Development of a Predictive Model of Effectiveness for the Implantation of Electrical Neurostimulators in Patients With Chronic Pain Using Imaging Biomarkers Extracted From Magnetic Resonance
Development of a Predictive Model of Effectiveness for the Implantation of Electrical Neurostimulators in Patients With Chronic Pain Using Imaging Biomarkers Extracted From Magnetic Resonance
Chronic pain is correlated with alterations in the structure and function of the brain, developed according to the phenotype of pain. Still today, the data on functional connectivity (FC), on chronic back pain, in patients with failed back surgery syndrome (FBSS), is limited. The selection process for the ideal candidate for spinal cord stimulation (SCS) is based on results from test and functional variables analysis as well as pain evaluation. In addition to the difficulties in the initial selection of patients and the predictive analysis of the test phase, which undoubtedly impact on the results in the middle and long term, the rate of explants is one of the most important concerns, in the analysis of suitability of implanted candidates. The hypothesis is that the structural and functional quantitative information provided by imaging biomarkers will improve the characterization of the patients compared to the characterization with the current clinical variables alone and this will allow establishing a CDSS that improve the effectiveness of the SCS implantation, optimizing human, economic and psychological resources.
A prospective, consecutive and observational, open-label, single-center study conducted at the Multidisciplinary Pain Management Department of our University Hospital. A total of 69 subjects were initially included in the study. The population split in 3 groups:
MR images were obtained in a 1.5T MR system (Ingenia, Philips, Best, The Netherlands) using an 8-channel head coil.Clinical variables were evaluated at two different time points baseline and 12 months after SCS implantation or conventional medication. An ad hoc database was created to evaluate the different variables involved in pain , including sociodemographic variables (age, gender, level of studies and marital status), clinical variables (anxiety, depression, sleeping hours, resilience, NRS, the Pain Detect Questionnaire (PD-Q)) , and the images obtained from the fMRI.
Primary objective:To develop a predictive model integrated in a clinical decision support system (CDSS) feed by neuroimaging quantitative information objectively extracted from Magnetic Resonance (MR) images, which maximizes the appropriate use and effectiveness of electrical stimulation devices surgically implanted in selected patients with chronic pain.
Exploratory Objectives:
Test Device:1.5 Tesla MR system (Philips Healthcare, Best, The Netherlands) Boston Scientific Neuromodulation (BSN) Precision Spectra™ Spinal Cord Stimulation System with Illumina 3D™ Software and 32 contacts.
Device Description: Precision Spectra™ system IPG is a multiple independent current controlled pulse generator, capable of delivering current through 32 contacts. It is powered by a 3D programming software that considers the anatomical position of the leads. Two models of SCS leads will be provided, featuring 8 or 16 contacts with 1.3 mm diameter, 3 mm contact length, and contact spacing of 1, 4 or 6 mm. The use of SCS extensions will be optional to connect the IPG.
fMR description: The MR experiment will be consistent with on-label requirements. The MR procedure will be performed prior to the device implantation to avoid bias. Even more, the Food and Drug Administration (FDA) does not recommend the examination of patients with this kind of devices for security reasons.
Examinations will be performed in a 1'5 Tesla MR system (Philips Healthcare, Best, The Netherlands) at the Quiron Hospital. Decision of magnetic field is based on the quality of the examinations and must rely on label products and approval of company for interaction with implanted system.
A head coil with 8 reception channels will be used. Once the patient has been positioned in the system, initial and fast localization images will be acquired in order to properly plan the MR sequences of the research study.
After planning, a resting-state functional MR (rs-fMR) imaging sequence will be acquired, asking to the patient to be quiet with the eyes closed and thinking in a blue sky. The acquisition parameters will consist of an Echo Planar (EPI) dynamic T2* sequence, full brain coverage with the following parameters: TR=2000 ms; TE=30 ms; voxel size, 1.8 × 1.8 x 3.5 mm; flip angle, 90º; 40 axial slices; acquisition time 5:20 min.
A DTI MR sequence will be acquired in order to analyze white matter microstructure and connectivity by tractography techniques with the following parameters: Spin-Echo Echo Planar Imaging (SE-EPI) sequence, single shot; full brain coverage; 64 gradient directions; b-value, 1300 s/mm2; TR=6200 ms; TE=67 ms; voxel size, 2 x 2 x 2 mm; 60 axial slices; acquisition time 9:40 min.
An additional anatomic sequence will allow overlying structural and functional results and, in addition, obtaining the volumetry values of each brain region. The sequence parameters are: T1-weighted 3D gradient echo sequence (GRE), full brain coverage; TR=11.6 ms; TE=5.69; voxel size, 0.48 x 0.48 x 0.50 mm; flip angle, 8º; 280 axial slices; acquisition time 5:36 min.
After image acquisition, all data sets will be sent to the Imaging Biomarkers Platform of the Biomedical Imaging Research Group (GIBI230) of the La Fe Research Institute.
The fMR images will be aligned in order to correct possible small patient's head movements during examination. For that, the open source SPM8 (Statistical Parametric Mapping, http://www.fil.ion.ucl.ac.uk/spm/) software tool will be used. After movement correction, a temporal correction will be applied optimizing the slice timing. Images will be then normalized to a standardized brain template in order to allow for the study of the oscillations between individuals. After such processes, data will be filtered by a3D-Gaussian kernel in order to increase signal-to-noise ratio (SNR) while minimizing inter-subject differences. Finally, the application of independent component analysis (ICA) algorithms will allow for the extraction of brain activation maps in the subject during the acquisition.
The analysis of the DTI MR data for extracting white matter tracts connectivity will be performed with the open-source FSL software tool (http://www.fmrib.ox.ac.uk/fsl/). An initial Eddy currents correction will be applied to the images, in order to minimize slight images displacements and geometry distortions. The brain will be then segmented using the BET algorithm and brain data of all patients will be normalized to a common template for the group-based analysis. After this process, fractional anisotropy (FA), diffusivity (D) and orientation maps will be obtained .
Once the MR images have been processed, the structural and functional connectivity properties have to be extracted from the regions of interest (ROI). The positioning of these regions will be obtained from the zones involved in the Default Mode Network (DMN). The DMN might take an important role in pain perception and shows a high correlation with the symptoms described by patients, which makes this network useful for the prediction of patient response after the implantation of electrical stimulation, either at a functional or at a structural level.
Since images will be normalized to a common template, the automated area labeling AAL-tool will be used to define the regions of interest of the study, that compound the DMN and are formed by the medial temporal lobe, prefrontal cortex, posterior cingulate, precuneus and the parietal cortex. After measurement the connectivity parameters in these regions, a predictive model will be developed by combining clinical variables (scales and symptoms of each patient) and neuroimaging information. These models will be initially adjusted with a total of 30 patients (training data) and later validated in a group of 30 patients (validation data), obtaining thereafter results of specificity, sensitivity and models precision.
Inclusion Criteria:
Exclusion Criteria: