stroke, traumatic brain injury and dementia. While BBB leakage is hypothesized to underlie the neuropsychiatric complications of these disorders¹, the link between BBB leakage and neuropsychiatric symptoms remains unclear. Here we set out to provide the first characterization of BBB permeability in bipolar depression, and to examine whether the affected brain regions reflect clinical symptomatology.

Methods

We studied the BBB of 36 bipolar patients and 14 controls (matched for sex and age), using dynamic contrast-enhanced MRI. Participants were intravenously injected with the magnetic contrast agent gadoteridol (0.1 mmol/kg, ProHance, Bracco Imaging Canada, Montreal, QC), and its dynamics in the brain were monitored for a period of 20 minutes using T1-weighted MRI (3T GE Discovery MR750). All participants also underwent diagnostic interviews, cognitive testing, anthropometric measurements and blood sampling. MRI analysis was performed as previously described^2–4. In brief, the accumulation rate of the contrast agent during the slow enhancement period of the scan (6-20 min) was derived for each voxel as a measure of BBB leakage. To identify voxels with elevated versus nominal permeability, an intensity threshold was applied (established as the 95th percentile of values in a previously scanned cohort of control subjects⁵). The percent of voxels with elevated permeability was used to reflect the extent of BBB leakage in each subject. Blinded clustering (K-means) was used to identify subjects with extensive BBB leakage. Generalized linear model-based feed-forward selection was applied to identify brain regions most predictive of depression severity. Classification accuracy was assessed using receiver operating characteristic (ROC) analysis. The Chi square test and Wilcoxon rank sum test were used for statistical comparisons of categorical and continuous clinical parameters, respectively.

Results

We identified extensive BBB leakage in 28% of bipolar patients and zero age/sex-matched controls (P<0.05, Chi square). Notably, all bipolar patients with extensive leakage also had insulin resistance and worse psychiatric, cognitive and functional outcomes (p<0.05). We found depression to be associated with region-specific BBB leakage, with the nucleus accumbens best predicting depression severity.

Discussion

Our findings highlight BBB damage as a mechanism contributing to the dysfunction of key brain regions associated with depression. Our results further suggest that insulin resistance increases the risk of extensive BBB leakage and the associated exacerbation of neuropsychiatric symptoms.

Conclusion

The described imaging approach is the first to identify a specific brain pathology in bipolar patients, and may therefore allow biomarker-based distinction between mood disorders. Moreover, the approach allows quantitative and objective diagnosis of region-specific BBB dysfunction, and the study of associated clinical manifestations. We thus assert that our approach may be useful for the diagnosis, prognosis and follow-up of mood disorders, potentially leading to more informed treatment decisions.

Acknowledgements

This study was supported by the European Union’s Seventh Framework Program (FP7/EPITARGET), the Nova Scotia Health Research Foundation (NSHRF), Canadian Institute for Health Research (CIHR), and the Brain & Behavior Research Foundation (NARSAD).