Using MRI to predict response to transcranial magnetic stimulation in schizophrenia

By Will Boggs MD

NEW YORK (Reuters Health) – Gray-matter changes in various brain regions predict how people with schizophrenia respond to repetitive transcranial magnetic stimulation (rTMS), researchers from Germany report.

rTMS has shown promise for treating schizophrenia and other psychiatric disorders, but results vary considerably from patient to patient.

Dr. Nikolaos Koutsouleris from Ludwig-Maximilians-Universitaet, Munich, Germany, and colleagues investigated whether machine learning analysis could identify pretreatment brain MRI structural predictors of a favorable response to rTMS. Their study, published online August 31 in Schizophrenia Bulletin, involved 92 participants with schizophrenia in the multicenter RESIS trial.

Reduced gray-matter density (GMD) in the dorsomedial and ventromedial prefrontal, frontopolar, and cingulate cortices; insular, opercular, temporopolar, and medial temporal cortices; and cerebellum predicted subsequent nonresponse to active rTMS.

In contrast, increased GMD in the left-hemispheric somatosensory and parietal cortices, with extensions to the lateral temporal and premotor structures, along with increases in the thalamic nuclei bilaterally, predicted nonresponse to rTMS.

Even these markers accurately predicted changes only in negative symptoms, not in positive symptoms, of schizophrenia in response to rTMS. Sociodemographic and clinical variables, antipsychotic treatment intensity, and type of psychopharmacological treatment did not appear to modify the predictors of rTMS response.

“This neuroanatomical baseline variance in patients with schizophrenia may represent an important, so far unknown biological factor driving the equivocal efficacy reported by previous rTMS studies in the field,” the researchers note.

“Our findings suggest that MRI-based machine learning derived from a careful randomized controlled trial design may foster a better understanding of and the ability to predict how different disease-related brain phenotypes contribute to the individual patient’s capacity to respond to brain stimulation,” they conclude.

Dr. Fabio Ferrarelli from the University of Pittsburgh School of Medicine, in Pennsylvania, who has used TMS to investigate the neurobiology of schizophrenia, told Reuters Health by email, “If these findings are confirmed and extended on larger groups of patients, they may contribute to identify responders from non-responders to rTMS, thus establishing a subset of schizophrenia patients who are especially suited for this type of non-pharmacological intervention.”

“These GMD patterns correctly predicted negative symptoms improvement only in patients receiving active rTMS, not in all patients,” he noted. “Indeed, in the sham rTMS group both ‘predicted responders and non-responders’ showed an improvement in their negative symptoms, which suggests that the GMD incorrectly predicted those patients who were not supposed to improve after sham treatment. To address this issue, future studies should make the ‘responder/non-responder prediction’ before randomizing patients to active versus sham rTMS.”

Dr. Ferrarelli added, “Factors such as duration of illness (before and after treatment) and exposure to pharmacological interventions can significantly affect brain volumes, including the GMD regions investigated in this study. It would therefore be important to assess the predictive value of these GMD patterns at illness onset, as well as before any other treatment intervention.”

Dr. Daniel M. Blumberger, medical head of University of Toronto’s Temerty Centre for Therapeutic Brain Intervention, in Canada, told Reuters Health by email, “The predictive accuracy of the machine learning algorithm was fairly high given the relatively small number of subjects used to train the algorithm. . . . Ultimately, we may be able to personalize treatments based on biological or illness characteristics.”

“Much larger samples are required to develop better algorithms, (which) will need to be tested in prospective controlled trials before they can be integrated into clinical practice,” he said.

SOURCE: http://bit.ly/2xkUsrJ

Schizophr Bull 2017.

(c) Copyright Thomson Reuters 2017. Click For Restrictions - http://about.reuters.com/fulllegal.asp