Reply: P300 amplitudes after concussions are usually decreased not increased

Sir, In their Letter to the Editor, Rosburg and Mager (2019) provide thoughtful commentary on our study in which the brain vital signs framework was used to monitor concussion-related effects in junior ice-hockey players (Fickling et al., 2019). The main comment relates to an apparent discrepancy between P300 amplitude changes after concussion. Fickling et al. (2019) reported increased P300 amplitudes relative to baseline in athletes in the early acute phase of concussion (<24 h), which does not agree with literature from traditional laboratory-based studies as reviewed by Brush et al. (2018). However, this may not be a discrepancy, but rather a reality. The complex cascade of physiological changes that occur following brain injury are highly dynamic over minutes, hours, days, weeks, months, and beyond (Giza and Hovda, 2001, 2015). As an important marker of information processing, P300 amplitude is influenced by a range of factors that affect attention, recognition, and context updating (Johnson, 1986; Connolly and D’Arcy, 2000). In the clinical translation into monitoring brain vital signs, assuming certain patterns of amplitude changes over time may not always be realistic with respect to the acute, subacute, and chronic phases. A basic vital sign framework would monitor differences both within and between individuals across these phases. This is a particularly important highlight of the Fickling et al. (2019) study, which to our knowledge represents the first observation of P300 changes in the early acute phase immediately following concussion (this equally applies to the N100 and N400 components). As Rosburg and Mager (2019) correctly point out, the closest comparison of post-concussive results occurred 1 week after injury (Candrian et al., 2018), with the remainder of P300 studies reviewed by Brush et al. (2018) taking place weeks (Gosselin et al., 2006), months (Dupuis et al., 2000; Gaetz et al., 2000; Lavoie et al., 2004; Thériault et al., 2009; Baillargeon et al., 2012), and the majority, years after injury (De Beaumont et al., 2007, 2009; Broglio et al., 2009; Pontifex et al., 2009; Ozen et al., 2013; Moore et al., 2014, 2015, 2017; Parks et al., 2015; Ledwidge and Molfese, 2016). It is necessary to fully characterize ERP amplitude changes across all time points during the post-concussive cascade of physiological injury and subsequent recovery. This is critical ahead of drawing any conclusions within individuals, between groups, or between studies. We are pleased that the brain vital signs framework has provided the initial evidence to encourage further investigations of this nature. We also agree with the fundamental concept that Rosburg and Mager (2019) advance in terms of the need to further validate this initial result under controlled conditions. The central issue relates to translating ERPs from the laboratory to clinical point-of-care settings, in order to develop an accessible and objective evaluation of brain function analogous to existing vital sign measures in accessibility at the point of care and applicability across a range of environments (Ghosh Hajra et al., 2016, 2018a; Pawlowski et al., 2019). We were excited by the authors’ excellent commentary around key issues in the clinical ERP translation, and certainly agree about the importance of these methodological factors (Connolly and D’Arcy, 2000; Gawryluk et al., 2010). Early acute post-concussive results can only be collected in environments outside of the laboratory. However, it is still possible to incorporate factors to address consistent noise and distractions. For the rink-side assessments, the Fickling et al. (2019) study consistently collected controlled data in a private room using comparable methodologies applied in the laboratory (e.g. noise control, fixations, consistent acquisition conditions between sessions, etc.). To address this challenge further, we are currently working on data analysis techniques to account for the effects of noise present in an ERP (Ghosh Hajra et al., submitted for publication). We agree that future studies should certainly record and publish measures of noise alongside ERP results for additional context. While the level of physical activity also varied between time points, athletes were completing their pre-season fitness exam and active practices during the baseline exam. Nonetheless, there will always be a degree of uncontrollable factors in point-of-care deployment and further studies will help to better understand the relative influences of this factor. The authors also raised a number of excellent technical ERP methodological questions. Given the translational objectives, there will always be methodological differences and challenges, which are equally opportunities to improve this proof of concept. In prior works (D’Arcy et al., 2011, 2016; Real et al., 2014; Ghosh Hajra et al., 2016, 2018a, b; Fleck-Prediger et al., 2018), we and others have explored different methodological factors that are particularly important for individual level analyses (e.g. filter settings, baseline adjustment, compressed sequences, etc.). This work has led to the current optimized implementation, benchmarked to traditional techniques (Ghosh Hajra et al., 2018b), and which will continue to be improved towards the application of individual-level monitoring. It is noteworthy that this application has required literature-derived and experimentally validated methodological modifications from traditional lab-based ERP studies (D’Arcy et al., 2011, 2016; Real et al., 2014; Ghosh Hajra et al., 2016, 2018a, b; Fleck-Prediger et al., 2018). Rosburg and Mager (2019) also raise an important point regarding ERP peak identification related to the P300, and whether it is in fact a preceding P200 component. We suspect the response contains both components, as they are both present when the low-pass filter setting is increased. The P300 label was used for a number of reasons: (i) examination of individual waveforms have shown that the bifurcated P200/P300 waveform is consistent within subjects but varies in relative amplitudes between the two peaks across subjects, and is therefore practically represented by reducing the low-pass filter setting; and (ii) the 10 Hz filter effectively merges these peaks into one, in which the P300 is most consistently and easily represented within the context of standard nomenclature. Importantly, as the brain vital sign framework evolves, the nomenclature and relative roles of the P200/P300 will be more fully characterized in terms of the basic attention metrics. It will be the relative changes within individuals that can be monitored over time that will matter most for the clinical translation to point-of-care. We were excited about the authors’ ideas around future experiments to better understand and improve the brain vital sign framework, particularly in optimizing the underlying ERP technical and research methods. Clinically, this initial implementation enables next steps in continuing to characterize longitudinal concussion-related changes at the individual level and to expand towards other applications where brain vital signs monitoring may be useful (e.g. dementia). Data sharing is not applicable to this article as no new data were created or analysed in this study. Financial support was provided by Mathematics of Information Technology and Complex Systems (MITACS, Grant #IT03240), Natural Sciences and Engineering Council Canada (NSERC), and Canadian Institutes for Health Research (CIHR) for this study. The research study was designed and carried out by the Mayo Clinic Sports Medicine Ice Hockey Research team, partially funded by USA Hockey and the Johannson-Gund Endowment. Some of the authors are associated with HealthTech Connex Inc., which may qualify them to financially benefit from the commercialization of a platform capable of measuring brain vital signs.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.