Posted by Phillipa on August 19, 2010, at 20:26:11
They have applied new algebraic interpreatation to EEG's and have found this can predict response to Clozapine for tx resistant schizophenia, also been applied to SSRI's and atypical antipsychiotics. Certainly could be very promising to many. It is better than rTms. Phillipa
From Medscape Medical News
EEG Testing May Predict Clozapine Response in Treatment-Resistant Schizophrenia
Deborah BrauserAugust 18, 2010 (Updated with comment August 19, 2010) Electroencephalography (EEG) testing may be able to predict response to clozapine in patients with treatment-resistant schizophrenia a new pilot study suggests.
In fact, the response prediction accuracy rate was over 85% during both the first part of the study, where a computer algorithm was "trained" using brainwave patterns from a group of clozapine-naive patients, and during the second part, where additional patients were tested, report researchers from McMaster University in Hamilton, Ontario.
"We found that certain brainwave patterns, particularly in the back half of the brain on the left side, seem to be predictive of whether a person would respond to clozapine or not," study coauthor Gary Hasey, MD, associate professor in the Department of Psychiatry and Behavioral Neurosciences at McMaster and director of the Transcranial Magnetic Stimulation Laboratory and the Mood Disorders Program at St. Joseph Hospital, told Medscape Medical News.
The investigators note that, although recognized as an effective treatment for schizophrenia, clozapine can produce serious adverse effects, including seizures, cardiac arrhythmias, bone marrow suppression, and even life-threatening blood problems. Because of this, the drug requires regular blood monitoring.
"The logistic difficulties for the patient and treatment team are substantial," they write. "A method that could reliably determine, before the onset of therapy, whether a given patient will or will not respond to clozapine would greatly assist the clinician in determining whether [its] risks and logistic complexity...are outweighed by the potential benefits.
"The take-home message is that this is very encouraging, positive pilot data," added Dr. Hasey. "If we can validate it, it could mean that clinicians could have a very simple, inexpensive way to predict response to a particular drug and is pointing toward individualized therapy, where we go beyond guess work. The current trial and error process can take a long time to get some people well, and that's costly to both the individual in terms of personal suffering and it's expensive to the insurance provider."
The study was published online June 17 in Clinical Neurophysiology.
EEG Noninvasive, Widely Available
Traditionally, EEG has been used to monitor for epilepsy and diagnose coma, brain death, tumors, stroke, and other brain disorders.
Although past studies have looked at using structural brain magnetic resonance images (MRIs) and functional MRI (fMRI) data to diagnose psychiatric conditions, such as schizophrenia and bipolar disorder, and pretreatment fMRI and positron emission tomography to predict treatment efficacy, "the clinical utility of these approaches is negatively impacted by the expense and unavailability of [the] complex methods," the investigators write.
However, "EEG is an inexpensive, noninvasive technique widely available in smaller hospitals and in community laboratories," study coauthor Duncan MacCrimmon, MD, said in a release. "Also, EEG readings take only 20 to 30 minutes of a patient's time, with no preparation required, so pose minimal inconvenience.
"The overall question was, 'Can you predict response to various types of psychiatric treatment using EEG?' That was our main starting point," explained Dr. Hasey. "My colleagues in psychiatry and I have been working on that issue for a number of years."
He noted that in the past, they were able to find some simple algebraic equations between certain aspects of the EEG and transcranial magnetic stimulation (TMS), a relatively new treatment for major depression. This approach changed when some engineering colleagues introduced the machine learning method to the psychiatric group.
"After looking at our EEG data, they found that we could get a much better, more accurate prediction using more sophisticated math," said Dr. Hasey.
For this study, the researchers collected resting EEG data from 23 chronically ill patients (52% men; mean age, 41.2 years) with schizophrenia who were clozapine naive. The computer algorithm then used these data to determine treatment-response predictions.
Positive and Negative Syndrome Scale scores were used to measure symptom severity after clozapine treatment and as treatment-response indicators, along with other scales, such as the Quantitative Clinical Assessment, after at least 1 year of patient follow-up.
The computer algorithm was also tested on 14 additional patients (57% male; mean age, 35.7 years) who underwent EEGs after being treated with clozapine.
Strong Predictor of Treatment Response
At the end of the study, the investigators found that 12 patients in the first group were responders to clozapine, whereas 11 were nonresponders. In all models used to evaluate this group, the prediction performance was well above 85%.
After training the computer program's classifiers using the first group's dataset, the average prediction performance for the second group was 85.7%.
In addition, the investigators found a list of 20 discriminating EEG features for treatment-efficacy prediction, which "could give clues about the locality and interconnection of neurological mechanisms associated with a positive response to clozapine," they write. "Further investigation of this matter remains a promising topic for future work."
Overall, "our findings support the potential utility of machine learning methods in clinical psychiatry," write the study authors. "We have been able to predict, in advance of the first dose, whether a treatment-resistant patient will or will not respond to a powerful but potentially toxic medication."
Dr. Hasey reported that the researchers have already gone on to successfully use these machine learning methods to analyze EEG signals in predicting response to other treatments for patients with other psychiatric conditions, especially with antidepressants.
"We found quite similar data for selective serotonin reuptake inhibitors," he explained. "We also used it to predict response to [TMS] and found that it works quite well."
Promising Ideas
"This is an interesting study with some promising ideas," Georgios Petrides, MD, associate professor at Albert Einstein College of Medicine and research psychiatrist at the Zucker Hillside Hospital of Northshore-Long Island Jewish Health System in Glen Oaks, New York, told Medscape Medical News.
"What they're proposing is actually a very sophisticated way to analyze EEG data. And they applied that to patients with treatment-resistant schizophrenia to see if they could predict response to clozapine," added Dr. Petrides, who was not involved with this study. "Their initial data shows that there is a promise there but it needs replication and larger numbers of patients."
In addition, he noted that he was "not sure how widely applicable" the findings are given the fact that clozapine is often a last resort medication for treatment-resistant patients. "Even if you could predict patients' response or nonresponse, people would still be willing to try it anyway.
"What I think is more interesting is if their methodology can be applied to various types of other medications which are [used as a] first option, which would then give you direction as to which ones are more likely to be beneficial to the patient," said Dr. Petrides. "For example, if someone can predict a patient's response to aripiprazole vs Risperdal or olanzapine, that would be very helpful.
"But as general application, it seems to me that [these findings] are not going to have a major impact on the way that people choose to prescribe clozapine or not," he concluded.
This study was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) and by The Magstim Company Ltd. The study authors and Dr. Petrides have disclosed no relevant financial relationships.
Clin Neurophysiol. Published online June 17, 2010.
poster:Phillipa
thread:959182
URL: http://www.dr-bob.org/babble/20100811/msgs/959182.html