BRAIN Publication Roundup – June 2018

First demonstration of adaptive DBS for Parkinson’s disease using motor cortex sensing… Coordinated articulator movements discretely encoded in sensorimotor cortex… Ethical considerations for guidelines on continued access to investigational brain implants…

Adaptive DBS for Parkinson’s disease detects and responds to adverse treatment effects

Deep brain stimulation (DBS) can be an effective treatment for Parkinson’s disease, especially for patients whose symptoms remain uncontrolled by medication. However, adverse treatment-related side effects like dyskinesia — abnormal involuntary movements — require patients and physicians to manually adjust settings in response to changing symptoms. This constant re-programming is time consuming and imprecise. To address this issue, Dr. Philip Starr and colleagues at the University of California, San Francisco, developed a technique to automatically detect dyskinesias and adjust DBS stimulation in real-time, an approach called adaptive DBS. The researchers noted that powerful oscillations in the gamma frequency range reliably occur in the motor cortex when dyskinesia is present, independent from other neural signals. In two patients with Parkinson’s disease, the scientists tested a novel DBS system using a stimulation lead in the subthalamic nucleus (STN) and a second electrocorticographic (ECoG) implant over the motor cortex for detecting the gamma signal, both connected to an implantable pulse generator (IPG) capable of recording as well as stimulation. The team first prototyped a control algorithm using an external computer, which sensed when the ECoG gamma signal exceeded a certain threshold — indicating dyskinesia — and implemented changes to the STN stimulation via radio signals. They subsequently implemented this algorithm using the implanted IPG, which recorded ECoG signals and then successfully bilaterally adjusted STN stimulation, increasing or decreasing stimulation if the gamma oscillations fell below or rose above the threshold, respectively. This innovative development resulted in a fully embedded self-tuning DBS device, offering a simple way to continuously manage side-effects while consuming significantly less energy than open-loop DBS. Though the group has yet to compare clinical efficacy of the two methods, this breakthrough in adaptive DBS is highly promising for the future of Parkinson’s disease treatment, especially for patients with severely fluctuating symptoms in the utmost need of DBS. For more information, see the NINDS press release on this work.

Starr_figure_high res
(A) Illustration of fully implanted, closed loop adaptive DBS system. (B) Top panel: spectrogram of signals recorded from motor cortex. Second panel: classifier state (gold), DBS voltage (blue), and gamma (60-90 Hz) power used as the control signal (red). Lower panels show a zoomed in view to demonstrate that transitions in DBS amplitude correspond appropriately to fluctuations in gamma band power.
Coordinated articulator movements are discretely encoded in sensorimotor cortex to provide the kinematic basis of human speech

To perform the extraordinary feat of speaking, human beings engage almost 100 muscles to power the articulators — our lips, jaw, tongue, and larynx — and rapidly reshape the vocal tract, producing the fluid sounds of speech. Neuroscientists have often wondered how the brain encodes the kinematics of speech. Research points to the ventral sensorimotor cortex (vSMC) as a candidate control region, but it is challenging to investigate the neural representation of articulatory movements during continuous speech, beyond isolated segments or phonemes. To overcome this challenge, Dr. Edward Chang and his team at the University of California, San Francisco, designed an innovative technique to capture vSMC encoding of natural speech kinematics. The researchers used data from 8 speakers reciting sentences to develop a deep learning computer model that then produced standardized estimates of articulator movement for all speech sounds. Next, they performed electrocorticographic (ECoG) recordings from the vSMC while 5 new participants read sentences aloud, using acoustic-to-articulatory inversion (AAI) to infer the associated articulatory kinematics. Subsequently, the team used modelling to obtain a final product of articulatory kinematic trajectories (AKTs), traces that each represent the characteristic vocal tract movement encoded by one electrode’s recorded vSMC activity. This approach culminated in insightful new findings, chiefly that each discrete neuronal population recorded by a single electrode encoded AKTs, a level of complexity that had not been previously observed. AKTs are the result of the coordination of articulator movements, each underlying a particular vocal tract shape — such as an alveolar constriction, which produces similar phonemes like /t/, /d/, and /z/ — rather than simply the movement of one articulator. There were four broad categories of AKTs, each of which represented a group of phonemes requiring similar articulator formation. The vSMC showed topographical organization by these four AKT categories, contrary to theories of a one-to-one mapping between cortical site and articulator. AKTs also encoded co-articulation effects, as electrode activity was dependent on the articulator formations that precede or follow an AKT in successive speech. These new insights, likely overlooked without the team’s pioneering technique, point to coordinated articulatory gestures as the basic cortical units of natural speech. This significantly improved understanding will accelerate new exploration into the realization of speech, sequential motor movement, and executive planning overall.

Chang_figure_high res
(A) Clustering of encoded AKTs for all 108 electrodes across 5 participants. Each column represents one electrode. AKT kinematics were described by maximal articulator displacement (purple to green) along their principal movement axis. Electrodes were clustered by their kinematic descriptions, resulting in four primary clusters. (B) Kinematically clustered electrodes also encoded four clusters of encoded phonemes differentiated by place of articulation (alveolar, bilabial, velar, and vowels). (C) Average AKTs across all electrodes in each cluster. Each trace represents kinematic trajectory of an articulator with a line that becomes thicker with time.
Ethical considerations demonstrate need for guidelines on post-trial access to investigational brain implants for clinical trial participants

Brain implant technologies like deep brain stimulation (DBS), brain-computer interfaces, and neuroprosthetics hold immense promise for treatment-resistant conditions, such as Parkinson’s disease and depression, and for restoring function, such as in traumatic brain injury and vision loss. Existing regulations mandate protections for participants during these high-risk and invasive procedures. However, there are currently no clear guidelines for post-trial maintenance and adjustment of brain implant devices when a device clearly benefits a patient participant. Dr. Gabriel Lázaro-Muñoz and colleagues from Baylor University raise this issue, noting the overall absence of guidelines concerning continued participant access to experimental brain implants. Citing recent examples, the authors explain that research sponsors and health insurers often deny coverage for maintenance or removal costs beyond trial completion, imposing the near-impossible burden on participants whose symptoms improve to rely on personal funds or advocacy. Urging prompt adoption of guidance for continued implant access, the group uses ethical grounds to argue that researchers enter into a relationship of trust with patients and thus owe a limited duty of care. First, they note, compassion requires assisting participants for whom the implant is the best and only treatment option. Second, the Belmont Report research ethics principle of “respect for persons” implies that research participants deserve recognition beyond research usefulness, and it is disrespectful to inflict costs that essentially deny further treatment. Finally, they posit that participants are research partners and merit reciprocation in the form of continued access to the brain implant that brings relief. The team proposes coordination between sponsors, Institutional Review Boards, and scientists to outline plans for continued device access, duration, and steps for determining whether to provide for individuals. These stipulations depend on costs, patient vulnerability, and the protection of future research, but the authors argue that it is ethically imperative to anticipate and build assurances for trial participants who benefit from brain implants. This group has importantly highlighted the necessity of neuroethics in informing responsible human neuroscience research, especially as brain implant devices blossom into a common medical and ethical reality.