Individual patient responses to electroconvulsive therapy (ECT) may be predicted by alterations in an area of the brain associated with functional and structural abnormalities in patients with major depressive disorder, say German researchers.
Although approximately one third of patients do not benefit from ECT, numerous attempts to identify factors associated with nonresponse, including those based on demographics, symptom severity, phenomenology, and treatment history, have been largely unsuccessful.
Ronny Redlich, PhD, Department of Psychiatry, University of Muenster, Germany, and colleagues found that the size of the subgenual cingulate cortex was significantly associated with response to ECT.
"The present study might be a promising step to provide biomarkers for correctly identifying patients who are likely to respond to ECT," the investigators write.
"Although determining which ECT recipients will respond remains difficult in clinical practice, a routine assessment with structural MRI before treatment could serve as a decision guide for clinical psychiatrists," they add.
The research was published online May 4 in JAMA Psychiatry.
Structural Brain Imaging
Investigators conducted a nonrandomized naturalistic study in which MRI and voxel-based morphometry was used to assess gray matter structure approximately 6 weeks apart in 67 individuals.
These included 23 patients with acute major depressive disorder who were treated with ECT along with antidepressants, 23 patients treated only with antidepressants, and 21 health control participants.
ECT consisted of brief pulse therapy administered three times a week. Initially, nine to 12 sessions were administered; more were administered if the patient did not achieve symptom relief. The mean number of sessions was 14.0.
In all, 13 patients were classified as responders and 10 as nonresponders. Nonresponse was defined as <50% symptom relief on the Hamilton Depression Rating Scale.
Machine learning–based algorithms were used to allow the multivariate differentiation of two or more groups on the basis of structural brain images. A support vector machine differentiated ECT responders and nonresponders with an accuracy of 78.3% (P = .009) and a sensitivity of 100%.
A Gaussian process classifier achieved an accuracy rate for differentiating ECT responders and nonresponders of 73.9% (P = .01), again with a sensitivity of 100%.
Further analysis of the binary multivariate classification revealed that the subgenual cingulate gyrus was the most significant contributor to the classification of ECT response. No associations were found between response and medication load, patient age, or sex.
Interestingly, whole-brain regression analysis showed that higher pretreatment subgenual cingulate gyrus gray matter volume was significantly associated with a better clinical response in the ECT group (P < .001), a finding not observed in the patients receiving only medication.
The team also found that patients in the ECT group had significant increases in gray matter in the left hippocampus during the course of the study (P < .001). This was not observed in the medication-only and healthy control groups. Moreover, reductions in whole-brain gray matter volume that were observed at baseline in the ECT group were normalized after treatment.
Network Approach
Dr Abbott noted that the focus on the cingulate "is a little but out of context," inasmuch as the finding suggests that the subgenual cingulate gyrus may be part of a combination of structures that form a network that predicts response.
"I think that that's encouraging, because we already know that the cingulate gyrus is really involved in a lot of pathophysiology with respect to depression, and it's the site of deep brain stimulation, and I think that adds some validity to the approach."
A network approach, he added, "really lends itself to some more of these sophisticated computational strategies to help us learn more about depression and how it's widely distributed, not just maybe a single area."
Dr Abbott described the longitudinal finding that ECT induces neural plasticity as "very exciting," particularly because the finding lends support to recent neuroimaging studies in ECT.
Taken together, he said, these investigations support the notion that neuroplasticity may be a part of the mechanism of action of ECT, similar to chemical antidepressants, although it may be "a far more potent form of neuroplasticity."
Although the classification had high sensitivity, the model had low specificity. This was due to the fact that six patients were designated as nonresponders despite having partial responses, which raises the question as to whether it is better to classify response on a continuous or a dichotomous basis.
"I think, moving forward, it would help to look at maybe several different endpoints to really interrogate the data adequately and not just put our best foot forward," said Dr Abbott.
"I think the tendency might be to just report on either response or remission rates as a dichotomous variable, but I think a far more clinically meaningful one is to look at a change in depression rating, because that might not be captured with a remission criteria, which is rather stringent."
Dr Redlich added: "In my opinion, defining response on an individual patient level is more accurate but also more demanding for the clinician, who will have to decide whether ECT is indicated or not.
"A continuous, individual prediction of symptom improvement...requires a decision with regard to whether a predicted symptom improvement of around 30% to 50% justifies an ECT treatment or not."
He noted that in some cases, symptom improvement of 30% or 40% may be enough to justify use of ECT, "considering that this patient does not benefit from pharmacotherapy or psychotherapy."
The next item on Dr Redlich's research agenda is to investigate the effects of ECT on brain function, particularly during emotion and reward processing tasks, using functional MRI.
"Besides investigating the limbic and mesolimbic system, this also allows to us investigate functional connectivity patterns in order to prove our hypothesis," he said.
The investigators intend to replicate their findings by a study of depressive patients treated with ECT in five neuroimaging centers and aim to transfer a trained classifier to independent sites.
Precision Medicine
In an accompanying editorial, Christopher C. Abbott, MD, Department of Psychiatry, University of New Mexico School of Medicine, Albuquerque, New Mexico, said that the findings "will have an effect on precision medicine."
Noting that, with further validation, the classification patterns identified by Dr Redlich and colleagues "may inform" healthcare professionals about the likelihood of response to ECT and thus "spare the ECT nonresponder from the unnecessary risks, time, and costs associated with the procedure."
However, Dr Abbott told Medscape Medical News that the timeline for being able to bring this kind of identification to the clinic is "indeterminate." For him, the next step is to test the prediction algorithm with an independent dataset derived from multiple sites.
With that, he said, "I think we’ll get a better feel for how close we are to a clinical application in that context."
This study was supported by a grant from the German Research Foundation and from the Innovative Medizinische Forschung, and Rolf-Dierichs-Stiftung. Dr Arolt has received compensation as a member of advisory boards and for presentations for AstraZeneca, Eli Lilly, Janssen-Organon, Lundbeck, Otsuka, Servier, and Trommsdorff during the past 3 years. That compensation had no relevance to the study. The work represented in the editorial was supported by a grant from the Centers of Biomedical Research Excellence.
JAMA Psychiatry. Published online May 4, 2016. Abstract, Editorial
Medscape Medical News © 2016 WebMD, LLC
Send comments and news tips to news@medscape.net.
Cite this: Structural Brain Imaging May Accurately Predict ECT Response - Medscape - May 12, 2016.
Comments