What boxing tells us about repetitive head trauma and the brain

Boxing and other combat sports may serve as a human model to study the effects of repetitive head trauma on brain structure and function. The initial description of what is now known as chronic traumatic encephalopathy (CTE) was reported in boxers in 1928. In the ensuing years, studies examining boxers have described the clinical features of CTE, its relationship to degree of exposure to fighting, and an array of radiologic findings. The field has been hampered by issues related to study design, lack of longitudinal follow-up, and absence of agreed-upon clinical criteria for CTE. A recently launched prospective cohort study of professional fighters, the Professional Fighters Brain Health Study, attempts to overcome some of the problems in studying fighters. Here, we review the cross-sectional results from the first year of the project.


Introduction
It is not surprising that the long-term neurological consequences of cumulative head trauma were initially recognized in professional boxers [1]. Th ese athletes are on the receiving end of thousands of blows to the head of varying intensity, in sparring and matches, over many years. Beginning in 1928, when Harrison Martland described the clinical features that constitute what is now known as chronic traumatic encephalopathy (CTE) [1], many articles have been written about the neurological consequences of boxing in both amateurs and professionals. Yet, there are still signifi cant gaps in our knowledge of the spectrum of chronic injuries that can occur in combat sports.
It is worth asking what can we achieve by studying those in combat sports, both boxing and the increasingly popular sport of mixed martial arts (MMA). In the broadest sense, these sports provide a human model to study the evolution of CTE, including understanding the natural history and modifying factors of the disease, along with identifying biomarkers. Th us, what is learned from combat sports may be applicable to various settings in which repetitive head trauma can occur, including other contact sports and the military. A more specifi c goal would be gathering knowledge that can be applied to improve long-term safety of boxing and MMA such as developing guidelines that can be used by regulatory agencies and the athletes themselves to better monitor their brain health.
In a review of the current body of literature on boxing and the brain, several caveats require mention. Th e sport itself has changed over the years, making comparison of studies from diff erent decades diffi cult [2]. Current fi ghters tend to have shorter careers and fewer career bouts and benefi t from rule changes such as limiting championship fi ghts to 12 rounds (instead of 15), use of larger glove size, and increased medical supervision. Moreover, a number of methodological issues cloud the interpretation of prior work: (a) reliance on retrospective or crosssectional design, (b) lack of, or inadequate, control group, (c) evidence based on small sample sizes or case reports, and (d) selection bias of boxers who are symptomatic or have an extraordinarily high amount of exposure.
With the goal of overcoming the methodological limita tions of prior research and addressing some of the important unanswered questions in the fi eld of cumu lative head trauma, we initiated a prospective cohort study of active and retired fi ghters in 2011, termed the Professional Fighters Brain Health Study (PFBHS). Whereas several excellent contemporary articles review the neurological eff ects of boxing, this article (a) will focus on how the current literature on fi ghters can inform us about the clinical and imaging features of CTE and (b) will describe the fi rst-year results from the PFBHS [3][4][5][6]. For the purpose of this review, we will use the term CTE to subsume a number of terms used in the literature to denote chronic neurological fi ndings in boxers, acknowledging that there is no way to know whether these individuals actually harbor the pathological changes we now attribute to CTE.

Epidemiology
A fundamental, but elusive, issue is just how common CTE is among those exposed to recurrent head trauma. In the absence of accurate clinical criteria or a large enough clinicopathological study group of symptomatic and asymptomatic individuals, this question cannot be answered satisfactorily. A commonly cited study of exprofessional fi ghters who were licensed to box in the UK from 1929 to 1955 found that 17% of subjects had CTE and that 40% of the remaining boxers had disequilibrium, dysarthria, or alcoholism [7]. No methodologically sound studies of incidence or prevalence have been published since. On the other hand, the risk factors that have been consistently linked to chronic brain injury across prior work are older age and higher levels of exposure to head trauma [7][8][9][10].
Among the major challenges in the fi eld of traumatic brain injury is being able to quantify exposure. In the absence of a direct measure of the cumulative trauma each subject is exposed to, several potential surrogates such as number of fi ghts, fi ghts per year, number of knockouts (KOs), and years of fi ghting have been used. However, each of these variables may actually have a slightly diff erent infl uence on the development of CTE. Number of fi ghts, for example, may act as a proxy for amount of training. Some have postulated that the eff ects of repeated blows to the head that occur during sparring, even at a subconcussive level, may play as important a role in causing cumulative brain injury as the match itself [11]. On the other hand, KOs may refl ect the more severe end of the spectrum of mild traumatic brain injury. Whereas the number of KOs sustained in sanctioned professional fi ghts can be tracked from commonly available records, the number of KOs that may have occurred at other times is harder to trace. Furthermore, frequency of fi ghting may be a complementary variable that requires consideration; fi ghting more frequently may reduce the time the brain has to fully recover from prior trauma and be a risk factor that interacts with number of fi ghts.
Increased exposure to head trauma in and of itself does not appear to be suffi cient to cause CTE. As in other neurodegenerative conditions, genetic factors may modify the risk of CTE. Some, but not all, studies have suggested that the apolipoprotein E 4 allele increases the risk of Alzheimer's disease in individuals with a history of head trauma [12][13][14]. In a study of boxers, Jordan and colleagues [15] demonstrated an increased risk of CTE in those who are E 4 -positive, although the study was retrospective in design.

Clinical features
A consistent picture of the clinical features of CTE in boxers has emerged over the years. However, whether these signs and symptoms develop in predictable stages is debated [4,5,16]. Th ere does appear to be agreement that behavioral changes, ranging from aff ective disorders to paranoia, irritability, and aggression, occur frequently as an early symptom [4,5,9,[17][18][19]. Progressively, cognitive dysfunction becomes noticeable with additional motor features such as dysarthria, parkinsonism, and gait disorder. Th ese clinical observations in boxers are not too diff erent from what was reported recently in a large clinicopathological study of athletes exposed to head trauma, in which headache, depression, and memory complaints were present in the early stages of CTE, followed by diffi culties in gait and dysarthria (which was associated with motor neuron disease) and parkinsonism in the later stages [20]. Th e clinical information in that study was obtained retrospectively, and the informant may have reported only those symptoms and signs that were strikingly apparent. It is possible and perhaps likely that prospective and precise measurement of motoric and other neurological function reveals a slightly diff erent sequence of signs and symptoms that occur with CTE associated with unarmed combat sports.
In regard to the specifi c neuropsychological domains eff ected in CTE, psychometric testing of former and active professional boxers has most frequently demonstrated defi cits in memory, information-processing speed, fi nger-tapping speed, complex attentional tasks, and frontal-executive functions [5,9]. In contrast to professional fi ghting, amateur fi ghting has rarely been shown to result in any long-term changes in cognitive function [21]; longitudinal studies did not show any eff ect of boxing on psychometric results in amateurs even up to 9 years [22]. Th e use of psychometric measures as a means to screen for developing CTE in active fi ghters does have its hazards. Performance on any single testing session, particularly in proximity to a competition, can be infl uenced by a number of factors, including the acute eff ects of recent sparring, rapid weight loss and dehy dration, pre-bout anxiety, and suboptimal eff ort. Moreover, the precision of psychometric tests used in this population may not be adequate to detect subtle changes given the variability of the tests themselves.

Imaging
Virtually every sort of imaging modality, ranging from pneumoencephalography to positron emission tomography (PET) scanning, has been studied in boxers [23]. Certainly, given its wide availability, lack of radiation exposure, and superior sensitivity over computed tomography imaging to detect subtle structural changes, magnetic resonance imaging (MRI) scanning has become the favored imaging modality for the evaluation of brain injury from head trauma. A number of MRI fi ndings recognized by visual inspection have been related to boxing [24]. Several of these fi nd ings, including lateral ventricular size, dilated peri vascular spaces, and diff use axonal injury, were associated with some measure of exposure, such as number of professional bouts or years of fi ghting. Moreover, studies using measures of diff u sivity on diff usion tensor imaging have shown changes at a group level between boxers and non-fi ghting groups [25][26][27].
Functional imaging has also been explored as a means of detecting brain injury that might not be seen on structural scanning. Studies employing single-photon emission tomography (SPECT) and PET imaging have reported diff erences between boxers and controls [28,29]. Despite a small sample size, there was a trend toward a relationship between number of fi ghts and number and extent of PET abnormalities.
Th e application of what we know of imaging in fi ghters, at the moment, is limited. Most published imaging studies are cross-sectional and do not include a clinical outcome, so the signifi cance of any one fi nding in predicting subsequent clinical change is unknown. In addition, the composition of the samples studied is usually not random, which may result in a bias of having more clincially symptomatic individuals participate. A number of other confounding factors in imaging studies have been reviewed by Moseley [23]. Although many states require MRI scans as part of licensing (and some states require the imaging to be repeated periodically), there is actually little evidence from well-designed studies to determine how to use the information to make decisions on fi tness to fi ght or the value of these measures in protecting fi ghter safety.

Professional Fighters Brain Health Study
Th e PFBHS is a longitudinal study of active professional fi ghters (boxers and MMA fi ghters), retired professional fi ghters, and age/education-matched controls. Th e main objective of the PFBHS is to determine the relationships between measures of head trauma exposure, along with other potential modifi ers, and changes in brain imaging and neurological/behavioral function over time. Th e study is designed to extend a minimum of 5 years, and an enrollment of more than 400 boxers and mixed martial artists is projected. Participants undergo annual evaluations to include 3-T MRI scanning, computerized cognitive assess ments, speech analysis, surveys of mood and impul sivity, and blood sampling for genotyping and exploratory biomarker studies. Information is collected on demographics, educational attainment, family and medical history, previous head trauma (whether related or unrelated to athletic activities), prior involvement in other contact sports, and their amateur fi ghting history. Th e fi ghters' professional record is obtained from commonly cited websites (boxrec.com [30] for boxers and mixedmartialarts.com [31] and sherdog.com [32] for MMA fi ghters) to determine number of years of professional fi ghting, number and outcome of professional fi ghts, number of rounds fought, weight class of each fi ght, frequency of professional fi ghting, and number of times knocked out (KOs and technical KOs). A composite fi ght exposure index was developed as a summary measure of cumulative traumatic exposure [33].
Several cross-sectional analyses have been performed on the baseline data obtained from the PFBHS to examine the association between fi ght exposure and various imaging measures. Repeated measures analysis of variance was employed to test the association between the outcome variables and fi ght exposure variables. Guided by the cutpoints (that is, tree branch splitting values) and deviance reduction values from the regression trees, we defi ned and tested fi ght exposure as follows: linear eff ect of total number of professional fi ghts, linear eff ect of total number of years of professional fi ghting, a threshold eff ect with brain volume reduction estimated separately for less than 5 years of professional fi ghting versus at least 5 years, and an exposure composite score as a function of number of professional fi ghts and number of professional fi ghts per year. In each model, we included the type of fi ghter (boxer or MMA fi ghter) and an interaction term for the type of fi ghter with the other exposure variable. Given the exploratory nature of this study, a signifi cance level of 0.05 was used to test the signifi cance of the regression coeffi cients of the exposure variables; no adjustments for multiplicity were applied. A secondary aim was to test for associations between imaging measures and cognitive test scores and between fi ght exposure and cognitive test scores. Generalized linear models were constructed with cognitive scores as the dependent variables and brain volume or fi ght exposure variables as the independent variables of interest. All analyses were adjusted for age (treated as a continuous variable), education (defi ned as no college-level versus some college-level), and race, which was defi ned as (a) Caucasian, (b) African-American, or (c) other (Asian, Pacifi c Islander, American Indian, or Alaskan Native).
Results from the baseline evaluations revealed fi ndings that support and extend previously published work. Complete data on 239 subjects -104 boxers and 135 MMA fi ghters -are currently available. Th e fi ghters' ages ranged from 19 to 43, and the median was 28.3 years. Close to 52% of the subjects had a high school education or less, and 48.2% had at least some college-level education. Th e mean total number of years of professional fi ghting was 4, and the median total number of professional fi ghts was 11 ( Table 1).
As might be expected, increasing exposure to head trauma, as measured by either number of professional fi ghts or years of professional fi ghting, was associated with lower volumes of several brain regions. Perhaps the most consistent relationship between exposure variables and brain volume was seen in the caudate and, less so, in the putamen [34]. Interestingly, for caudate and amygdala volumes, there was no eff ect of increasing number of years of professional fi ghting up to 5 years. However, above 5 years, there was a 1% reduction in caudate volume per additional year of professional fi ghting (P <0.001) (Figure 1). Th is raises the possibility that the relationship between fi ght exposure and reduction in brain volume is not linear; one might predict that a sequence of pathophysiologic changes occurs with repeated head trauma and that actual drop-out of neurons (and thus reduced volume) comes in a delayed fashion.
Similar associations between exposure and MRI measures of diff usivity and resting-state connectivity are seen. Like previous investigators, we found a signifi cant relationship between number of fi ghts and mean diff usivity values in the posterior corpus callosum. In addition, the number of times a fi ghter has been knocked out in his career predicted increased longitudinal and transverse diff usivity in white matter and subcortical gray matter regions including corpus callosum, cingulate, pericalcarine, precuneus, and amygdala, leading to in creased mean diff usivity and decreased fractional aniso tropy values in the corres ponding regions in ROI analysis [35]. Preliminary analysis of resting-state functional MRI from a left posterior cingulate cortex seed showed that greater number of fi ghts and KOs was associated with more impairment in the functional connectivity in anterior cingulate and cingulate gyrus (Figure 2).
Early results from a limited computerized cognitive battery found that only speed of processing was related to volume and exposure. Decreasing volumes of the thalamus, amygdala, left caudate, and hippocampus were associated with lower scores on speed of processing measures (Figure 3). On the other hand, processing speed was related to exposure to head trauma only at the extremes of exposure. Th e fact that the association between exposure and processing speed was seen only between highest and lowest quartile is consistent with what is seen in other neurodegenerative diseases; the clinical expression of underlying pathology may not appear in a measurable way until a substantial amount of structural damage has occurred.
Level of education may have a modifying eff ect on the relationship between exposure and structural and cognitive changes. In the PFBHS, fi ghters with a high school education or less showed negative associations between fi ght exposure (number of fi ghts and years of fi ghting) and cognitive tests scores (Figure 4). Th e relation ship between brain structure volume and exposure did not diff er on the basis of education. Th ese results are interpreted as putatively showing a protective eff ect of education on functional, but not structural, integrity in fi ghters.  Th e roles of several factors that might infl uence exposure to head trauma in fi ghters have been examined. Diff erences in the type of fi ghting on volumetric measures were seen. Boxers, in general, had lower thalamic and hippocampal volumes than MMA fi ghters and had worse scores on diff usion measures. However, both groups showed a negative association between exposure and volume or diff usivity.
On the other hand, in an initial assessment, the weight of the fi ghter did not infl uence volumetric results. Specifi cally, the interaction between weight class and fi ght exposure did not signifi cantly predict brain volume. For the caudate, there was a trend for an interaction between the number of professional fi ghts and weight class (P = 0.051). For lower weight classes, the relationship between reduction in caudate volume with increasing number of fi ghts is greater than for higher weight classes.
Th ere are several limitations to consider in interpreting the cross-sectional fi ndings from the PFBHS. Th ough all the analyses were adjusted for age and education, we are only now recruiting an age-and education-matched control group, and so we did not have a control group for comparison. Th us, whether the associations between measures of exposure and imaging and clinical outcomes are related to the exposure or other factors is not clear. Th e study group was not a random sample of fi ghters; participants were self-selected and may be less skilled or more susceptible to be knocked out. However, in comparison with all those who fought in Nevada over the same year, subjects in the PFBHS diff ered only in their slightly younger age and their slightly smaller number of professional fi ghts but not in winning percentage or times knocked out. As mentioned above, all of our measures of exposure to head trauma are indirect and may not truly refl ect the actual degree of head trauma each subject experienced.   Figure 4. Estimated psychomotor speed scores after adjustment for age and race. Scores are plotted against total years of fi ghting for fi ghters with a high school (HS) education or less (n = 73) versus those with more than a HS education (n = 75) (P = 0.021).

Conclusions
It is generally conceded that there is still much work to be done in CTE, understanding its natural history, determin ing its risk factors, developing diagnostic methods including predictive biomarkers, and ultimately discovering therapeutic measures. Information gathered from epidemiological studies in groups exposed to repetitive head trauma, such as those engaged in combat sports, may guide us in the directions needed to answer the many outstanding questions of CTE. From the established literature on the brain eff ects of boxing (much of which has design limitations) come a picture of the clinical features of CTE and the recognition that greater exposure to head trauma is associated with increased risk of long-term neurological disease and that a variety of imaging fi ndings can be seen in fi ghters. Th ere are a number of large initiatives, either already launched (such as the PFBHS) or in the planning stage, to more rigorously study the eff ects of repetitive head trauma in both the sports or military arena. Learning from the experience and methods used in investigating other neurodegenerative disease, such as Alzheimer's or Parkinson's disease, we hopefully accelerate our knowledge and treatment of CTE.