Pittsburgh Brain Health Initiative (BHI): protocol and methods for an observational study of cognitive function in former professional football players and controls Authors: David O Okonkwo, Michael W Collins, Anthony P Kontos, Oscar Lopez, James M Mountz, Stephen R Wisniewski, Kathryn L Edelman, Steven Benso, Cyndi Holland, Sue R Beers, Ryan J Soose, Tina Harrison, Anne Mucha, Ava Puccio, Mike Mancinelli, Allison Borrasso, Bedda L Rosario, Charles Laymon, Brian Lopresti, Davneet Minhas, Vikas Agarwal, Sarah Ostop, Oussama Khalifa, Sarah Berman, Beth Snitz, Shawn R Eagle Published: 2025 ABSTRACT AbstractIntroductionLong-term brain health profiles following exposure to repetitive head impacts and/or concussions in contact sports are a public health focus and the subject of a national debate. The true prevalence rates of mild cognitive impairment (MCI) or neurobehavioural dysregulation are unknown in the nearly 20 000 current/living former professional football players. Here, we describe the procedures and methodology of the prevalence study of cognitive function in former professional football players from the Brain Health Initiative at the University of Pittsburgh. The objective is to define the prevalence of normal cognitive function versus neurodegeneration in former professional football players through clinical, neuroimaging and biomarker assessments.Methods and analysisParticipants include former professional football players aged 29–59 years at study onset who played a minimum of three professional football games in three professional seasons and non-exposed controls. Participants are recruited by two mechanisms, a random and non-random sample. The full study protocol includes a 3–4-day, multidomain assessment (eg, neurological, neurocognitive, psychiatric, sleep, vestibular, orthopaedic and cardiovascular) for neurodegenerative disease and overall health and function, including MRI, positron emission tomography scans, analysis of blood plasma and cerebrospinal fluid, neurocognitive assessments, applanation tonometry, overnight sleep study and informant interview. A multidisciplinary clinical panel conducts a blinded diagnostic consensus conference to adjudicate the presence of MCI and/or traumatic encephalopathy syndrome, which serve as the study’s primary and secondary outcomes, respectively. Point prevalence of these for both the exposed and unexposed cohorts will be calculated as the primary statistical analysis.Ethics and disseminationThe University of Pittsburgh Institutional Review Board approved the study prior to recruiting human subjects (protocol numbers STUDY19010008: sIRB - Brain Health Initiative (Part 1) and STUDY19030211: sIRB - Brain Health Initiative (Part 2)). The results will be disseminated in peer-reviewed journals and as presentations at national and international scientific conferences. STRENGTHS AND LIMITATIONS OF THIS STUDY This study combines both a random and non-random sample of former professional football players allowing for comparison between players who were self-selected and those who were selected at random. This study combines both a random and non-random sample of former professional football players allowing for comparison between players who were self-selected and those who were selected at random. The control group was designed to most closely match demographics (age, race, socioeconomic status, body habitus, etc.) of the players in the absence of collegiate or professional football exposure. The control group was designed to most closely match demographics (age, race, socioeconomic status, body habitus, etc.) of the players in the absence of collegiate or professional football exposure. In some cases, data collection of imaging and physical function was limited due to body habitus and claustrophobia. In some cases, data collection of imaging and physical function was limited due to body habitus and claustrophobia. The time requirement of participation made it more challenging for former players who were employed and/or had family commitments to participate. The time requirement of participation made it more challenging for former players who were employed and/or had family commitments to participate. Introduction Long-term brain health profiles following exposure to repetitive head impacts and/or concussions in contact sports are a public health focus and the subject of a national debate.1 Participation in sports provides significant benefits to youth including physical health, social development and leadership skills that promote long-term health and wellness.2 Nevertheless, there is a concern about the long-term implications of brain health for participants of numerous contact sports, including football.3 Since the release of the first case report of chronic traumatic encephalopathy (CTE) in a former National Football League (NFL) American football player in 2005,4 research investigating trauma-related neurodegeneration has markedly risen.1 Trauma-related neurodegeneration was initially described in professional boxers, with a prevalence as high as 47% in boxing careers longer than 10 years.5 6 Initially termed ‘punch drunk syndrome’ in 1928 and, later, CTE in 1949,7 trauma-related neurodegeneration has received intense attention with its detection in former professional American-style football (hereby simply referred to as ‘football’) players.4 8 CTE is a neuropathological disease that can only be diagnosed at autopsy.9 Traumatic encephalopathy syndrome (TES) is the purported in vivo clinical manifestation of underlying CTE, which was first described with consensus criteria from the National Institute of Neurological Disorders and Stroke (NINDS) in 2021.10 The core clinical features of TES are cognitive impairment and/or neurobehavioural dysregulation. While recent studies have identified higher rates of self-reported mild cognitive impairment (MCI) than expected in former professional football players,11 the true prevalence rates of MCI or neurobehavioural dysregulation are unknown in the nearly 20 000 current/living former professional football players. Other available data from postmortem studies of former football players are limited by recall and selection bias.9 12 Statistically valid prevalence determination will form the foundation for powering and statistically analysing future studies of preventative, diagnostic, prognostic and treatment modalities for brain health regarding football and other contact sport exposure. Here, we describe the procedures and methodology of the prevalence study of cognitive function in former professional football players from the Brain Health Initiative (BHI) at the University of Pittsburgh. The objective is to define the prevalence of normal cognitive function versus neurodegeneration in former professional football players through clinical, neuroimaging and biomarker assessment. The specific aims of this study are the following: Specific aim 1 Establish the prevalence of cognitive impairment and neurodegeneration (ie, TES) through blinded review of a comprehensive assessment of clinical (cognitive, neuromotor, behavioural function), neuroimaging (PET, MRI) and biomarker (neurodegenerative disease metrics in blood and cerebrospinal fluid) measures in former professional football players using a random and non-random sample of eligible former players. Specific aim 2 Compare the prevalence and characteristics of neurodegeneration in former professional football players to unexposed groups ((1) friend and brother controls and (2) community controls) using the comprehensive assessment. Specific aim 3 Analyse factors (exposure to contact sports, concussion history, substance abuse, psychiatric disorders, cardiovascular disease, etc.) associated with increased cognitive impairment and neurodegeneration in former professional football players and unexposed groups ((1) friend and brother controls and (2) community controls). Methods and analysis The BHI received regulatory approval from the University of Pittsburgh Institutional Review Board on 18 October 2019 (STUDY19010008: sIRB - Brain Health Initiative-Part 1). This approval covered all study procedures outlined below apart from AV1451 PET imaging. AV1451 imaging was submitted under a second regulatory protocol to facilitate ease of Food and Drug Administration (FDA)/Investigational New Drug (IND) monitoring; approval was received on 1 November 2019 (STUDY19030211: sIRB - Brain Health Initiative-Part 2). The first participant, a community control, was enrolled on 18 November 2019. Due to challenges related to the global COVID-19 pandemic, all study-related procedures were paused indefinitely. BHI was subsequently registered on clinicaltrials.gov on 1 January 2021 (NCT04708613), and the first former football player participant was enrolled on 8 February 2021. Because the rates of MCI and TES in populations under 60 years of age are unknown,13 14 a pre-planned interim analysis on the first 50 controls and 50 former players was performed mid-2024 to conduct a sample size re-estimation using preliminary data. The interim analysis determined that the total number of participants needed to achieve statistical power is 300. At press, 204 participants have completed the study and enrolment is ongoing. The projected end date for the study is 2027, with data collection concluding late 2026. Participants Participants include former professional football players aged 29–59 years at study onset (year 2020) who played a minimum of three professional football games in three professional seasons. We selected this ‘younger’ age range as the rate of MCI and TES is not well understood in the general population below 60.13 14 Participants are recruited by two mechanisms, a random and non-random sample. Recruitment of random-sample former professional football players is conducted via collaboration with the Football Players Health Study (FPHS) at Harvard University.15 The FPHS includes a cohort of former professional football players who played since 1960. To ensure accurate and valid prevalence estimates, potential participants include those randomly selected from a list of retired players 29–59 years old and who played a minimum of three professional football seasons. Stratified random sampling is used to minimise potential self-selection bias in the professional football group. Strata were defined based on age (29–44 years and 45–59 years), race (African American vs other races) and football position (offensive/defensive line, linebacker/quarterback/running back, defensive back/tight end/wide receiver/special teams, kicker/punter) at a ratio of 3:3:3:1, respectively. The age strata were chosen to evenly distribute players into ‘younger’ and ‘older’ cohorts within the broader study’s inclusion criteria. Player positions were grouped based upon typical exposures to repetitive head impacts in practice and competition. A simple random sample is obtained from each stratum and former players are contacted by staff at Harvard University in the random order in which they were generated. After a brief eligibility confirmation and with verbal consent, interested participants are transferred to the study team at the University of Pittsburgh via a secure Health Insurance Portability and Accountability Act (HIPAA)-compliant and limited-access database available only to trained study staff, which is designed to facilitate study screening, data collection and confidential participant coordination. A coordinator at the University of Pittsburgh then contacts the player, completes a full eligibility screening and schedules the player for participation if eligible. Not all players in the random sample choose to participate. We also enrol a non-random cohort of former professional football players. This cohort is comprised of former players who approach the research team at the University of Pittsburgh with interest in participating. The same screening process as the random sample is conducted, after which eligible players are invited to participate. The strategy for this non-random cohort stems from a phenomenon observed in a publication by Weinstein et al,16 where the importance of non-randomised comparisons of self-reported outcomes may be important in understanding the results. In addition to the former player groups, we also enrol a control group comprised of mostly friends or brothers of the former players. Following former player screenings, eligible players are asked if they have a male sibling, relative or friend who might be interested in participating as an unexposed control volunteer. This is not required. The rationale for including a male sibling/relative or friend unexposed to football beyond high school is that they would have similar genetic predispositions (if a relative) and/or theoretically similar sociocultural experiences. Control referrals are also assessed for eligibility prior to scheduling in the same way former players are screened and scheduled. In addition to friend and brother controls, a small supplemental community control group is recruited through advertising and the University of Pittsburgh research registry. Inclusion and exclusion criteria for both groups are detailed in table 1. Table 1Inclusion and exclusion criteria for both groupsInclusion criteriaExclusion criteriaFormer professional football players (fPFP)fPFP aged 29–59 years at study onsetPrior history of strokePlayed, at a minimum, three professional football games in each three professional seasonsNon-fluent English speakersNo known history of severe traumatic brain injury (defined as Glasgow Coma Scale score <9)Prior history of unmanaged or untreated psychosis and/or major neurological disorder* that would interfere with valid neuropsychological testingControlsAge 29–59 years at study onsetPrior history of stroke or major neurological disorder*No history of organised football beyond high school (if at all)Non-fluent English speakersNo known history of severe traumatic brain injury (defined as Glasgow Coma Scale score <9)Prior history of unmanaged or untreated psychosis and/or major neurological disorder* that would interfere with valid neuropsychological testing*Including diagnosed schizophrenia, bipolar disorder, amyotrophic lateral sclerosis, multiple sclerosis, Alzheimer’s disease, Parkinson’s disease and frontotemporal dementia. We did not exclude for pre-existing depression and/or anxiety disorders. Inclusion and exclusion criteria for both groups Including diagnosed schizophrenia, bipolar disorder, amyotrophic lateral sclerosis, multiple sclerosis, Alzheimer’s disease, Parkinson’s disease and frontotemporal dementia. We did not exclude for pre-existing depression and/or anxiety disorders. Former players and controls participate in parallel, and up to two participants are enrolled per week for 3–4 months out of the year. Informed consent is completed on day one prior to any study procedures. This involves thoroughly reviewing the study agenda and discussing each study procedure in detail along with risks, benefits and payments. The participant is encouraged to ask questions during this process. Typically, the informed consent process takes 30–60 min. If a former player does not refer a control volunteer, the second slot is filled with a different participant. Full confidentiality is maintained in these instances with staggered transportation and participation schedules. An example agenda for participant testing can be viewed in online supplemental table 1. Patient and public involvement We engaged with key stakeholders to assist in the design of this study and its dissemination. First, we assembled a ‘Steering Committee’ comprised of multidisciplinary clinicians from external institutions who are considered experts in the fields of neuropsychology, neurology, blood biomarkers, neuroimaging, sports medicine and/or neurodegenerative disease. Committee members are not co-investigators of the study but provided their unbiased recommendations for how to best design and operationalise the study. The steering committee also included former NFL players who assisted by improving the impact of study findings by ensuring we were capturing data related to issues most relevant to the former player community. Study design and protocol Consented participants meeting inclusion and exclusion criteria participate in the study. Participation includes either the full protocol (3–4 days of assessments) or a condensed version of the protocol (2–3 days of assessments). The protocol version (full vs condensed) is determined both by participant preference and principal investigator discretion. The condensed version includes all assessments of the full version with the exception of PET imaging, lumbar puncture and applanation tonometry. The condensed version was created for a variety of reasons, including limited information gained from some assessments and high likelihood of participants declining to undergo certain procedures. Most of the study assessments are available in current clinical practice (ie, health and physical exam, vestibular/ocular motor testing, sleep assessment, cognitive testing). All risks and discomforts associated with assessments are thoroughly reviewed with each participant at consent and on an ongoing basis throughout participation. Adverse events associated with unlikely or rare risks are documented in an AE log and reported to the Principal Investigator and to the University of Pittsburgh Institutional Review Board in accordance with reporting requirements. If a participant has an incidental finding during any part of this testing (ie, obstructive sleep apnoea or high blood pressure), the clinical staff provides clinical follow-up information for the finding outside of the study as applicable. Any emergent findings will be addressed immediately; non-emergent findings will be relayed to the participant at their study debrief or sooner. Specific assessments included within each domain are detailed in figure 1, and an overview of procedures is detailed below: Health and physical exam (included in both the full and condensed protocol) List of outcome assessments by domain. H&P, Health and Physical Exam. Participants undergo a comprehensive history and physical exam by a physician or advanced practice provider. The assessment includes family history, social history, medical history, psychiatric history, drug history and current medications. Current symptomatology is discussed during this procedure. The assessment also includes various psychological health, drug and alcohol questionnaires and screening tools. Vital signs are performed and recorded. As part of the comprehensive health assessment, the Unified Parkinson’s Disease Rating Scale (UPDRS) and Alcohol Use Disorders and Identification Test are performed.17 The UPDRS is a neurological and motor exam which assesses motor movements, including rigidity, tremor and slowness.18 An electrocardiogram (EKG) is also performed to understand current cardiovascular health, which is an important risk factor for neurodegenerative disease.19 Sports medicine assessment (neuropsychological, vestibular, ocular motor, exertion) (included in both the full and condensed protocol) Participants undergo three separate evaluations with sports medicine clinicians. The evaluations include neuropsychological, vestibular/ocular motor and physical function. Sports medicine evaluation and neuropsychological testing The sports medicine assessment includes several tests and questionnaires that are completed using computers (including the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT))20 and via interview. Questions are asked about sport and concussion history, and a detailed traumatic brain injury (TBI) history (including the validated Ohio State University TBI (OSU-TBI)21 history) is recorded by the research staff. Several assessments and questionnaires regarding daily symptoms are also conducted with participants during their neuropsychological evaluation. These include the Post-Concussion Symptom Scale,22 the Concussion Clinical Profiles Screening,23 the Barratt Impulsiveness Scale24 and Buss Perry Aggression Questionnaire.25 A trained neuropsychologist oversees data collection by certified staff who interview the participant to assess for current concussion-related impairments and symptoms. Vestibular/ocular motor evaluation and testing Vestibular and ocular dysfunction is commonly observed in patients with neurodegenerative disease.26 27 These assessments are conducted by a vestibular physical therapist and include tests and questionnaires that focus on physical symptoms of TBI and include assessments on dizziness (Dizziness Handicap Inventory, Visual Vertigo Analogue Scale),28 29 balance (Modified Balance Error Scoring System, Functional Gait Assessment Dynamic Gait Index),3032 ocular motor coordination (Vestibular Ocular Motor Screening)33 and ocular (accommodation and ocular alignment/cover tests) measures. Exertion evaluation Current physical fitness can impact risk for neurodegenerative disease and/or poorer cognitive performance.19 Participants engage in a dynamic movement assessment which takes 15–30 min to complete and includes four different exercises—a 3 min aerobic step test, Sit to Stand Test, Ball Toss and Arrow Agility.34 Each participant’s blood pressure is tested pre-exertion and post-exertion. Participants with high blood pressure, those with relevant EKG abnormalities or those who have not been cleared for exercise by their physician do not participate in the exertion component. The exertion protocol is modified as needed to accommodate orthopaedic injuries. This judgement will be made on a case-by-case basis by the certified staff members at the sports medicine clinic. The exertion portion of the sports medicine visit is completed by a physical therapist. PET imaging (full protocol only) PET data acquisition Subjects enrolled in the full protocol undergo PET imaging using [11C]PiB for evaluation of Aβ and [18F]AV-1451 for tau at the University of Pittsburgh PET Center. The synthesis [11C]PiB has been previously described35 36 and is in accordance with US FDA IND 72200. [18F]AV-1451 is prepared according to methods described in US FDA IND 123396. PET imaging data are typically collected on day 2 of the study session. Participants undergo PET imaging sessions sequentially starting with [11C]PiB followed by [18F]AV-1451. A period of 15 min is allotted between the end of the 70 min [11C]PiB scan and the injection of [18F]AV-1451, during which time subjects are removed from the scanner for a comfort break and to allow additional decay of residual 11C (t1/2 = 20.4 min) remaining from the [11C]PiB injection. All [18F]AV-1451 scanning is performed using a Siemens Biograph mCT Flow TrueV PET/CT scanner (mCT) (Siemens Medical Solutions USA, Malvern, Pennsylvania). For scheduling efficiency, [11C]PiB scanning is split between the mCT and a Siemens ECAT HR+ scanner (HR+). For mCT scans, a low-dose non-diagnostic CT is acquired prior to the start of the PET scan and is required for attenuation and scatter correction during image reconstruction. For HR+ scans, similar additional data are acquired in a transmission scan using the scanner’s built-in 68Ge/68Ga rod sources. All PET data are acquired in frames of 5 min duration spanning the window of 50–70 min post injection for [11C]PiB and 80–100 min for [18F]AV-1451. Scans are reconstructed using the manufacturer’s software and include corrections for deadtime, decay, attenuation and scatter. PET data analysis The multiframe PET image sets are inspected for frame-to-frame motion and, if necessary, corrected using PMOD software (PMOD Technologies LLC, Industriestrasse 26, 8117 Fällanden, Switzerland). Each image set is averaged over frames into a single frame representing tracer uptake over the windows listed above. A separately acquired T1 MRI image is used for anatomical reference in the analysis of the PET data. It is reoriented along the anterior-posterior commissure and the single-frame PET image is rigidly registered to it using PMOD. The T1 MRI is then processed using FreeSurfer (FS) software (http://surfer.nmr.mgh.harvard.edu/) to produce an atlas (Desikan Killiany atlas)37 of anatomical regions matched to the MRI.3842 In our implementation, we replace the FS striatum with the more finely detailed parcellation from the Imperial College London Clinical Imaging Centre atlas.43 44 Our final result is an atlas (DKCIC atlas) aligned to each individual’s PET image. PMOD is used to extract regional concentrations from the PET scan for each of the DKCIC atlas regions. Custom quantitation regions are assembled by performing a volume weighted average of concentrations in various DKCIC regions. For [11C]PiB, the quantitation regions are the following: anterior cingulate, anterior ventral striatum, superior frontal, orbitofrontal, insula, lateral temporal, parietal, posterior cingulate and precuneus. Additionally, tracer concentration is determined in a global region consisting of the weighted average of the individual quantitation regions. For [18F]AV-1451, quantitation regions correspond to the six Braak regions. Additionally, [18F]AV-1451 tracer concentration is extracted for a meta-region consisting of parts of the Braak 1, 3 and 4 quantitation regions.45 For both [11C]PiB and [18F]AV-1451, specific uptake value ratios (SUVR) are produced by normalising quantitation-region tracer concentrations to the tracer concentration of cerebellar grey matter. Amyloid status (Aβ(+/−)) is assessed using [11C]PiB PET SUVR thresholds for the global region and each of the individual regions that were previously determined.46 47 Blood draw (included in both the full and condensed protocol) A single venipuncture for blood collection is performed to examine certain proteins as well as standard clinical lab measures. Blood samples are collected by venipuncture or intravenously by a nurse or phlebotomist from the study team. Plasma collection, processing and analyses Blood samples are collected to polypropylene vacutainer (Greiner Bio-One, No. 456038) 6 mL tubes containing ~1.8 mg K3EDTA per mL of blood. Vacutainer tubes are inverted carefully 10 times to mix blood and anticoagulant, and samples undergo centrifugation using Eppendorf 5702R centrifuge (3200 rotations per minue (RPM); 1500 × g; A-8-17 swinging-bucket rotor) for 15 min at room temperature. Plasma is separated and transferred to 1.5 mL polypropylene cryovials (Olympus, No.: 21-363), divided into 0.5 mL aliquots. The samples are immediately transferred to the laboratory and stored at −80°C. Blood and cerebrospinal fluid (CSF) samples will be analysed for neurodegenerative markers in both groups, including amyloid beta 40 and amyloid beta 42 (Aβ42), glial fibrillar acidic protein, hyperphosphorylated tau 181 and total tau.48 Samples will also be stored for exploratory analysis of new analytes as evidence becomes available. APOE genotyping DNA is isolated from blood using the QIAmp Blood DNA Maxi Kit protocol (Qiagen, Valencia, California). Genotypes for two Apolipoprotein E (APOE) Single Nucleotide Polymorphism (SNPs), rs429358 (E*4) and rs7412 (E*2), will be determined using TaqMan SNP genotyping assays (Applied Biosystems, Foster City, California). Lumbar puncture (full protocol only—omitted for condensed protocol) The lumbar puncture is done following standard clinical procedures and performed by physicians in the department of neurological surgery. CSF samples (6 mL) are stored in the Neurotrauma Clinical Trials Center (NCTC) Biorepository until shipment to the Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory. Inflammatory markers in the CSF are measured using a Luminex assay (Millipore, Billerica, Massachusetts) which can analyse multiple inflammatory markers. CSF (via lumbar puncture) and serums (via venipuncture) samples are obtained from each subject. Thus, our [11C]PiB and [18F]AV-1451 PET imaging studies in TBI subjects are complemented by quantitative measures of Aβ and tau protein concentrations in CSF samples from the same imaged subjects in this protocol. CSF collection, processing and analyses Cerebrospinal fluid samples (6 mL), collected by lumbar puncture, are gently inverted to avoid possible gradient effects and aliquoted into polypropylene tubes prior to freezing at −80°C. CSF samples are not thawed until the start of the assay. Samples are analysed for total tau, phosphorylated tau181 and Aβ1-42 by enzyme-linked immunosorbent assay (Innotest; Innogenetics, Ghent, Belgium). Neuropsychological testing (included in both the full and condensed protocol) A comprehensive neuropsychological evaluation is administered using standardised assessments via paper and pencil and computer. The cognitive domains evaluated include verbal learning and memory, visual learning and memory, language, attention, working memory, visual-spatial, processing speed and executive function. The battery takes 2 hours to complete. See table 2 for a list of all assessments. Neuropsychological testing administrators were trained and certified by a licensed neuropsychologist to administer and score the neuropsychological battery. Certification requires satisfactory test administration and accurate completion of CRFs/record forms. Certification for independent administration is granted when staff display proficiency with test administration as determined by the neuropsychologist. Quality assurance checks are conducted regularly to monitor accuracy and consistency. Neuropsychological testing also includes the Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF), which aids in the assessment of mental disorders, personality characteristics, behavioural problems and major symptoms of psychopathology.49 This test is administered and scored on the computer and usually given during the wait time associated with neuroimaging studies. The MMPI-2-RF normative sample is drawn from the MMPI-2 normative sample. It consists of 2276 men and women between the ages of 18 and 80 from several regions and diverse communities in the USA. The MMPI-2-RF T scores are non-gendered and non-K-corrected. Table 2Neuropsychological batteryAttention, working memory and executive functionWisconsin Card Sorting TestHRNB Trail Making Tests A and BStroop Color/Word Interference Test(Golden)WAIS-IV Working Memory Index (Digit Span, Letter/Number Sequencing) Language Memory and learning Visual-spatial Motor and processing Speed North American Adult Reading TestCalifornia Verbal Learning TestRey Figure CopyWAIS-IV Processing Speed Index (Coding, Symbol Search)Controlled Oral Word Association TestWMS-IV Logical Memory Story B Immediate and DelayKaplan Clock DrawingAnimal NamingRey Figure Immediate and Delayed Recall Effort, functional status and symptom scales Victoria Symptom Validity ScaleModified GOSEMMPI-2-RFEmbedded validity measures: CVLT Forced Choice; WAIS-IV Reliable Digit Span; WAIS-IV Digit Span Scaled Score.CVLT, California Verbal Learning Test; GOSE, Glasgow Outcome Scale Extended; HRNB, Halstead-Reitan Neuropsychological Battery; MMPI, Minnesota Multiphasic Personality Inventory; WAIS-IV, Wechsler Adult Intelligence Scale Fourth Edition; WMS-IV, Wechsler Memory Scale, Fourth Edition. Neuropsychological battery CVLT, California Verbal Learning Test; GOSE, Glasgow Outcome Scale Extended; HRNB, Halstead-Reitan Neuropsychological Battery; MMPI, Minnesota Multiphasic Personality Inventory; WAIS-IV, Wechsler Adult Intelligence Scale Fourth Edition; WMS-IV, Wechsler Memory Scale, Fourth Edition. MRI of the brain (included in both the full and condensed protocol) Assessments of the nature, location and extent of any brain injury are conducted. Most scans occur using a 3T scanner in the dedicated MRI Research Center (MRRC). In the event a participant is unable to undergo an MRI at the MRRC, for example, due to size or claustrophobia, attempts are made to facilitate MRI in the University of Pittsburgh Medical Center (UPMC) Department of Radiology where a 1.5T wide-bore unit is available. Following the three-plane localiser, sagittal Magnetization Prepared Rapid Gradient Echo (MPRAGE), multishell diffusion imaging, sagittal 3D T2, axial SWI and sagittal 3D T2 Flair sequences are obtained for a total scan time of approximately 52 min. In the abbreviated protocol, three-plane localiser, sagittal MPRAGE, single-shell diffusion imaging, sagittal 3D T2 FLAIR, axial T2* and sagittal 3D T2 space sequences are obtained with a total scan time of approximately 30 min. A trained radiologist investigates the scans for incidental findings which may be clinically significant, such as a brain mass, aneurysm or Chiari malformation. Participants are notified in the event of clinically relevant incidental findings. Sleep assessment (included in both the full and condensed protocol) All participants undergo a comprehensive sleep medicine history and physical exam that includes sleep habits and timing, related nocturnal and daytime symptoms, review of outside sleep records, if applicable, and any sleep-related medical comorbidities, psychiatric conditions, social history and medications. Participants also complete validated questionnaires that assess multidimensional self-reported perceptions of sleep and the associated impact on daytime function and quality of life: Epworth Sleepiness Scale,50 Insomnia Severity Index,51 Pittsburgh Sleep Quality Index52 and three PROMIS scores (PROMIS-SD, PROMIS-SRI, PROMIS-Fatigue).53 All participants are issued a portable sleep apnoea testing device (WatchPAT, Zoll Itamar) to wear at their home or hotel to evaluate for sleep-related breathing disorders. The overnight assessment may be repeated if the dataset is technically limited or inadequate on the first night. Participants receive pretest instructions on how to use the device and are encouraged to sleep as they normally would (eg, usual position, pillow, medications and treatment devices), to obtain a recording representative of their usual night of sleep. The sleep medicine history, validated questionnaires and overnight testing are synthesised to determine if the participant meets diagnostic criteria for one or more sleep disorders, according to the American Academy of Sleep Medicine International Classification of Sleep Disorders, third edition.54 Applanation tonometry (continuous arterial pressure) (full protocol only—omitted for condensed protocol) Arterial stiffness is a known risk factor for dementia,55 so we measured arterial stiffness and pulse wave reflections using a high-fidelity applanation tonometer (SphygmoCor Xcel, Atcor Medical, Australia). Vascular function is characterised using measurements of central aortic pulse pressure, augmentation index (AI) and pulse wave velocity (PWV) as these are well-validated surrogates of arterial stiffness, pulse wave reflections and cardiovascular disease risk. After 5 min of rest in a supine position, applanation tonometry-derived indices are obtained. To assess AI, brachial artery compression waveforms are obtained by partially inflating a blood pressure cuff over the brachial artery approximately midway between the shoulder and the elbow. The brachial waveforms are obtained using cuff-measured brachial systolic and diastolic pressures, which are then used to generate central aortic pressure waveforms by applying proprietary digital signal processing and transfer function. The central aortic pulse wave is used to determine AI: the ratio of wave reflection amplitude to central pulse pressure. To assess PWV, carotid pulse waves are measured by applanation tonometry (high-fidelity micromanometer; Millar Instruments) and femoral pulse waves are simultaneously obtained by a partially inflated cuff over the femoral artery at the leg midway between the hip and the knee. Aortic PWV is determined by calculating the ratio of the corrected distance between the pulse measuring sites to the time delay between the carotid and femoral pulse waves. Informant interview (included in both the full and condensed protocol) Participants are asked to provide an informant that the study team may contact (by phone or in person). This conversation takes approximately 20–30 min and includes questions related to the participant’s cognition, drug and alcohol use. Participants may still participate in the study if they elect not to provide an informant. Ideally, the informant is someone who has known the participant for 10 years and can comment on memory and performance, typically a family member or close friend. The informant interview involves various questionnaires to assess if the participant has had any notable changes over time in the areas of cognition, behaviour and/or daily function. The informant interview is conducted by a study clinician or nurse practitioner. Clinical de-brief (included in both the full and condensed protocol): After the conclusion of the study assessments, the participant meets with the principal investigator either in person or virtually via Pitt Zoom (HIPAA-compliant version). The purpose of this discussion is to relay clinical findings and make recommendations (if applicable) for follow-up outside the scope of the study. This de-brief also allows the study team to ensure the participant is not experiencing any adverse events from the assessments, and to answer any questions/concerns prior to study completion. A written summary of findings is provided to each participant after study completion. Blinded and unblinded Diagnostic Consensus Conference (DCC) The diagnostic adjudication process is completed first with an unblinded group and later re-evaluated with a blinded group of clinicians. Both adjudications are structured the same way; study clinicians from each discipline summarise each participant’s data from all of the following assessments: history and physical exam, blood clinical labs, neuropsychological testing, sports medicine, concussion management and sleep medicine. After the clinical data are discussed, a neuroradiologist reports the visual MRI read and the qualitative assessment of PET images (if completed). Following the summary of findings, a consensus is reached with the DCC team across four domains of functioning pertaining to TES (cognitive, affective, behavioural, motor). Each domain is determined to be normal or abnormal based upon the review of data, ancillary study results and clinical impressions. The DCC team also determines whether the participant meets diagnostic criteria for mild cognitive impairment. After this information is recorded, the blinded adjudication committee is unblinded to group and trauma history. If the DCC consensus (in both the blinded and unblinded adjudications) determines that a participant is impaired in any of the four domains (cognitive, affective, behavioural, motor), the team reviews research criteria and makes a consensus decision on the presence of TES using three separate criteria described by Reams,25 Montenigro26 and the NINDS) TES consensus criteria.7 Outcomes of interest Cognitive impairment criteria (primary outcome) The indication of cognitive impairment is based on the ADRC National Alzheimer’s Coordinating Center (NACC) MCI criteria (Petersen56; Lopez57) and uses the chart in the NACC manual (Petersen58): (1) on review of all clinical and cognitive information, the DCC team judges that a participant does not have normal cognition for age; (2) the participant does not meet the criteria for dementia; (3) cognitive decline is determined to be present and (4) activities of daily living are normal. Cognitive impairment criteria will be defined as a dichotomous outcome (yes/no). Trauma-related neurodegeneration criteria (secondary outcome) The determination of trauma-related neurodegeneration is based on published criteria for TES:7 25 26 (1) presence of cognitive deficits, and/or neuromotor manifestations, and/or behavioural/mood-related symptoms; (2) PET scans demonstrating abnormal amyloid and/or tau deposition; (3) quantitative MR evidence of cortical volume loss and/or white matter injury and (4) elevated CSF tau/Aβ42 ratio. The analysis of trauma-related neurodegeneration takes into consideration history of TBI/concussion per the OSU-TBI survey.11 Both CSF and blood samples will be analysed for neurodegenerative markers. DNA is isolated from blood for APOE genotyping and other future genetic testing that may include whole-genome sequencing. Trauma-related neurodegeneration criteria are defined as a dichotomous outcome (yes/no). Statistical analysis plan Descriptive statistics will be presented for all study measurements overall and by group (former professional football players and unexposed controls). Distributional assumptions will be examined to identify potential influential observations and/or outliers. Categorical variables will be summarised by frequencies and percentages. Percentages will be calculated according to the number of patients where data are available. Where values are missing, the actual denominator will be stated. Continuous variables will be summarised using standard measures of central tendency and dispersion, using the mean and SD for normally distributed continuous data or median and IQR for non-normally distributed continuous data. Categorical characteristics will be compared with a χ2 test or Fisher’s exact test. Continuous characteristics will be compared with a two-sample t-test (or Mann-Whitney U test). All statistical analyses will be two-sided and the significance level will be 0.05. Significance level will be adjusted for multiple comparison correction by using a Bonferroni correction for multiple hypothesis testing. Specific aim 1. The point prevalence of cognitive impairment (primary outcome) and of neurodegeneration (secondary outcome) will be defined as the proportion of former professional football players with cognitive impairment based on the comprehensive assessment of clinical measures. That is, point prevalence will be defined as the number of former professional football players in the sample with cognitive impairment divided by the total number of former professional football players in the sample. The point prevalence of neurodegeneration (TES; secondary outcome measure) will be similarly defined as the proportion of former professional football players with a clinical syndrome of neurodegeneration based on the comprehensive assessment of clinical, neuroimaging and biomarker measures. We will also construct a 95% CI for the estimated proportions using the Wald approach.59 Specific aim 2. Analyses will begin by comparing baseline and clinical (eg, age, race, BMI and APOE status) characteristics of the exposed group (former professional football players) and unexposed group ((1) friend and brother controls and (2) community controls). Two-sample t-test (continuous data, eg, age) and χ2 test (categorical data, eg, race) will be used to determine differences in baseline and clinical characteristics between exposed and unexposed control groups. Mann-Whitney U test will be used if the assumptions for the parametric test are not met for continuous measures. Fisher’s exact test will be used for sparse data discrete measures. Prevalence of cognitive impairment (primary outcome) and of neurodegeneration (secondary outcome) for the unexposed control groups will be defined as in specific aim 1. The χ2 test will be used to compare the proportion of participants with cognitive impairment (or neurodegeneration) between the exposed (former professional football players) and unexposed control groups (defined by (1) friends and brothers and (2) friends and brothers and community controls) groups. Fisher’s exact test will be used for sparse data. It is possible that subject characteristics may not be balanced between the two groups. To control for these potential confounding effects, a propensity score will be used. The propensity of being in each group (exposed/unexposed) will be estimated and will be used in a propensity score-weighted logistic regression model (yes/no outcome), main effects of exposure (former professional football players/control) and propensity score as a weight to estimate the independent effect of exposure, controlling for the possible confounding effects. A combination of propensity score weighting and covariate adjustment (ie, doubly robust estimation) will be performed if the propensity scores were not able to completely balance all the characteristics. Specific aim 3. Two-sample t-test (or Mann-Whitney U test, continuous data, eg, hippocampal volume) will be used to determine differences in biomarkers between exposed and unexposed control groups. Point-biserial correlation (or rank-biserial correlation) will be used to measure the strength of the relationship between continuous (or ordinal) data imaging biomarker and clinical evaluation of cognitive and behavioural dysfunction (dichotomous data) in former professional football players. Sample size and power justification We calculated the total sample size needed a priori to appropriately power the primary outcome measure (ie, cognitive impairment). Based on a χ2 test, a total of 300 participants (150 former professional football players and 150 unexposed controls) achieve 80% power to detect a difference between the group proportions of 9.6%. This calculation assumes 5.0% of unexposed controls show cognitive impairment (based upon national prevalence estimates)14 and a two-sided significance level of 0.05. Prespecified interim analysis A preplanned interim analysis was conducted when 33% of exposed cases and controls had been evaluated (N=50 each). The interim analysis was conducted to re-estimate the sample. Based on the interim analysis and proportion of cognitive impairment observed in the unexposed control group, the sample size and power re-estimation showed that if we enrolled 150 former NFL players and 150 unexposed controls, we will be able to detect a difference in proportions of 9.6%, 13.4% or 15.0% (or effect size W=0.16) assuming 5.0%, 15.7% and 24.1% of unexposed controls show cognitive impairment with 80% power and significance level of 0.05. Ethics and dissemination The University of Pittsburgh Institutional Review Board approved the study prior to recruiting human subjects (protocol numbers STUDY19010008: sIRB - Brain Health Initiative (Part 1) and STUDY19030211: sIRB - Brain Health Initiative (Part 2)). The results will be disseminated in peer-reviewed journals and as presentations at national and international scientific conferences. All data are curated into a secure web-based database hosted by QuesGen, Inc. QuesGen is HIPAA compliant and stores identifiable/de-identified data on two separate secured servers. QuesGen, Inc. manages the servers which are located in a data centre compliant with Statement on Auditing Standards-70 behind a dedicated firewall. Supplementary material