Table 1.
TBI Phenotypes
| Study | Phenotype domain | Dim. reduction | Age | Population | Injury stage at assessment | Severity | Measures | Subgroups or clusters | Approach |
|---|---|---|---|---|---|---|---|---|---|
| 1. DeJong and Donders, 2010 | Cognitive | No | 16–79 | Civilian | Chronic | All severities | CVLT-II | Six subgroups: four replicated in mTBI; five replicated in severe TBI | Clustering (fastclus); Agglomerative clustering on variance; k-means labeling |
| 2. Mottram and Donders, 2006 | Cognitive | No | 6–16 | Civilian | Chronic | All | CVLT-C | Four clusters based on level and pattern of performance | Clustering (fastclus); Complete linkage procedure (reliability check) |
| 3. Sherer et al., 2017 | Cognitive | No | 16–70 | Civilian | Chronic | All | TBI-QOL, NSI, EQOL, FADGF, PART-O | Five clusters that had significantly different results on PART-O scale | Clustering; agglomerative clustering on variance; k-means labeling |
| 4. Pugh et al., 2019 | Comorbid factors | No | 18–65 | Military/Veterans | Chronic | Mild | GCS, diagnosed health conditions | Five comorbidity trajectories: moderately healthy stable; moderately healthy decline; mental health, polytrauma stable; polytrauma with improvement | LCA, logistic regression |
| 5. Lumba-Brown et al., 2020 | Post-concussive symptom focused | No | 6+ | All populations | Acute | Mild | Concussion Symptom Scales and other indicators | Five concussion subtypes: cognitive, ocular-motor, headache/migraine, vestibular, anxiety/mood (sleep disturbance also associated) | Literature review and meta-analysis |
| 6. Maruta et al., 2018 | Symptom | No | 12–30 | Civilian athletes | Acute | Mild | RPQ | Six classes of symptoms: cognitive/fatigue, vestibular, oculomotor, anxiety/mood, migraine, cervical/sleep | Binomial tests; expert assignment; subtype prevalence overlap analysis |
| 7. Velikonja et al., 2010 | Emotional and behavioral | No | 15–69 | Canadian clinical cohort | Chronic | All | PAI | Seven clusters: multiple symptoms, somatic/depressive symptoms, normal, depression, substance use/antisocial, normal (minimizing), multiple symptoms/bipolar | Split sample; three clustering methods; hierarchical, agglomerative, k-means, linkage, iteration |
| 8. Warriner et al., 2003 | Emotional and behavioral | No | 15–69 | Canadian clinical cohort | Chronic | Mild-moderate (75% of patients) | MMPI | Six injury outcome subtypes: normal function, mild somatic/pain concerns, disinhibition/externalizing behavior, internalizing behavior, externalizing and somatic behavior | Split sample; three clustering methods; hierarchical, agglomerative, k-means, linkage, iteration |
| 9. Juengst et al., 2017b | Emotional, cognitive, and behavioral | No | 16–70 | Civilian | Chronic | All | PHQ-9; PANAS; NTB; FSBS | Temporal evolution of emotional, cognitive and behavioral clusters; <6-month injury: clustered along continuum of emotion/behavior symptoms; >6-month injury: complex symptom patterns | Cross-lagged panel analysis, structural equation modeling |
| 10. Nielson et al., 2017 | Outcome, biomarkers | Yes | 43.3 ± 18.5 | Civilian | Chronic | All | Injury character, neuroimaging, PTSD Checklist; WAIS; CVLT | Two broad topological node groups, six nodal extrema. One mTBI node reflecting unfavorable outcomes on GOSE 3–6 months, (including PTSD) associated with PARP1, ANKK1, COMT, and DRD2 | Topological data analysis with third-party software, linear models |
| 11. Goldsworthy and Donders, 2019 | Personality | No | 18–75 | Civilian | Chronic | All | MMPI-2-RF | Four clusters: clusters 1 and 4 differed by profile elevations; clusters 2 and 3 varied in pattern. Pre-morbid factors separated clusters. | Clustering (fastclus) |
| 12. Kennedy et al., 2015 | Personality | No | 19–49 | Military | Chronic | Mild | PAI | Four clusters: high distress, moderate distress, somatic distress, no distress | Clustering, hierarchical, and k-means |
| 13. Hellstrom et al., 2013 | Symptom | No | 16–55 | Civilian | Chronic | Mild | RPQ | Four clusters: low symptoms, high symptoms, cognitive, somatic | Clustering, hierarchical, and k-means |
| 14. Polimanti et al., 2017 | Symptoms | No | 18–46 | Military post-9/11 | Chronic | Unknown | GWAS; PCS | No significant association of post-concussive symptoms (PCS) with any genetic components; high infant HC-PRS was correlated with better recovery from concussion. | Genome-wide cross-phenotype analysis with PRSice, linkage disequilibrium, enrichment, regression |
| 15. Stein et al., 2016 | Symptom | No | 18–46 | Military post-9/11 | Chronic | Mild | PCS | Severity of PCS associated with five traits: history of TBI, stress, more severe deployment-related events, LOC lapse of memory vs. LO attention | Zero-inflated negative binomial regression |
| 16. Ensign et al., 2012 | Psychosocial | No | 6–20 | Civilian | Chronic | Mild-severe | BASC-2 | Six: two primary: Normal, Pervasive emotional difficulties; four less reliable: Mild Externalizing with 1) Depression, 2) Attention Problems, 3) Mild Depression, and 4) Mild Anxiety | Agglomerative hierarchical cluster analysis and simple UPGMA, Ward's methods |
| 17. Hayman-Abello et al., 2003 | Psychosocial | No | 12–18 | Civilian | Chronic | Mild-severe | CBCL | Four groups: Normal, Attention, Delinquent, and Withdrawn-Somatic | Q-factor analysis |
| 18. Folweiler et al., 2020 | Severity | Yes | 18–70 | Civilian | Acute | Mild-severe | GCS | Three patient phenotypes, two replicated across studies | GLRM; Gower's dissimilarity matrix K-nearest neighbor |
| 19. Gravesteijn et al., 2020 | Severity | Yes | 50a [30, 66] | Civilian | Acute | Mild-severe | Injury Mechanism Extracranial Injury GCS | Four clusters of severity associated with differential long-term outcomes | Bootstrap resampling with replacement; PCA and Gower's distance |
| 20. Masino et al., 2018 | Severity | Yes | 18–70 | Civilian | Chronic | Mild-severe | Baseline <24 h, CT scan, other intake history | Four distinct patient phenotypes that were associated with 90-day outcomes on 12 assessments | GLRM feature selection, Gower's distance, dissimilarity matrix |
| 21. Si et al., 2018 | Severity | No | 16+ | Civilian | Acute | Mild | GCS, clinical variables, GOSE, WAIS | Five mTBI subgroups: general, cognitive, functional, emotional, and somatic | Sparse hierarchical clustering with automated feature rejection/selection |
| 22. Kucukboyaci et al., 2018 | Comorbid and vulnerability | No | 16–70 | Civilian | Unspecified | Not specified | Demographics, psychosocial | Four clusters: 1) high substance use and psychiatric history; 2) race/ethnic minority, limited English proficiency; 3) minority with substance use, incarceration, and homelessness; and 4) elderly with complex comorbidity | Two-step clustering with log-linear differences and k-means |
| 23. Yeates et al. 2019 | General post-concussion | No | 8–18 (8–12, 13–18) | Civilian | Acute - chronic | Mild | Pre-morbid, clinical. ACE, PCSI, SAC, and BESS | Four clusters found using pre-morbid history. Four clusters found using clinical data. Assessment at 4 and 12 weeks. Age, female sex, (anxiety), phenotypes increase PPCS risk. | LCA |
| 24. Bailie et al. 2016 | Psychosocial, cognitive, behavioral | No | 18–56 | Military | Chronic | Mild | Neurobehavioral Symptom Inventory and PTSD Checklist-Civilian Version (PCL-C) | Four subtypes: primarily psychiatric (post-traumatic stress disorder) group, a cognitive group, a mixed symptom group, and a good recovery group | Two-step clustering procedure (hierarchical clustering and k-means labeling) using average linkage/Pearson correlation as the proximity measure |
| 25. Zimmermann et al. 2015 | Cognitive, executive function | No | 18+ | Civilian | Chronic | All severities | Multiple Executive Function tasks | Three clusters: 1) inhibition, flexibility, and focused attention; 2) inhibition, flexibility, working memory, and focused attention; and 3) no expressive executive deficits | Hierarchical cluster analysis, tasks Z-scores, ANOVA |
| 26. Howell et al. 2019 | Post-concussive symptom focused | No | 7–30 | Civilian | Acute-chronic | Mild | Post-Concussion Symptom Scale | Five symptom domains: 1) somatic, 2) emotional, 3) sleep, 4) cognitive, and 5) vestibular-ocular | Linear regression model |
| 27. Kontos et al. 2019 | Post-concussive symptom focused | No | 11–40 | Civilian | Unspecified | Mild | Medical history, injury, clinical interview/examination notes, cognitive/vestibular/ocular tests | Six profiles: 1) cognitive/fatigue, 2) vestibular, 3) ocular, 4) post-traumatic migraine, 5) anxiety/mood, and 6) cervical | Blinded chart reviews by six clinicians determining the primary and secondary clinical profiles |
| 28. Feddermann-Demont et al. 2017 | Post-concussive symptom focused | No | Teens-adults (mean age 17) | Athletes | Unspecified | Mild | Symptom scales, neurocognitive tests, balance | Five domains: cognition, dizziness and balance, emotions, headache, and vision | Clinical comparison of post-concussive symptoms; subanalysis of predominant symptoms |
a
Interquartile range.
AUDIT, Alcohol Use Disorders Identification Test; BDI-II, Beck Depression Inventory II; BASC-2, Behavior Assessment System for Children, Second Edition; UPGMA, between-group linkage; CVLT-C, California Verbal Learning Test–Children's Version; CVLT-II, California Verbal Learning Test–Second Edition; CBCL, Child Behavior Checklist; EQOL, Economic Quality of Life Scale; ESL, English as Second Language; FADGF, Family Assessment Device General Functioning Scale; FSBS, Frontal Systems Behavior Scale; GLM, general linear model; GLRM, generalized low-rank models; GWAS, genome-wide association study; GCS, Glasgow Coma Scale; GOSE, Glasgow Outcome Score-Extended; IHC, infant head circumference; LCA, latent class analysis; LOOCV, leave one out sample cross-validation; MMPI-2-RF, Minnesota Multiphasic Personality Inventory–2–Restructured Form; NSI, Neurobehavioral Symptom Inventory;,NTB, Neuropsychological Test Battery; PART-O, Participation Assessment with Recombined Tools-Objective; PHQ-9, Patient Health Questionnaire-9; PAI, Personality Assessment Inventory; PRS, Polygenic Risk Score; PANAS, Positive and Negative Affect Schedule; PCS, post-concussive symptoms; PTSD, post-traumatic stress disorder; PCA, principal component analysis; ROI, region of interest; RAVLT, Rey Auditory Verbal Learning Test; RPQ, Rivermead Post-Concussion Symptoms Questionnaire; SNP, single-nucleotide polymorphism; SHC, sparse hierarchical clustering; TBI-QOL, TBI Quality of Life; TDA, topological data analysis; UPGMC, unweighted pair-group method using centroid averages; WAIS, Wechsler Adult Intelligence Scale; ZNB, zero-inflated negative binomial.