Development and Validation of a Prognostic Neuroimaging Biomarker for Depression

Lay Summary

MRI imaging may yield biomarkers for diagnosing and optimally treating major depressive disorder

Researchers will use functional magnetic resonance imaging (fMRI) techniques in patients with major depressive disorder (MDD) to try to identify atypical connectivity patterns in the brain that are biomarkers for diagnosing MDD and for predicting patients’ treatment responses and long-term prognoses.

MDD patients are highly heterogeneous. They have varied clinical symptoms and responses to treatment. In fact, only about a third of MDD patients who receive anti-depressants fully recover within the first three months, and many of these patients will experience repeated episodes and require long-term treatment. There currently is no way to predict which individual patients will respond acutely to treatment, and which of these patients will subsequently require long-term therapy. Instead, many treatment decisions are currently trial and error.

Several studies using “resting state” fMRI (rsfMRI, where participants are left to their own thoughts and undertake no tasks while being scanned to visualize the functioning of brain circuits) implicate abnormal functional connectivity patterns in particular brain circuits in MDD patients. Yet patterns differ among patients, and some patterns overlap with those seen in healthy controls. MDD, therefore, may reflect a syndrome that encompasses several different patterns of brain dysfunction, where each pattern pertains to a specific MDD subgroup. If this is the case, biomarkers that differentiate specific subgroups of patients may be more effective for informing individual diagnoses and identifying which patients are most likely to respond to which antidepressants. They also have the potential to accelerate clinical research by enabling investigators to quantify antidepressant responses in homogenous subgroups of patients.

Identifying homogenous subgroups, however, requires that large numbers of MDD patients and healthy volunteers be studied, in order to define subgroups of patients who have similar patterns of atypical connectivity. The Cornell researchers are in the process of doing just that. With collaborators at other institutions, they have amassed a large, multi-site data set of rsfMRI scans from MDD patients and matched controls. The Cornell researchers hypothesize that depressive symptoms and recurrence risk arise from distinct patterns of abnormal connectivity in limbic brain circuits. Additionally they hypothesize that the unique patterns detected by rsfMRI can serve as a biomarker to diagnose MDD and identify patients who are most likely to respond to specific antidepressants.

They will test their hypotheses by identifying subgroup-specific biomarkers of MDD in the large existing dataset and then test the biomarkers’ diagnostic and prognostic utility in a prospective study of an additional 120 MDD patients. Specifically, they will: 1) Analyze connectivity patterns in the patients imaged to date and identify subtype-specific biomarkers that differentiate homogenous subgroups from healthy volunteers. 2) Undertake rsfMRI and anatomical MRI in a new group of 120 MDD patients to test the ability of these diagnostic biomarkers to identify patients in homogenous subgroups, and also determine whether the subgroups are clinically distinguishable on measures of symptoms such anxiety, sleep disturbances and cognitive function. 3) Follow a subset of the 120 MDD patients for 18 months to test the biomarkers’ utility for predicting each patient’s prognosis.

Significance: If future large-scale studies validate the biomarkers of MDD subgroups, this research may help to improve treatment and clinical management by facilitating early diagnosis of MDD and more effective treatment selection for patients in each subgroup.

Abstract

Development and validation of a prognostic neuroimaging biomarker for depression

Translational studies and clinical outcomes in major depressive disorder (MDD) are hindered by the lack of objective biomarkers, linked to a specific pathophysiology, and capable of quantifying or predicting treatment outcomes. Only about ⅓ of patients achieve full remission during acute phase treatment, and many, but not all, will require long-term maintenance therapy. Clinical variables have been associated with recurrence risk, but at present, there is no reliable biomarker for definitively identifying probable treatment responders on an individual basis or identifying those most likely to require long-term maintenance therapy. Converging evidence from multiple sources implicates abnormal patterns of metabolic activity and functional connectivity in frontolimbic circuits—particularly within the subgenual cingulate cortex—in the pathophysiology of depression. However, these measures show significant intersubject variability and significant overlap between patients and healthy controls, limiting their clinical utility. This variability is probably an accurate reflection of a syndrome that encompasses multiple, heterogeneous but clinically indistinguishable neuropathologies. If so, then the key to developing biomarkers may lie in assembling large datasets to empirically identify distinct neural correlates of depression in homogeneous subgroups of patients. The goal of this project is to develop diagnostic and prognostic neuroimaging biomarkers of putative depression subtypes in a large, retrospective dataset and then validate them in a rigorously characterized, prospective sample. We will test the hypothesis that varying constellations of depressive symptoms arise from distinct patterns of abnormal connectivity in corticolimbic circuits, which can be quantified using functional magnetic resonance imaging and exploited as a biomarker for diagnosing depression and predicting prognosis.