The innovation developed in the context of HBP is a Perturbational Complexity Index (PCI), which measures the complexity of electroencephalographic (EEG) responses to transcranial magnetic stimulation (TMS); it showed a remarkable sensitivity in detecting minimal signs of consciousness.
Read our interview with Professor Marcello Massimini who leads the PCI team:
What is the clinical problem addressed and the advantages of the solution provided?
Disorders of consciousness (DOC) after severe brain injury affect more than 1 million people worldwide each year and present a formidable diagnostic challenge. Typically, the level of consciousness in these patients is assessed based on their ability to connect to the surrounding environment and produce appropriate motor responses. A patient that reacts to different inputs, i.e. sensory stimuli and verbal commands, with complex motor outputs that are specific and reproducible is conscious, whereas a patient that remains unresponsive or only shows automatic behaviors is declared unconscious. This behavioral input-output approach at the bed-site by a trained observer currently represents the gold-standard for assessing the presence or absence of consciousness in hospitals and rehabilitation centers. Yet it suffers from fundamental limitations. First, because consciousness can be entirely generated within the brain, even in the absence of any input-output interaction with the external world; this occurs almost every night, while we dream or during certain forms of anesthesia. Second, because brain injury can result in severe blockage of sensory and motor functions leading to conditions in which a brain might be conscious but disconnected and not accessible from the world outside. In practice, failure to detect consciousness through the typical input-output paradigm results in a high rate of misdiagnosis (estimated between 20% and 40%), represents a burden for caregivers and families, as well as an ethical challenge and a major hurdle for rehabilitation strategies. The perturbational Complexity Index represents a strategy to overcome this problem. In a nutshell, the idea is that, instead of judging consciousness by the complexity of behavior, one should by-pass sensory inputs and motor outputs to directly measure the internal dynamic complexity of the brain to a brief perturbation. Now, we have developed techniques and algorithms to perform this kind of measure at the bedside of patients, and they show an unprecedented sensitivity in detecting consciousness.
Can you describe the equipment, the technology and the software solution involved?
Put simply, the technology works like this: we give the cerebral cortex a knock and then quantify the spatio-temporal complexity of the cause-effects chain triggered by this initial perturbation at the whole-brain level. That initial knock or zap is delivered via transcranial magnetic stimulation (TMS), a technique that activates a local population of cortical neurons in a non-invasive manner, through the skull. For perturbations to be controlled and reproducible, we use SmartFocus® TMS, a navigation system developed by the Nexstim company that enables surgical-degree precision targeting based on the online view of stimulating electro-magnetic-field superimposed onto the 3D reconstruction of the patient’s MRI. Then, we use a 60-channels EEG system to record the overall brain response, which typically lasts for a few hundred milliseconds; this post-zap electrical echo is the signal of interest. Taken in isolation, TMS, navigation systems, and EEG amplifiers – have been around for decades; what is novel is the integration of state-of-the-art systems and their optimization for obtaining high signal-to-noise EEG responses to controlled cortical perturbations.
Photo by Russ Juskalian
Photo by Russ Juskalian
The other key element of novelty is how we analyze this internal brain echo, to compute its complexity and extract an index of consciousness. Indeed, this represents an interesting example of how concepts from theoretical neuroscience can be directly translated into a number that doctors can use at the bedside. Back in 1998 Tononi and Edelman proposed in an article in Science that consciousness depends on the brain capacity to sustain complex patterns of interactions that are both integrated (due to tight causal interactions among neurons) and differentiated (rich in information). To translate this concept from theory to practice, we quantify the information content of the causal chain triggered in the brain by the targeted TMS zap by applying a classic compression algorithm (Lempel and Ziv), normally used to zip files. In the end, this “zap and zip” procedure yields a number, the so-called Perturbation Complexity Index (PCI), that is high for spatiotemporal patterns that are both integrated and differentiated, and low for brain responses that are either local (low integration) and/or stereotypical (low differentiation) such as those that occur during deep sleep, anesthesia and coma. Theoretical neuroscience aside, what is most exciting about all this is that PCI works extremely well when tested in real-life conditions.
How is the technology positioned in relation to other trends and solutions in the area?
In principle, this technology differs from other measuring approaches because it by-passes sensory inputs and motor outputs and because it combines perturbational measures of causality with measures of information in a single number. In this perspective, it clearly diverges from classic approaches based on sensory-evoked potentials, spontaneous EEG, and commercial consciousness monitors, such as the Bispectral Index. In practice, PCI is unique because it allows detecting recovery of consciousness with unprecedented sensitivity, even in challenging conditions. Over the last 15 years, we have been testing its performance in collaboration with different centers - the University of Wisconsin in Madison, the Coma Science Group in Liege, Fondazione Don Gnocchi in Milan and the Research Center of Neurology in Moscow - in hundreds of patients, spanning all the clinical conditions that follow coma. Here, PCI affords a sharp and reliable stratification of patients, even in the grey zone represented by subjects with little or no behavioral output. For example, when compared to other techniques, PCI shows the highest sensitivity (94% vs 14-70%) in detecting minimally conscious patients and it allows revealing recovery of consciousness in patients that would be otherwise declared in a vegetative state (also called unresponsive wakefulness syndrome), an astounding 1 out of 5 such patients. Such reliable detection occurs also when other solutions fail, such as when sensory evoked potentials are absent/inconclusive or when the spontaneous EEG is severely abnormal.
How was the innovation conceived and by whom? We would like to know the role played by HBP in this conception.
The initial idea was conceived almost 20 years ago when I visited Giulio Tononi in Wisconsin. I was interested in measuring how communication within the brain changes between wakefulness and sleep and he was working on a theory suggesting that the complexity of these interactions is a basic substrate of consciousness. The first time we met, we both pulled out of our backpacks the same paper: the first report by Risto Ilmoniemi, now Distinguished professor at Aalto University, showing that measures of cortico-cortical communication were technically feasible by combining TMS and EEG. We took this serendipity quite seriously and immediately contacted Nexstim, at the time a newborn spin-off company out of Helsinki, Finland, asking them to test their first prototype of TMS-compatible EEG amplifier. Subsequent experiments in sleep and anesthesia showed that idea was very promising. Then, a fantastic group of HBP partners, Steven Laureys, Olivia Gosseries and Melanie Boly at the University of Liege and Mario Rosanova, Silvia Casarotto, Adenauer Casali and Simone Sarasso at the University of Milan successfully applied the approach to patients emerging from coma. From SGA1 to SGA3, the team was able to further refine measurements and to validate them in larger cohorts of patients. Here, a key role was played by Fondazione Don Gnocchi in Milan, which opened a new brain injury division where PCI is used as a routine exam under the supervision of dr. Angela Comanducci. Yet, the contribution of HBP went well beyond support to algorithm development and clinical validation; thanks to a coordinated consortium effort, PCI has been explored across scales and models, from cortical slices (Mavi Sanches-Vives) to rodents (Johan Storm and Mavi Sanches-Vives), intracranially in humans (Andrea Pigorini) and in computer models (Jennifer Goldman and Alain Destexhe). Such a thorough multiscale exploration is rather unique for a bedside clinical index, a truly translational approach that can provide a solid mechanistic background to medical decisions.
Could you summarize what have been, in general, the most important technical difficulties and barriers found?
The first technical challenge has been managing and minimizing the EEG artifacts produced by the magnetic pulse. TMS injects for a fraction of millisecond tens of volts onto the EEG electrodes, enough to saturate the amplifier and obliterate brain responses for tens of milliseconds. This was solved by adopting EEG amplifiers with special circuits, a wide dynamic range and EEG caps with special electrodes. The second challenge was controlling the strength of the initial impact of TMS on cortical neurons. With TMS, the brain can be stimulated with a huge combination of parameters (location, angle and intensity of the stimulating electric field) and not all of them are effective in eliciting the high signal-to-noise responses needed to compute PCI. Together with the research and development team at Nexstim, we solved this problem by a combination of strategies. First, by maximizing a priori information about the cortical target, using individual MRIs and neuronavigation that enable the stimulating electric field targeting with accurate and reproducible TMS pulse delivery. Second, by developing an EEG-based data visualization software that allows the operator to titrate stimulation parameters in real time to attain the desired level of initial neuronal activation. Third, by developing new hardware solutions to maximize the effects of the TMS pulse on the cortex.
Are there companies involved? What kind of further technical development are you undertaking towards commercialization?
Yes, as I said, a constant collaboration with Nexstim has been pivotal since the start of this endeavour. Now, we are at a stage where we are testing a brand new TMS-EEG prototype that has just been shipped from Helsinki to Milan. The machine implements in a compact package all the features that I have described above. Although the precise nature of the technical solutions are covered by a non-disclosure agreement, I can say that this prototype clearly goes in the direction of transforming TMS-EEG from an investigational device into a user-friendly clinical tool. In essence, we have turned this electromagnetic perturb-and-measure approach into a tool akin to ultrasound echography (the size is also similar), whereby the operator holds a (magnetic) probe in his hands and performs accurate measurements via an informative readout (the electrical echo of the brain). The expectation is that, thanks to effective probes, real-time readout and training, this electromagnetic probe will become standardized to the point of supporting routine clinical evaluations, just like echography.
How will you standardize and disseminate this approach? Are you planning a clinical trial?
Indeed, we are devoting most of our current efforts towards standardization and dissemination. The goal is to extend testing of TMS-EEG and PCI beyond Milan, Liege and Moscow and to involve other European and US-based institutions. In Europe, we plan to provide normative data, analysis tools and training services to both researchers and clinicians with the coordination and support of the EBRAINS infrastructure. In the US, we are starting a large multicentric study. Here, thanks to the support of the Tiny Blue Dot Foundation, a Santa Monica based Foundation, TMS-EEG devices have been shipped to Massachusetts General Hospital (dr. Darin Dougherty and dr. Brian Edlow), to the Department of Psychology and Neurosurgery at UCLA (dr. Martin Monti), the Medical University of South Carolina (dr. Mark George) and the Department of Neurology of the University of Wisconsin (dr. Melanie Boly). After some delay due to the pandemic, data acquisition has started, and these centers are now applying for NIH support and are connecting to the FDA to set up a clinical trial. Clearly, this whole process will take some time, but we are already at a stage where training, protocols, equipment and data acquisition are defined and standardized.
Have you explored any venture capital (or similar) investing options to get funding?
Besides HBP, we have received so far significant support from, the Tiny Blue Dot Foundation (TBD) that has been fundamental in financing the development and distribution of the current TMS-EEG prototypes. In the near future, the goal is to find investors that are motivated in supporting the development and certification of a commercial version of the device that is ready for hospitals and rehabilitation centers. Along these lines we have started a coordinated state-of-the-art analysis of the European market in coordination with the HBP Innovation Team led by Gonzalo León and with Guillermo Velasco. Parallel explorations are being made in the US by Zeinab Barati, an expert in the exploitation of clinical research, with the support of the Tiny Blue Dot Foundation. In the previous development of the navigated TMS technology and these current projects, an active role is played by Henri Hannula, Vice President, International Sales and Marketing at Nexstim, through the European and the US-based branches of the company.
What is the market potential of the technology? Can you foresee applications beyond the field of disorders of consciousness?
Disorders of consciousness (DOC) are the source of intense burden on caregivers and patients’ families, due to the high cost of life-sustaining therapy, rehabilitation, and diagnostic/prognostic uncertainties. Given the prevalence of these disorders (about one million patients/year worldwide) this can be considered as a relatively small niche, but we expect that successful clinical application in this field will make a significant difference and be the entry point towards the broader domains. Along these lines, two TMS-EEG studies published last year in Brain (one by the Milan/Liege team and one by the group of dr. Christian Grefkes in Cologne) have shown a local reduction of perturbational complexity in the perilesional cortex of stroke patients, which predicted functional impairment and subsequent recovery. Hence, it is likely that TMS-EEG and PCI-like applications may turn out to be valuable in a much larger population of patients with focal brain injury. Indeed, in this clinical field there is an urgent need for reliable electrophysiological readouts to guide rehabilitation and neuromodulation. In general, this kind of standardized real-time readout from the brain status may also help clinicians titrate TMS-based neuromodulation therapies in larger groups of patients who have decreased or altered cortical reactivity (e.g. due to depression or neuropathic pain).