Cumulative effects of subsequent concussions on the neural patterns of young rugby athletes: data from event-related potentials

ABSTRACT

Our study aimed at detecting a potential cumulative effect of subsequent concussions on the neural activation patterns of young rugby athletes with or without concussion history. Event-related brain potential (ERP) data from 24 rugby players, 22-year-old on average, were retrospectively examined. All underwent a Sport Concussion Assessment Tool (SCAT2) during preseason and an on-site ERP task (P300) following a recent concussion event (<48 hours). Sixteen players suffered at least one concussion in the previous 3 years and eight were without self-reported past concussion. While no differences were reported between groups regarding symptom appraisal on the SCAT2 assessment, ERP revealed significantly decreased P3b amplitude and a trend for increased P3b latency in players who experienced prior concussions. Our data thus support the cumulative effect of concussions on neuroelectric events in young rugby players, highlighting the importance of managing player’s concussion load to reduce the risk of long-term injuries.

KEYWORDS:

• Brain concussion
• electroencephalogram
• event-related potential
• P300 wave

Introduction

Mild Traumatic Brain Injury (mTBI) resulting from sport-related concussions is a major health concern affecting ~1.6 to 3.8 million people every year in the USA (Centers for Disease Control and Prevention, 2019). Among contact sports, rugby has one of the highest concussion rate, particularly among the 5- to 18-year-olds, with tackling being the most common cause of brain injury (Bayt & Bell, 2016; Gardner et al., 2014). A meta-analysis revealed that the incidence of concussions varied considerably between practice levels, with the sub-elite level recording the greatest incidence (2.08/1,000 player match hours) (Gardner et al., 2014). Despite concussion rules developed for immediate control and return-to-play decisions (Chermann et al., 2014), the numbers are not decreasing leading Rugby authorities to request major changes in concussion management based on objective biomarkers.

Diffuse axonal injury, due to the head trauma or rotational acceleration, appears to be a central pathogenic mechanism of concussion (Zetterberg et al., 2019). Consequences seem, however, primarily functional without any apparent structural damage detectable using neuroimaging techniques. Clinical symptoms, including headache, fatigue, attention deficit, memory impairment or anxiety, can develop hours after injury and are usually transient (1–12 weeks). Nevertheless, 10–15% of concussed players still report symptoms and cognitive deficits months or even years after the event (Rice et al., 2018; Zetterberg et al., 2019). Due to their still-developing brains, young players appear particularly vulnerable to repeated trauma or long-term repercussions (Narayana et al., 2019). Moreover, accumulating evidence suggests that repeated head injuries should be of even greater concern because they are associated with an increased risk of developing dementia with features of both chronic traumatic encephalopathy (CET) and Alzheimer’s disease (AD) (Agrawal et al., 2022; Hay et al., 2016; Lee et al., 2019; Mez et al., 2017; Shively et al., 2012). A study of retired athletes found that those reporting three or more concussions during their career had a three-fold increase in self-reported memory complaints and a five-fold increase in mild cognitive impairment (MCI) diagnosis relative to non-concussed athletes (Guskiewicz et al., 2005).

Currently, the Sport Concussion Assessment Tool (SCAT) is the most widely used instrument in the evaluation of concussion by the rugby union (McCrory et al., 2009). However, current methods might be limited because symptom termination might not be indicative of the resolution of underlying injuries. Additional markers, as neuroimaging, fluid biomarkers or eye tracking, are available but not validated against pathological diagnosis (Marion et al., 2020; Zetterberg et al., 2019). In the absence of a validated biomarker for concussion, objective biomarkers, other than symptomatology, are thus required to measure the full extent of an injury and to make optimal return-to-play decisions.

Event-related brain potentials (ERPs) are an attractive candidate. They consist of a series of electroencephalographic (EEG) waves, non-invasively recorded on the scalp, reflecting specific cognitive tasks performed by individuals in response to sensory stimuli. The P300 wave is an ERP that could be induced by an auditory oddball task (two-stimulus discrimination) taking the shape of a large positive wave within a few milliseconds (~250–500 ms). P300 is considered as an objective measure and a clinically useful index of cognitive function with P300 amplitude reflecting the quality of information processing and P300 latency directly associated with cognitive capabilities (Goodin et al., 1994; Polich & Herbst, 2000). The P300 complex is divided into two subcomponents: a “P3a” component generated in the frontal cortex, involved in focal attention and mediated by dopaminergic pathways; a most studied P3b component generated from the temporoparietal junction and posterior parietal cortex, involved in context-updating working memory and decisional processes, mediated by the locus-coeruleus pathway (Bennys et al., 2007, 2011).

Detection of subtle neurocognitive deficits, such as impairment of daily living activity or social function, might be reliably detected using highly sensitive electrophysiological measures such as P300 (Li et al., 2021). A reduction in P300 amplitude was observed in symptomatic athletes who recently (<6 months) suffered sports-related concussion (Dupuis et al., 2000; Lavoie et al., 2004). Bernstein (2002) also showed a significant decrease of P3b amplitude in athletes who suffered at least one concussion within the 8 previous years. This decrease in P3b amplitude/latency was later shown to persist up to three decades after concussion in former athletes (De Beaumont et al., 2009). It suggested that P300 abnormalities being long-lasting, could be cumulative and associated with lower performances on neuropsychological tests of episodic memory and response inhibition. Intriguingly, Gosselin et al. (2006) reported that both symptomatic and asymptomatic athletes showed delayed P300 response challenging the validity of return-to-play guidelines relying on the absence of symptoms only. The purpose of our study was thus to assess the effects of subsequent concussions in young rugby players evaluating P300 parameters and confronting them to SCAT2 baseline performances.

Materials and methods

Subjects

Twenty-four players from a French regional rugby team were referred to our memory clinic (Montpellier hospital) between 2012 and 2013 for a routine P300 examination after a recent concussion event (<48 h). Concussion history was assessed on-site through a medical history interview. Sixteen athletes reported at least one concussion (CH group), resulting in a game withdrawal, over the past 3 years and eight did not report prior concussion (NCH group). All players were free from neurological diseases or physical disabilities. Physician and athletes were not blinded to P300 results, as they were used to make return-to-play decision. Of the 16 players of the CH group, three did not return to baseline after a second P300 examination 1 month later, leading to the decision not to resume playing.

Design

This study was conducted retrospectively from data obtained for clinical purposes, in which medical procedures are reported in the usual way without additional diagnostic procedure. All athletes gave their consent for participation and on-site evaluation. Only registration numbers were employed to maintain anonymity and medical confidentiality. The study was approved by the legal representative of the French Data protection authority (Commission Nationale Information et Libertés) through a declaration of conformity to the MR-004 reference methodology (CNIL reference #: 2223988v0).

Methods

Sport Concussion Assessment Tool

All athletes underwent SCAT2 at preseason. SCAT2 usually combines subcomponents of a clinical assessment, including symptom appraisal (graduated symptom checklist), cognitive evaluation (five-word immediate recall/delayed recall), modified Maddocks questions and neurological screening. Only the total number of symptoms (/22) and their total severity score (/132) were used for analysis because cognitive and balance assessment were removed from our analysis due to missing data.

Event-related potential recording

Neuroelectric activity was recorded from four midline scalp derivations (frontal: Fz, central: Cz, parietal: Pz, occipital: Oz) according to the international 10/20 standards, with, as reference, two linked electrodes attached to the right and left earlobes (A1-A2). The impedances were kept below 5kΩ. EEG activities (sampling rate of 512 Hz) were amplified with a 40,000 gain, processed with bandpass filters of 0.5 to 150 Hz and visualized on the screen of an ERP device (Myoquick Matrix line of Micromed SAS, France). Recording started 100 ms before stimulation, as reference, and continued 900 ms after that. On average, according to subjects, 3 to 10 EEG scans were rejected due to ocular artefacts.

Two-tone auditory oddball task

A classical auditory oddball paradigm was used. The two stimuli were 1000 Hz and 2000 Hz tones (60 dB HL, 100-ms duration), designated as “non-target” and “target” stimuli, respectively, and delivered binaurally via headphones. The presentation rate was random with 20% occurrence of “target” stimuli. Participants were asked to count only “target” stimuli during two trials. Individual recordings corresponding to “non-target” and “target” stimuli were processed online and averaged separately. A classical approach based on automatic peak detection and a spatio-temporal PCA reconstructed from ERP data were applied. The grand average of the ERP was calculated from the two trials for Fz, Cz and Pz electrodes. P300 scalp distribution is defined as the amplitude change over the midline electrodes (Fz, Cz, Pz), which typically increases in magnitude from the frontal to parietal electrode sites. Thus, for the sake of data reduction, as well as reducing the signal bandwidth, these electrodes were used for further analysis. Latency corresponds the delay between sensory stimulus and highest positive peak within a ~ 250–500 ms time window, while amplitude is the maximal value of peaks relative to the pre-stimulus baseline. Latency and amplitude of late subcomponent P3b were used as primary outcome measure for the analysis.

Statistical analysis

Analyses were performed using R (v4.1.1) (R Core Team, 2021). For quantitative variables (age, number and severity of the symptoms, ERP amplitude and latency), normality was examined using Shapiro–Wilk test. Student's t-test (Welch’s for unequal variances), or Mann–Whitney test, according to normality, were used to compare symptoms (number and severity) and ERP parameters (amplitude and latency) between the CH and NCH groups. Cut-offs were computed using expectation – maximization (EM) algorithms for mixtures of univariate normal distributions. An ROC (receiver operating characteristic) analysis was used to determine P3b performances. A predictive formula was built using a generalized linear model (GLM) (binomial logistic regression analysis). Correlation between ERP parameters and symptoms were assessed using Pearson’s or Spearman's $\rho$ according to normality. Significance level was set at p < 0.05.

Results

SCAT2 modalities

Our first main result indicates that young rugby athletes who previously suffered from concussions reported, during their preseason SCAT2 assessment, as many symptoms, with identical severity, as the group without a history of concussion (Figure 1 & Table 1). The group with a concussion history (CH) had an average number of 3.56 ± 2.85 symptoms with a severity score of 7.56 ± 8.43 (Figure 1a). The group with no concussion history (NCH) had 2.63 ± 2.07 symptoms with a severity score of 6.38 ± 6.88 (Figure 1b). No difference was observed between the two groups regarding symptom number (p = 0.37) and symptom severity (p = 1). Age was similar between the two groups (~22-year-old, p = 0.9, Table 1) and no correlation was found between age and the number of symptoms (p = 0.24) or symptom severity (p = 0.57).

Figure 1. Number of symptoms (a) and symptom severity score (b) according to concussion history.

Table 1. Age, SCAT2 results and ERP (P3b) parameters according to the concussion history of the athletes.

	CH (n = 16)	NCH (n = 8)	P-value^a
Age, y	22.06 ± 5.53	21.88 ± 4.32	0.93
SCAT2
Number of symptoms	3.56 ± 2.85	2.63 ± 2.07	0.37
Symptoms’ severity	7.56 ± 8.43	6.38 ± 6.89	1
EPRs
Amplitude, V
P3b Fz	2.95 ± 1.52	5.38 ± 2.21	0.022*
P3b Cz	3.07 ± 1.40	6.64 ± 3.34	0.019*
P3b Pz	3.17 ± 1.09	7.1 ± 2.76	0.005**
P3b Fz/Pz ratio	0.99 ± 0.56	0.8 ± 0.26	0.37
Latency, ms
P3b Fz	341.74 ± 13.27	329.97 ± 14.93	0.066
P3b Cz	336.79 ± 12.98	328.2 ± 16.33	0.17
P3b Pz	337.75 ± 10.24	329.88 ± 14.55	0.14

Abbreviations: CH, concussion history; NCH, no concussion history; y, year; SCAT, sports concussion assessment tool; ERP, event-related potential; Fz, frontal; Cz, central; Pz, parietal recording sites.

a: Student, Welch t-test or Mann-Whitney according to normality.

Event-related potential characteristics

Our second significant finding was that rugby players with a history of concussion who experienced a new event had a significant decrease in P3b amplitude compared to those without trauma history. Individual ERP responses to “target” stimuli were averaged to compute P3b amplitude and latency for each groups (Figure 2a,b). P3b amplitudes exhibited significant decrease at frontal (p = 0.022), central (p = 0.019) and parietal recording sites (p = 0.005), in the CH group as compared to NCH group (Figure 1c & Table 1). Provisional P3b amplitude cut-offs below which athletes could be considered with concussion history could be established at 4.1 (3.7–4.4) for the Fz location, 3.6 (3.4–3.8) for Cz and 6.7 (6.4–7) for Pz (Figure 2c, dotted lines). P3b amplitudes were able to discriminate CH patients from NCH ones with excellent abilities for frontal and central recording sites (AUC: 0.800 and 0.825) and outstanding abilities in the case of parietal recording with and AUC of 0.942 (Figure 2d). P3b latencies at the frontal recording site appeared longer in the CH group (341.7 ms ±13.2) relative to the NCH group (330 ms ±14.9); even if only a statistical trend was observed (p = 0.066) (Table 1).

Figure 2. Grand average ERPs to target stimuli recorded in concussion history group (a) and in group without history of concussion (b). Amplitude, in µv, is represented along Y-axis while latency is along the X-axis and in ms. ERPS are represented with the N1 exogenous followed by endogenous responses: N2, P3a and P3b. (c) P3b amplitude according to recording site and concussion history. (d) ROC/AUC analysis of the ability of P3b amplitudes to discriminate athletes with a history of concussion from without.

Correlation between symptoms and ERPs

No significant correlation was found between number of symptoms or severity and ERP parameters; only a trend was observed between symptom severity and amplitude of P3b under Fz (r = 0.395; p = 0.061) (Table 2 & Figure 3).

Figure 3. Scatter plot of symptom severity and ERP P3b amplitude.

Table 2. Correlation between endorsed symptoms and ERP parameters.

	Amplitude (µV)			Latency (ms)
SCAT2	P3b Fz	P3b Cz	P3b Pz	P3b Fz	P3b Cz	P3b Pz
Number of symptoms	r = 0.27	r = 0.21	r = 0.3	r = − 0.13	r = 0.03	r = 0.04
	p = 0.22	p = 0.33	p = 0.16	p = 0.57	p = 0.91	p = 0.85
Symptom severity	r = 0.3	r = 0.27	r = 0.4	r = − 0.29	r = − 0.09	r = 0.003
	p = 0.16	p = 0.21	p = 0.06	p = 0.19	p = 0.66	p = 0.99

Discussion

Young rugby athletes, with concussion history, experienced significantly reduced ERP P3b amplitudes after a recent concussion event while reporting identical symptoms at preseason (SCAT2) relative to those without history of injury. Our results suggest subtle abnormalities in ERP components that could reveal an objective biomarker of cumulative effects of recurrent concussions in neural networks connectivity and help clinicians to take return-to-play decisions.

Our results corroborate previous works reporting suppressed P300 amplitudes in university football players with multiple-concussion history, compared to those with a single concussion or without this history, and confirm, in rugby athletes the cumulative effects of multiple concussions on P3b (De Beaumont et al., 2007). For three athletes with CH, P3b amplitude was not back to baseline one month after the initial P300 recording thus suggesting effects of longer duration. Considering that these players might be at higher risk to concuss again or for second-impact syndrome (Cantu, 1998; Gerberich et al., 1983), the medical staff took the decision to provisionally exclude them from playing.

P3b latency showed a statistical trend for longer duration at frontal recording site in athletes with a history of concussion that could reflect a decrease in stimulus classification speed (Polich & Herbst, 2000). This result is consistent with increased P3b latencies observed in retired athletes who suffered their last concussion ~30 years earlier (De Beaumont et al., 2009). Our results support the hypothesis that concussion might not be a transient injury for some populations and could impact on the long run the sensitivity to future concussions and the state of neural activity patterns (Broglio et al., 2009, 2011). These prolonged changes could suggest that athletes with concussions do not allocate the same level of attentional resources nor are they able to initiate and monitor their actions in the same way than those without concussion history (Broglio et al., 2011).

No difference in the number or severity of symptom was observed during preseason between the groups with or without concussion history. Our results are somewhat different from previous research in college athletes as a history of concussion was associated with greater baseline symptom and severity appraisal (Bruce & Echemendia, 2004). This discrepancy could be explained by the limited size of our groups as well as the small number and severity of symptoms, close to baseline. It is also possible that players who previously suffered concussions underestimate or minimize their symptoms to enter their next season. The challenge of competition combined with their strong motivation to keep playing could lead them to underreport concussions. In addition, the frequency of preseason clinical assessment, particularly among those who have suffered a concussion, could lead to a habituation effect reducing symptom appraisal. Another variable one should consider is the delay between prior concussion(s) and preseason/baseline assessment. In our case, concussion history could be up to 3 years which may give longer time for symptom recovery. Moreover, SCAT symptoms could remain unnoticed in young players due to their cognitive reserve and only appear later with the ageing process (Satz, 1993). This is consistent with the lack of correlation we found between self-reported symptoms and ERP parameters and the report by Gosselin et al., showing that both symptomatic and asymptomatic athletes exhibit P300 amplitude changes (Gosselin et al., 2006).

Taken together, those results indicate that return-to-play decisions based solely on self-reported symptoms should be considered with care, especially in young athletes with full cognitive reserve and high motivation to play.

These findings are important to guide clinicians in their assessment of performance changes in athletes with concussions, particularly in younger players. Thus, changes in performance assessed during preseason must be interpreted with caution when a player has already suffered a concussion, taking into account the time elapsed between the injury and the assessment. By implementing a very simple and useful test like the P300 that has the ability to detect subtle changes in brain activity, we highlight its important role in managing injury risk in the training environment and game.

Our study presents some limitations. First, no P300 exploration was done during pre-season, we thus cannot exclude that differences could have been present prior to the most recent concussion event, especially for the three players with longer modifications. Future pre-post within-subject design should help untangle this interrogation and determine if previous concussion history exaggerated the consequences of subsequent concussions. Second, the reduced number of participants might limit the extrapolation of our results. Larger follow-up studies, including a concussion-free control group, might be necessary to further confirm our results. Eventually, other SCAT modalities, such as cognitive performances, were unfortunately missing and could have a significant clinical meaning.

Conclusion

While young rugby athletes have similar symptom appraisal during preseason, stimuli-evoked P300 amplitude after a recent concussion event is significantly reduced in players with a history of concussion. This could potentially reflect subtle changes in neural network connectivity. Our results suggest that preseason symptom self-evaluation should be interpreted with care in young athletes with a history of concussion, especially with full cognitive reserve. P300 is an easily implementable test that could reveal itself an objective adjunct to help discriminate athletes with a concussion history and to avoid early return-to-play decisions and potential long-term injuries.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

No external funding was used in the preparation of this study.