Neurophysiologic Predictors of Outcome With rTMS Treatment of Major Depressive Disorder

A critical challenge in the management of major depressive disorder (MDD) is the selection of treatment for each individual patient. Although treatments with depression can restore people's lives, with any treatment modality there are some individuals who do not achieve complete remission of symptoms, whether the intervention is pharmacological, psychological, or somatic. While predictors for some treatments have been proposed for groups of patients, the translation of these predictors to individualized patient care has remained elusive. In an analysis of data from the NCT 00104611 multi-site, randomized, sham-controlled trial of TMS, it was found that a larger number of prior treatment failures, longer duration of the current episode, and the presence of comorbid anxiety were individual patient characteristics associated with poorer acute outcomes with TMS treatment in the randomized period (Lisanby et al., in press). This publication did not report standard predictor metrics (e.g., sensitivity, specificity, positive- or negative-predictive accuracy, ROC curves), so it is difficult to assess the value of these clinical factors in treatment planning for individual patients. A predictor that could distinguish between individuals likely to remit with TMS versus those likely to need a different intervention would be of great use to clinicians and patients in making treatment decisions.

Our prior work (Cook et al., 2001, 2002, 2005; Leuchter et al. 2002) has studied a new physiologic biomarker of response to SSRI and mixed-action antidepressants. The EEG-based cordance biomarker can detect the physiologic effects of successful antidepressant treatment at 48 hours, 1 week, and 2 weeks of treatment; in contrast, symptom differences between responders and non-responders did not separate until 4 weeks of treatment in our placebo-controlled trials. Additionally, the magnitude of early physiologic change was associated with the completeness of clinical response. Our biomarker has been independently studied and our findings replicated (Kopecek et al., 2006; Bareš et al., 2007, 2008). The cordance biomarker can be considered as a leading indicator or predictor of treatment outcome. As a non-invasive probe of brain physiology, it may detect early neurophysiologic changes associated with accelerated clinical response from TMS.

More recent work with a related EEG-based measure, the Antidepressant Treatment Response Index (ATR) has led to a simplified monitoring system; a physician can record clinically-useful data from a 15-minute in-office procedure with electrodes located on the forehead and ears (Leuchter et al., in submission). The ATR uses physiologic data collected prior to treatment and after one week of exposure, and was shown to be predictive of outcome with antidepressant medication. We are able to assess both cordance and ATR measures with EEG measurements made prior to treatment and after 5 treatment sessions with TMS to evaluate the predictive properties of both metrics.

On a related issue, some of the variation in outcome may be related to treatment factors. Quantitative models and direct in vivo measurements (Wagner et al, 2004, 2008) indicate that the electrical currents induced by TMS are predominantly confined to a brain region directly under the treatment coil. The procedure for positioning the coil over the cortical target is described in the NeuroStar TMS System User Manual (volume 2, sections 6 and 7) and involves first determining a location where stimulation leads to a contraction of the abductor policis brevis muscle (visualized with a thumb twitch on the right hand) and then positioning the coil 5.5 cm anterior to that position along the left Superior Oblique Angle line. While this target can be located with good reproducibility and was associated with therapeutic outcome in the NCT trial, it is not clear that this positions the coil over the best target within the DLPFC for all patients. Indeed, individual differences in gyral anatomy and in gross brain size both add variability to the specific neuroanatomic region being stimulated, and this may impact treatment efficiency.

Exposure to even a brief train of TMS pulses can elicit an acute physiologic change (cf Siebner and Rothwell, 2003), and so a test procedure can be performed that will assess the distance from the standard treatment position to the point eliciting a maximal acute physiologic response. We propose a 9-locus mapping procedure, involving the assessment of changes in brain activity from stimulation at locations including and around the standard treatment target. The nine locations will be the usual treatment location and 8 other points, 1.5 and 3.0 cm anterior, posterior, rostral, and caudal of the primary target. Test stimulation will be for 15 seconds (=150 pulses @ 10 Hz) at each location, followed by 5 minutes of continuous EEG recording to examine acute changes in regional brain activity in response to a brief stimulation exposure. All therapeutic stimulations will take place in the standard location, and we will be able to evaluate what proportion of variance in clinical outcome is explained by distance from the location of maximal acute physiologic response.

Based on these previous studies, we propose to assess patients during treatment with TMS, using clinical symptom ratings and brain physiology with EEG.