Jejo Koola

Serial Vagus Nerve Stimulation Functional MRI in Treatment-Resistant Depression

Vagus nerve stimulation (VNS) therapy has shown antidepressant effects in open acute and long-term studies of treatment-resistant major depression. Mechanisms of action are not fully understood, although clinical data suggest slower onset therapeutic benefit than conventional psychotropic interventions. We set out to map brain systems activated by VNS and to identify serial brain functional correlates of antidepressant treatment and symptomatic response. Nine adults, satisfying DSM-IV criteria for unipolar or bipolar disorder, severe depressed type, were implanted with adjunctive VNS therapy (MRI-compatible technique) and enrolled in a 3-month, double-blind, placebo-controlled, serial-interleaved VNS/functional MRI (fMRI) study and open 20-month follow-up. A multiple regression mixed model with blood oxygenation level dependent (BOLD) signal as the dependent variable revealed that over time, VNS therapy was associated with ventro-medial prefrontal cortex deactivation. Controlling for other variables, acute VNS produced greater right insula activation among the participants with a greater degree of depression. These results suggest that similar to other antidepressant treatments, BOLD deactivation in the ventro-medial prefrontal cortex correlates with the antidepressant response to VNS therapy. The increased acute VNS insula effects among actively depressed participants may also account for the lower dosing observed in VNS clinical trials of depression compared with epilepsy. Future interleaved VNS/fMRI studies to confirm these findings and further clarify the regional neurobiological effects of VNS.

Motor threshold in transcranial magnetic stimulation: the impact of white matter fiber orientation and skull-to-cortex distance.

The electrophysiology of transcranial magnetic stimulation (TMS) of motor cortex is not well understood. In this study, we investigate several structural parameters of the corticospinal tract and their relation to the TMS motor threshold (MT) in 17 subjects, with and without schizophrenia. We obtained structural and diffusion tensor MRI scans and measured the fractional anisotropy and principal diffusion direction for regions of interest in the corticospinal tract. We also measured the skull‐to‐cortex distance over the left motor region. The anterior–posterior trajectory of principle diffusion direction of the corticospinal tract and skull‐to‐cortex distance were both found to be highly correlated with MT, while fractional anisotropy, age and schizophrenia status were not. Two parameters—skull‐to‐cortex distance and the anterior component of the principle diffusion direction of the corticospinal tract as it passes the internal capsule—are highly predictive of MT in a linear regression model, and account for 82% of the variance observed (R2 = 0.82, F = 20.27, P < 0.0001) in measurements of MT. The corticospinal tracts anterior–posterior direction alone contributes 13% of the variance explained. Hum Brain Mapp, 2009.

The maximum-likelihood strategy for determining transcranial magnetic stimulation motor threshold, using parameter estimation by sequential testing is faster than conventional methods with similar precision.

Background: The resting motor threshold (rMT) is the basic unit of transcranial magnetic stimulation (TMS) dosing. Traditional methods of determining rMT involve finding a threshold of either visible movement or electromyography (EMG) motor-evoked potentials, commonly approached from above and below and then averaged. This time-consuming method typically uses many TMS pulses. Mathematical programs can efficiently determine a threshold by calculating the next intensity needed based on the prior results. Within our group of experienced TMS researchers, we sought to perform an illustrative study to compare one of these programs, the Maximum-Likelihood Strategy using Parameter Estimation by Sequential Testing (MLS-PEST) approach, to a modification of the traditional International Federation of Clinical Neurophysiology (IFCN) method for determining rMT in terms of the time and pulses required and the rMT value. Methods: One subject participated in the study. Five researchers determined the same subject’s rMT on 4 separate days–twice using EMG and twice using visible movement. On each visit, researchers used both the MLS-PEST and the IFCN methods, in alternating order. Results: The MLS-PEST approach was significantly faster and used fewer pulses to estimate rMT. For EMG-determined rMT, MLS-PEST and IFCN derived similar rMT, whereas for visible movement MLS-PEST rMT was higher than for IFCN. Conclusions: The MLS-PEST algorithm is a promising alternative to traditional, time-consuming methods for determining rMT. Because the EMG-PEST method is totally automated, it may prove useful in studies using rMT as a quickly changing variable, as well as in large-scale clinical trials. Further work with PEST is warranted.

Estimating resting motor thresholds in transcranial magnetic stimulation research and practice: a computer simulation evaluation of best methods.

Objectives: Resting motor threshold is the basic unit of dosing in transcranial magnetic stimulation (TMS) research and practice. There is little consensus on how best to estimate resting motor threshold with TMS, and only a few tools and resources are readily available to TMS researchers. The current study investigates the accuracy and efficiency of 5 different approaches to motor threshold assessment for TMS research and practice applications. Methods: Computer simulation models are used to test the efficiency and accuracy of 5 different adaptive parameter estimation by sequential testing (PEST) procedures. For each approach, data are presented with respect to the mean number of TMS trials necessary to reach the motor threshold estimate as well as the mean accuracy of the estimates. Results: A simple nonparametric PEST procedure appears to provide the most accurate motor threshold estimates, but takes slightly longer (on average, 3.48 trials) to complete than a popular parametric alternative (maximum likelihood PEST). Recommendations are made for the best starting values for each of the approaches to maximize both efficiency and accuracy. Conclusions: In light of the computer simulation data provided in this article, the authors review and suggest which techniques might best fit different TMS research and clinical situations. Lastly, a free user-friendly software package is described and made available on the world wide web that allows users to run all of the motor threshold estimation procedures discussed in this article for clinical and research applications.

Development of an automated phenotyping algorithm for hepatorenal syndrome

OBJECTIVE Hepatorenal Syndrome (HRS) is a devastating form of acute kidney injury (AKI) in advanced liver disease patients with high morbidity and mortality, but phenotyping algorithms have not yet been developed using large electronic health record (EHR) databases. We evaluated and compared multiple phenotyping methods to achieve an accurate algorithm for HRS identification. MATERIALS AND METHODS A national retrospective cohort of patients with cirrhosis and AKI admitted to 124 Veterans Affairs hospitals was assembled from electronic health record data collected from 2005 to 2013. AKI was defined by the Kidney Disease: Improving Global Outcomes criteria. Five hundred and four hospitalizations were selected for manual chart review and served as the gold standard. Electronic Health Record based predictors were identified using structured and free text clinical data, subjected through NLP from the clinical Text Analysis Knowledge Extraction System. We explored several dimension reduction techniques for the NLP data, including newer high-throughput phenotyping and word embedding methods, and ascertained their effectiveness in identifying the phenotype without structured predictor variables. With the combined structured and NLP variables, we analyzed five phenotyping algorithms: penalized logistic regression, naïve Bayes, support vector machines, random forest, and gradient boosting. Calibration and discrimination metrics were calculated using 100 bootstrap iterations. In the final model, we report odds ratios and 95% confidence intervals. RESULTS The area under the receiver operating characteristic curve (AUC) for the different models ranged from 0.73 to 0.93; with penalized logistic regression having the best discriminatory performance. Calibration for logistic regression was modest, but gradient boosting and support vector machines were superior. NLP identified 6985 variables; a priori variable selection performed similarly to dimensionality reduction using high-throughput phenotyping and semantic similarity informed clustering (AUC of 0.81 - 0.82). CONCLUSION This study demonstrated improved phenotyping of a challenging AKI etiology, HRS, over ICD-9 coding. We also compared performance among multiple approaches to EHR-derived phenotyping, and found similar results between methods. Lastly, we showed that automated NLP dimension reduction is viable for acute illness.