CLINICAL ARTICLE J Neurosurg 131:596–603, 2019

Long-term outcome in patients with midline shift: a secondary analysis of the Phase 3 COBRIT clinical trial

Ross C. Puffer, MD,1 John K. Yue, BS,2 Matthew Mesley, MD,2 Julia B. Billigen, RN,2 Jane Sharpless, MS,2 Anita L. Fetzick, RN, MSN,2 Ava Puccio, PhD,2 Ramon Diaz-Arrastia, MD, PhD,3 and David O. Okonkwo, MD, PhD2

1Department of Neurosurgery, Mayo Clinic, Rochester, Minnesota; 2Department of Neurosurgery, UPMC, Pittsburgh; and 3Department of Neurosurgery, University of Pennsylvania, Philadelphia, Pennsylvania

OBJECTIVE Following traumatic brain injury (TBI), midline shift of the brain at the level of the septum pellucidum is often caused by unilateral space-occupying lesions and is associated with increased and worsened morbidity and mortality. While outcome has been studied in this population, the recovery trajectory has not been re- ported in a large cohort of patients with TBI. The authors sought to utilize the Citicoline Brain Injury Treatment (COBRIT) trial to analyze patient recovery over time depending on degree of midline shift at presentation. METHODS Patient data from the COBRIT trial were stratified into 4 groups of midline shift, and outcome measures were analyzed at 30, 90, and 180 days postinjury. A recovery trajectory analysis was performed identifying patients with outcome measures at all 3 time points to analyze the degree of recovery based on midline shift at presentation. RESULTS There were 892, 1169, and 895 patients with adequate outcome data at 30, 90, and 180 days, respectively. Rates of favorable outcome (Glasgow Outcome Scale–Extended [GOS-E] scores 4–8) at 6 months postinjury were 87% for patients with no midline shift, 79% for patients with 1–5 mm of shift, 64% for patients with 6–10 mm of shift, and 47% for patients with > 10 mm of shift. The mean improvement from unfavorable outcome (GOS-E scores 2 and 3) to favor- able outcome (GOS-E scores 4–8) from 1 month to 6 months in all groups was 20% (range 4%–29%). The mean GOS-E score for patients in the 6- to 10-mm group crossed from unfavorable outcome (GOS-E scores 2 and 3) into favorable outcome (GOS-E scores 4–8) at 90 days, and the mean GOS-E of patients in the > 10-mm group nearly reached the threshold of favorable outcome by 180 days postinjury. CONCLUSIONS In this secondary analysis of the Phase 3 COBRIT trial, TBI patients with less than 10 mm of midline shift on admission head CT had significantly improved functional outcomes through 180 days after injury compared with those with greater than 10 mm of midline shift. Of note, nearly 50% of patients with > 10 mm of midline shift achieved a favorable outcome (GOS-E score 4–8) by 6 months postinjury. https://thejns.org/doi/abs/10.3171/2018.2.JNS173138 KEYWORDS TBI; midline shift; outcome; clinical trial; trauma; traumatic brain injury

ollowing traumatic brain injury (TBI), midline shift line shift can be the result of heterogeneous morphologies of the brain at the level of the septum pellucidum consisting of hemorrhage, infarction, cerebral edema, and is often caused by unilateral, space-occupying le- hydrocephalus, or a combination of these factors. Current Fsions and is associated with increased intracranial pressure TBI guidelines suggest that mass lesions causing > 5 mm and worsened morbidity and mortality.1,2,4,6,13,16,20,21,27,33 The of midline shift associated with abnormal neurological ex- degree of shift is associated with abnormal neurological amination results that are potentially reversible should be examination results due to progressive dysfunction of the considered for surgical evacuation.3,11,12,17,18,22,23,29,31,32,36 diencephalon subsequent to mass effect.5,9,27 In TBI, mid- Patients with TBI who present with substantial midline

ABBREVIATIONS COBRIT = Citicoline Brain Injury Treatment; CONSORT = Consolidated Standards of Reporting Trials; FITBIR = Federal Interagency Traumatic Brain Injury Research; GCS = Glasgow Coma Scale; GOS-E = Glasgow Outcome Scale–Extended; TBI = traumatic brain injury. SUBMITTED December 14, 2017. ACCEPTED February 16, 2018. INCLUDE WHEN CITING Published online August 3, 2018; DOI: 10.3171/2018.2.JNS173138.

596 J Neurosurg Volume 131 • August 2019 ©AANS 2019, except where prohibited by US copyright law

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

FIG. 1. CONSORT diagram. shift and abnormal neurological examination results may (FITBIR) informatics system. All patient information have a poorer prognosis with significant implications for within the FITBIR system for this trial is de-identified, family discussions and surgical decision-making. Many of and therefore institutional review board approval was not these decisions are made based on incomplete information required. Our institutional business officer (per FITBIR and speculation, and withdrawal of care is a leading cause requirements) evaluated the method of data acquisition of in-hospital death in patients with severe TBI.19,25,28 How- and usage to ensure that data access protocols and data ever, there is a paucity of long-term outcome data regard- management were performed in accordance with regula- ing improved rates of recovery after TBI.20,27 tions. The Citicoline Brain Injury Treatment (COBRIT) tri- The full details of the COBRIT trial, a multicenter, al34,35 was a Phase 3, double-blind, randomized controlled double-blind randomized controlled trial performed at 8 trial within the clinical trials network that was designed Level I trauma centers within the US, have previously been to evaluate the efficacy of citicoline to improve functional published.34,35 Inclusion criteria were age 18–70 years, non- and cognitive outcomes after TBI; the trial included pa- penetrating TBI classified as complicated mild (CT posi- tients with complicated mild (positive acute intracranial tive, GCS scores 13–15), moderate (GCS scores 9–12), and pathology on the initial head CT scan but Glasgow Coma severe (GCS scores 3–8). Additional CT inclusion criteria Scale [GCS] scores 13–15), moderate (GCS scores 9–12), were the presence of intraparenchymal hemorrhage > 10 and severe (GCS scores 3–8) TBI. The trial was stopped mm in any diameter, midline shift > 5 mm, acute extra- for futility, as it failed to demonstrate improved functional axial hematoma > 5 mm, intraventricular hemorrhage on or cognitive outcomes with citicoline treatment; however, at least 2 contiguous 5-mm slices, and subarachnoid hem- the large number of enrolled patients and detailed func- orrhage on at least 2 contiguous 5-mm slices. Patients were tional and cognitive outcome measures studied at 30, 90, not required to have > 5 mm of midline shift to be included and 180 days made this enriched data set a valuable source in the COBRIT trial, but this finding on CT scanning was considered positive if it was the only finding present at for subset analyses. We utilized the prospective COBRIT 34,35 trial data to analyze the degree of midline shift on ad- baseline. Patients were excluded if they had bilateral mission CT scanning and its association with functional fixed and dilated pupils, imminent death, or history of outcomes at 30, 90, and 180 days after TBI. serious psychiatric or neurological illness that would af- fect neuropsychological outcomes.34,35 The Consolidated Standards of Reporting Trials (CONSORT) diagram for Methods COBRIT is included in Fig. 1. Access to the COBRIT trial database was granted by Demographic, radiological, and outcome measures for the Federal Interagency Traumatic Brain Injury Research each TBI patient were downloaded from the FITBIR sys-

J Neurosurg Volume 131 • August 2019 597

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

TABLE 1. Selected demographics mean age of this cohort was 40.4 ± 0.5 years, and 651 No. of Patients (73%) patients were male. The GOS-E was calculated at Variable (%) a mean of 31.5 ± 0.2 days from injury. The cohort was separated into 4 categories of midline shift: no midline Total 1213 shift (n = 697), 1–5 mm of midline shift (n = 133, mean Age, yrs 2.8 ± 0.1 mm), 6–10 mm of midline shift (n = 49, mean 7.5 18–30 434 (36) ± 0.04 mm), and > 10 mm of midline shift (n = 18, mean 31–45 266 (22) 15.1 ± 0.15 mm). 46–60 344 (28) GOS-E scores at 30 days are presented in Fig. 2 and >60 171 (14) Table 2. The differences among midline shift groups were statistically significant by chi-square analysis (p < 0.0001). Sex In patients with no midline shift on the admission CT scan Female 310 (26) (n = 692), 31 (4.5%) patients died, 215 (31% had an unfa- Male 903 (74) vorable outcome, and 446 (64.5%) patients had a favorable Mechanism of injury outcome. In patients with 1–5 mm of midline shift at base- Motor vehicle collision 594 (49) line CT (n = 133), 8 (6%) died, 54 (41%) had an unfavor- Struck by vehicle 72 (6) able outcome, and 71 (53%) had a favorable outcome. In patients with 6–10 mm of midline shift (n = 49), 4 (8%) Fall 378 (31) died, 28 (57%) had an unfavorable outcome, and only 17 Sports 7 (0.5) (35%) had a favorable outcome. In patients with > 10 mm Assault 113 (9) of midline shift (n = 18), the mortality rose to 28% (n = 5), Other mechanism 49 (4.5) the unfavorable outcome rate rose to 44% (n = 8), and the Admission GCS score favorable outcome rate at 30 days was 28% (n = 5). 13–15 w/ imaging abnormality (complicated mild) 659 (54) 9–12 (moderate) 165 (14) 90-Day Outcomes 3–8 (severe) 389 (32) There were 1169 patients with adequate follow-up in- formation to be included in the assessment of outcomes by the GOS-E (unfavorable: scores 2 and 3; favorable: scores 4–8) at 90 days postinjury. The greatest number of patients tem and consolidated into separate databases for 30-, 90-, had 90-day outcomes, as this was the primary endpoint of and 180-day analyses and a joint database for recovery the COBRIT trial, with 30- and 180-day outcomes being trajectory analysis. The data were analyzed using JMP secondary endpoints. The mean age of the 90-day cohort statistical software (SAS Institute). Selected demograph- was 40 ± 0.5 years, and 871 (74%) patients were male. The ic information from the COBRIT trial is shown in Table GOS-E score was calculated a mean of 83.9 ± 0.7 days 1. Statistical variables were categorical, and the degree from injury. The cohort was separated into 4 categories of midline shift was separated into 4 categories (none, of midline shift: no midline shift (n = 898), 1–5 mm of 1–5 mm, 6–10 mm, and > 10 mm). These variables were midline shift (n = 178, mean 2.8 ± 0.1 mm), 6–10 mm of analyzed against neurological outcomes as measured by 14,24,30 midline shift (n = 68, mean 7.3 ± 0.1 mm), and > 10 mm of the Glasgow Outcome Scale–Extended (GOS-E). midline shift (n = 25, mean 14.6 ± 0.2 mm). GOS-E scores were dichotomized as unfavorable (GOS- GOS-E outcomes at 90 days are presented in Fig. 2 E score 2 and 3) and favorable (GOS-E score 4–8) out- and Table 3. The differences among midline shift groups comes, consistent with the determination by the investi- were statistically significant by chi-square analysis (p < gators and steering committee of the recent RESCUEicp 0.0001). In patients with no midline shift on the admis- (Randomized Evaluation of Surgery with Craniectomy for sion CT scan (n = 898), 37 (4.1%) died, 132 (14.7%) had an Uncontrollable Elevation of Intracranial Pressure) trial, as unfavorable outcome, and 729 (81.2%) patients had a favor- well as other hemicraniectomy trials for the treatment of 7,10,26 able outcome. In patients with 1–5 mm of midline shift on malignant ischemic . Mean values are presented the baseline CT scan (n = 178), 11 (6.2%) died, 38 (21.3%) as the mean ± SEM. Pearson’s chi-square test was utilized had an unfavorable outcome, and 129 (72.5%) patients had for the primary outcome analysis, and a repeated-mea- a favorable outcome. In patients with 6–10 mm of shift sures multivariate ANOVA test including within-subjects (n = 68), mortality and unfavorable outcome increased; 6 and between-subjects analyses was performed for the re- (8.8%) died and 22 (32.2%) had an unfavorable outcome, covery trajectory analysis. A p value of < 0.05 was consid- and 40 (59%) of patients had a favorable outcome. In pa- ered significant. tients with > 10 mm of midline shift (n = 25), mortality increased, and 6 (24%) died. Seven (28%) patients had an Results unfavorable outcome, and only 12 (48%) patients had a fa- 30-Day Outcomes vorable outcome. For the analysis of midline shift, 892 patients had ad- equate follow-up information to be included in the assess- 180-Day Outcomes ment of outcomes by the GOS-E (unfavorable: scores 2 There were 895 patients with adequate follow-up infor- and 3; favorable: scores 4–8) at 30 days postinjury. The mation to be included in the assessment of outcomes by

598 J Neurosurg Volume 131 • August 2019

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

FIG. 2. The 30-day (A), 90-day (B), 180-day (C) postinjury GOS-E outcomes (D) by degree of midline shift. the GOS-E (unfavorable: scores 2 and 3; favorable: scores (17%) died, 10 (19%) had an unfavorable outcome, and 34 4–8) at 180 days postinjury. The mean age of this cohort (64%) patients had a favorable outcome. In patients with was 40.4 ± 0.5 years, and 655 (73%) patients were male. > 10 mm of midline shift (n = 17), mortality and unfa- The GOS-E was calculated a mean of 182.4 ± 0.4 days vorable outcome were substantial, with 6 (35%) patients from injury. The cohort was separated into 4 categories having died and 3 (18%) having an unfavorable outcome. of midline shift: no midline shift (n = 690), 1–5 mm of Despite the increase, just under half (47%, n = 8) of the midline shift (n = 135, mean 2.8 ± 0.1 mm), 6–10 mm of patients had a favorable outcome at 180 days after injury. midline shift (n = 53, mean 7.5 ± 0.2 mm) and > 10 mm of midline shift (n = 17, mean 15 ± 0.3 mm). Recovery Trajectory Analysis GOS-E outcomes at 180 days are presented in Fig. 2 and Table 4. The differences among midline shift groups Outcomes were available for 713 patients at all 3 time were statistically significant by chi-square analysis (p < points, and a recovery trajectory analysis is demonstrated 0.0001). In patients with no midline shift on the admission in Fig. 3. There was a significant difference in the between- CT scan (n = 690), 42 (6%) died, 51 (7%) had an unfa- subjects analysis [F(3, 713) = 13.9, p < 0.001] and within- vorable outcome, and 597 (87%) patients had a favorable subjects analysis [F(3, 713) = 62.6, p < 0.001]. The mean ± outcome. In patients with 1–5 mm of midline shift on the SEM GOS-E score for patients with no midline shift was baseline CT scan (n = 135), 11 (8%) died, 17 (13%) had an 4.7 ± 0.07 at 30 days, 5.6 ± 0.08 at 90 days, and 5.9 ± 0.08 unfavorable outcome, and 107 (79%) patients had a favor- at 180 days. The mean GOS-E score for patients with 1–5 able outcome. In patients with 6–10 mm of midline shift mm of midline shift was 4.2 ± 0.16 at 30 days, 5.2 ± 0.19 (n = 53), mortality and unfavorable outcome increased; 9 at 90 days, and 5.6 ± 0.21 at 180 days. The mean GOS-E

TABLE 2. 30-day outcomes by brain midline shift on the admission CT scan (n = 892) GOS-E Score* Shift 1 2 3 4 5 6 7 8 None (n = 692) 31 (4.5%) 20 (2.9%) 195 (28.2%) 66 (9.5%) 129 (18.6%) 127 (18.4%) 75 (10.8%) 49 (7.1%) 1–5 mm (mean 2.8 mm; n = 133) 8 (6%) 9 (6.7%) 45 (33.8%) 5 (3.8%) 27 (20.3%) 22 (16.5%) 11 (8.3%) 6 (4.6%) 6–10 mm (mean 7.5 mm; n = 49) 4 (8%) 11 (22.5%) 17 (34.7%) 2 (4.1%) 9 (18.4%) 5 (10.2%) 0 (0%) 1 (2.1%) >10 mm (mean 15.1 mm; n = 18) 5 (27.8%) 4 (22.2%) 4 (22.2%) 1 (5.6%) 3 (16.6%) 1 (5.6%) 0 (0%) 0 (0%) * 1, dead; 2, vegetative state; 3, lower severe disability; 4, upper severe disability; 5, lower moderate disability; 6, upper moderate disability; 7, lower good recovery; 8, upper good recovery.

J Neurosurg Volume 131 • August 2019 599

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

TABLE 3. 90-day outcomes by brain midline shift on the admission CT scan (n = 1169) GOS-E Score Shift 1 2 3 4 5 6 7 8 None (n = 898) 37 (4.1%) 14 (1.6%) 118 (13.1%) 58 (6.5%) 168 (18.7%) 183 (20.4%) 180 (20%) 140 (15.6%) 1–5 mm (mean 2.8 mm; n = 178) 11 (6.2%) 6 (3.4%) 32 (17.9%) 10 (5.6%) 36 (20.2%) 33 (18.5%) 24 (13.5%) 26 (14.7%) 6–10 mm (mean 7.3 mm; n = 68) 6 (8.8%) 6 (8.8%) 16 (23.5%) 7 (10.3%) 14 (20.6%) 9 (13.2%) 5 (7.4%) 5 (7.4%) >10 mm (mean 14.6 mm; n = 25) 6 (24%) 3 (12%) 4 (16%) 1 (4%) 7 (28%) 2 (8%) 2 (8%) 0 (0%) score for patients with 6–10 mm of shift was 3.4 ± 0.23 at provement from 1 month to 6 months, which would sug- 30 days, 4.1 ± 0.29 at 90 days, and 4.8 ± 0.38 at 180 days. gest that some patients improve over time; however, the The mean GOS-E score for patients with > 10 mm of shift majority of patients had a favorable outcome already at was 2.3 ± 0.38 at 30 days, 3 ± 0.52 at 90 days, and 3.8 ± 1 month postinjury. Similarly, patients with 1–5 mm of 75 at 180 days. Recovery trajectories were similar in pa- midline shift at presentation (mean 2.8 mm) had favorable tients with 5 mm of shift or less, with mean GOS-E scores outcome rates of 53% at 1 month, 72.5% at 3 months, and improving approximately 1.3 points between 30 and 180 79% at 6 months. The improvement in favorable outcomes days. Patients with > 5 mm of shift had slightly better re- from 1 month to 6 months (26%) is encouraging and sug- covery trajectories, improving 1.5 GOS-E points over the gests that a significant number of patients with this level same interval; however, their starting GOS-E score was of midline shift at presentation and an impaired functional lower, and they also had a lower mean GOS-E score at 180 status at 1 month after injury will improve over time. days. The mean GOS-E score in patients with 6–10 mm Patients with 6–10 mm of midline shift at presentation of midline shift crossed the favorable outcome line by 90 (mean 7.5 mm) had favorable outcome rates of 35% at 1 days postinjury, while the mean GOS-E score in patients month, 59% at 3 months, and 64% at 6 months postin- with > 10 mm of midline shift did not cross the favorable jury. In patients with > 10 mm of midline shift (mean 15.1 outcome line at 180 days. The 95% confidence interval of mm for all time points), favorable outcome rates were 28% the mean GOS-E was noted to cross into favorable out- at 1 month, 48% at 3 months, and 47% at 6 months. In- come at the 180-day time point in patients with > 10 mm terestingly, the improvement in favorable outcome rates of midline shift. at 6 months for patients with 6–10 mm and > 10 mm of midline shift is again substantial, with one-fifth to nearly one-third of the cohort (20%–29%) improving from unfa- Discussion vorable to favorable outcome between those time points. In this secondary analysis of the completed Phase 3 There is, however, a noticeable difference in favorable out- COBRIT clinical trial for TBI, we report the 30-, 90-, and come rates at 6 months when midline shift increases over 180 day outcomes on the GOS-E stratified by the degree 5 mm (< 5 mm: 79%–87% vs > 5 mm: 47%–64%). of midline shift present on the admission CT scan. Mor- tality increased significantly based on degree of midline Recovery Trajectory shift at admission at all subsequent time points. Trajectory Recovery trajectories between 30 and 180 days were analysis reveals that it may require more than 180 days to similar for all groups studied; however, there was a very determine true functional outcomes after TBI. slightly better recovery trajectory in patients with > 5 mm of midline shift at presentation. The mean GOS-E score Midline Shift and Functional Outcome improvement in this group was approximately 1.5 points Patients who presented with no midline shift had high from 30 to 180 days, compared with a mean of 1.3 points of rates of favorable outcome (GOS-E scores 4–8) at 1, 3, and improvement in patients with 5 mm or less midline shift. 6 months of 64.5%, 81%, and 87%, respectively. While a This is likely a statistical anomaly, given that patients with rate of 87% at 6 months might be lower than expected, it > 5 mm of midline shift started with lower mean GOS-E reinforces that outcome after TBI is affected by many fac- scores at 30 days and may not have had as much skew tors, and even in patients with no midline shift, favorable as the group with 5 mm of shift or less. Regardless, both long-term outcome is not assured. There is observed im- groups experienced significant improvement between 30

TABLE 4. 180-day outcomes by brain midline shift on the admission CT scan (n = 895) GOS-E Score Shift 1 2 3 4 5 6 7 8 None (n = 690) 42 (6%) 3 (0.4%) 48 (6.9%) 26 (3.8%) 107 (15.5%) 133 (19.3%) 161 (23.3%) 170 (24.8%) 1–5 mm (mean 2.8 mm; n = 135) 11 (8.2%) 2 (1.5%) 15 (11.1%) 5 (3.7%) 23 (17%) 26 (19.3%) 23 (17%) 30 (22.2%) 6–10 mm (mean 7.3 mm; n = 53) 9 (16.9%) 5 (9.4%) 5 (9.4%) 5 (9.4%) 7 (13.2%) 7 (13.2%) 8 (15.3%) 7 (13.2%) >10 mm (mean 15 mm; n = 17) 6 (35.3%) 0 (0%) 3 (17.7%) 0 (0%) 2 (11.6%) 3 (17.7%) 0 (0%) 3 (17.7%)

600 J Neurosurg Volume 131 • August 2019

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

FIG. 3. Recovery trajectory of the mean GOS-E score by degree of midline shift. and 180 days, with the mean GOS-E score of patients in previous randomized, controlled trial designed to evaluate the 6- to 10-mm group crossing into favorable outcome at the efficacy of a medication for treatment of TBI outcomes. 90 days and the mean GOS-E of patients in the > 10-mm We utilized the negative results of the trial to combine the group nearly reaching that threshold at 180 days (95% con- entire cohort, but it should be recognized that a proportion fidence interval does reach favorable outcome). This high- of the study population received a drug (citicoline) that was lights the fact that trials attempting to study TBI should hypothesized to have an effect on outcome. Nonetheless, continue follow-up to at least 1 year after injury and likely the fact that the trial did not show any effect mitigates the beyond. risk of bias. Another limitation is that only midline shift While midline shift is known to be associated with out- is utilized, and patients are not further stratified by type come based on studies including data from the Traumatic of intracranial pathology, which almost certainly has an Coma Data Bank,6,8,15 our study helps to stratify patients effect on outcome. The large number of patients at each by degree of midline shift and demonstrate their outcomes. outcomes time point (892, 1169, and 895) helps to decrease It has been demonstrated that midline shift as a continu- the risk of bias introduced by differences in intracranial ous variable is correlated with outcome, but this variable pathology, coexistent systemic injuries, and other potential can be difficult to utilize when considering prognosis for confounding variables. Furthermore, this was a longitudi- individual patients. Our study further reports the recovery nal study, and the cohorts at 30, 90, and 180 days contained trajectory analysis in these patients, which may help pro- a large proportion of the same patients. The primary end- viders counsel patients and family members on potential point of the original trial was at 90 days postinjury, and, in for recovery for an individual patient falling into one of our analysis, there were approximately 300 fewer patients the reported categories of midline shift at presentation. We with GOS-E evaluation at the secondary endpoints of 30 have demonstrated that not only do patients with 5 mm of and 180 days, which could potentially skew the results and midline shift or less have excellent recovery trajectories, subsequent trajectory analysis. Our trajectory analysis in- but also that patients in the 6–10 mm of midline shift cat- cluded only patients with data at all 3 time points, but this egory may also have good recovery potential, going from excluded nearly 500 patients, and that substantial number nearly two-thirds unfavorable outcome at 1 month to nearly could greatly affect our results. Lastly, the study is limited two-thirds having a favorable outcome by 6 months. Even by only demonstrating outcomes at 6 months postinjury. patients with > 10 mm of midline shift demonstrated a It would be more informative to demonstrate the 1-year significant recovery trajectory, with nearly half of patients postinjury outcomes. Based on the data presented here, reaching a favorable outcome by 6 months. The recovery there is a decreased rate of improvement from 3 months trajectory does not appear to flatten, suggesting a possible to 6 months compared with rates of improvement from 1 continued improvement if these patients were closely fol- month to 3 months. It would be interesting to determine lowed longer than 6 months postinjury. at what point this rate of improvement ceases to continue.

Limitations Conclusions This study is limited in that it is a subset analysis of a In this secondary analysis of the Phase 3 COBRIT trial,

J Neurosurg Volume 131 • August 2019 601

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

TBI patients with less than 10 mm of midline shift on the M, Eisenberg H, Jane JA, et al: The diagnosis of head injury admission head CT scan had significantly improved func- requires a classification based on computed axial tomogra- tional outcomes through 180 days after injury compared phy. J Neurotrauma 9 (Suppl 1):S287–S292, 1992 17. Massaro F, Lanotte M, Faccani G, Triolo C: One hundred and with those with greater than 10 mm of midline shift. How- twenty-seven cases of acute subdural haematoma operated ever, nearly 50% of patients with 10 mm of midline shift on. Correlation between CT scan findings and outcome. Acta will achieve a favorable outcome (GOS-E scores 4–8) by Neurochir (Wien) 138:185–191, 1996 6 months postinjury. 18. Mathew P, Oluoch-Olunya DL, Condon BR, Bullock R: Acute subdural haematoma in the conscious patient: outcome with initial non-operative management. Acta Neurochir References (Wien) 121:100–108, 1993 1. Aldrich EF, Eisenberg HM, Saydjari C, Foulkes MA, Jane 19. Plaisier BR, Blostein PA, Hurt KJ, Malangoni MA: With- JA, Marshall LF, et al: Predictors of mortality in severely holding/withdrawal of life support in trauma patients: is there head-injured patients with civilian gunshot wounds: a report an age bias? Am Surg 68:159–162, 2002 from the NIH Traumatic Coma Data Bank. Surg Neurol 20. Quattrocchi KB, Prasad P, Willits NH, Wagner FC Jr: Quan- 38:418–423, 1992 tification of midline shift as a predictor of poor outcome fol- 2. Athiappan S, Muthukumar N, Srinivasan US: Influence lowing head injury. Surg Neurol 35:183–188, 1991 of basal cisterns, midline shift and pathology on outcome 21. Rapenne T, Lenfant F, N’Guyen Kim L, Honnart D, Beaurain in head injury. Ann Acad Med Singapore 22 (3 Sup- J, Freysz M: [Predictive factors of short-term mortality in pl):452–455, 1993 patients with severe head injury.] Presse Med 26:1661–1665, 3. Bullock MR, Chesnut R, Ghajar J, Gordon D, Hartl R, Newell 1997 (Fr) DW, et al: Surgical management of acute subdural hemato- 22. Sakas DE, Bullock MR, Teasdale GM: One-year outcome mas. Neurosurgery 58 (3 Suppl):S16–S24, Si–iv, 2006 following craniotomy for traumatic hematoma in patients 4. Cook RJ, Dorsch NW, Fearnside MR, Chaseling R: Out- with fixed dilated pupils. J Neurosurg 82:961–965, 1995 come prediction in extradural haematomas. Acta Neurochir 23. Servadei F, Faccani G, Roccella P, Seracchioli A, Godano U, (Wien) 95:90–94, 1988 Ghadirpour R, et al: Asymptomatic extradural haematomas. 5. Duhaime AC, Gennarelli LM, Yachnis A: Acute subdural Results of a multicenter study of 158 cases in minor head hematoma: is the blood itself toxic? J Neurotrauma 11:669– injury. Acta Neurochir (Wien) 96:39–45, 1989 678, 1994 24. Teasdale GM, Pettigrew LE, Wilson JT, Murray G, Jennett 6. Eisenberg HM, Gary HE Jr, Aldrich EF, Saydjari C, Turner B: Analyzing outcome of treatment of severe head injury: a B, Foulkes MA, et al: Initial CT findings in 753 patients with review and update on advancing the use of the Glasgow Out- severe head injury. A report from the NIH Traumatic Coma come Scale. J Neurotrauma 15:587–597, 1998 Data Bank. J Neurosurg 73:688–698, 1990 25. Turgeon AF, Lauzier F, Simard JF, Scales DC, Burns KE, 7. Hutchinson PJ, Kolias AG, Timofeev IS, Corteen EA, Czos- Moore L, et al: Mortality associated with withdrawal of nyka M, Timothy J, et al: Trial of decompressive craniecto- life-sustaining therapy for patients with severe traumatic my for traumatic intracranial hypertension. N Engl J Med brain injury: a Canadian multicentre cohort study. CMAJ 375:1119–1130, 2016 183:1581–1588, 2011 8. Jacobs B, Beems T, van der Vliet TM, Diaz-Arrastia RR, 26. Vahedi K, Hofmeijer J, Juettler E, Vicaut E, George B, Algra Borm GF, Vos PE: Computed tomography and outcome A, et al: Early decompressive surgery in malignant infarc- in moderate and severe traumatic brain injury: hematoma tion of the middle cerebral artery: a pooled analysis of three volume and midline shift revisited. J Neurotrauma 28:203– randomised controlled trials. Lancet Neurol 6:215–222, 215, 2011 2007 9. Jenkins A, Mendelow AD, Graham DI, Nath FP, Teasdale 27. Valadka AB, Gopinath SP, Robertson CS: Midline shift after GM: Experimental intracerebral haematoma: the role of severe head injury: pathophysiologic implications. J Trauma blood constituents in early ischaemia. Br J Neurosurg 4:45– 49:1–10, 2000 51, 1990 28. Vedantam A, Robertson CS, Gopinath SP: Clinical char- 10. Jüttler E, Unterberg A, Woitzik J, Bösel J, Amiri H, Sakowitz acteristics and temporal profile of recovery in patients OW, et al: Hemicraniectomy in older patients with extensive with favorable outcomes at 6 months after severe trau- middle-cerebral-artery stroke. N Engl J Med 370:1091–1100, matic brain injury. J Neurosurg 129:234–240, 2018; DOI: 2014 10.3171/2017.3.JNS162720] 11. Kase CS, Williams JP, Wyatt DA, Mohr JP: Lobar intracere- 29. Wilberger JE Jr, Harris M, Diamond DL: Acute subdural bral hematomas: clinical and CT analysis of 22 cases. Neu- hematoma: morbidity, mortality, and operative timing. J rology 32:1146–1150, 1982 Neurosurg 74:212–218, 1991 12. Lee EJ, Hung YC, Wang LC, Chung KC, Chen HH: Factors 30. Wilson JT, Pettigrew LE, Teasdale GM: Structured inter- influencing the functional outcome of patients with acute views for the Glasgow Outcome Scale and the extended epidural hematomas: analysis of 200 patients undergoing Glasgow Outcome Scale: guidelines for their use. J Neu- surgery. J Trauma 45:946–952, 1998 rotrauma 15:573–585, 1998 13. Lehmann U, Regel G, Ellendorf B, Rickels E, Lorenz M, 31. Wong CW: CT and clinical criteria for conservative treat- Becker H, et al: [Initial cranial CT for evaluating the progno- ment of supratentorial traumatic intracerebral haematomas. sis of craniocerebral trauma.] Unfallchirurg 100:705–710, Acta Neurochir (Wien) 135:131–135, 1995 1997 (Ger) 32. Yamaki T, Hirakawa K, Ueguchi T, Tenjin H, Kuboyama T, 14. Levin HS, Boake C, Song J, Mccauley S, Contant C, Diaz- Nakagawa Y: Chronological evaluation of acute traumatic Marchan P, et al: Validity and sensitivity to change of the intracerebral haematoma. Acta Neurochir (Wien) 103:112– extended Glasgow Outcome Scale in mild to moderate trau- 115, 1990 matic brain injury. J Neurotrauma 18:575–584, 2001 33. Young B, Rapp RP, Norton JA, Haack D, Tibbs PA, Bean JR: 15. Levin HS, Gary HE Jr, Eisenberg HM, Ruff RM, Barth JT, Early prediction of outcome in head-injured patients. J Neu- Kreutzer J, et al: Neurobehavioral outcome 1 year after se- rosurg 54:300–303, 1981 vere head injury. J Neurosurg 73:699–709, 1990 34. Zafonte R, Friedewald WT, Lee SM, Levin B, Diaz-Arrastia 16. Marshall LF, Marshall SB, Klauber MR, Van Berkum Clark R, Ansel B, et al: The citicoline brain injury treatment (CO-

602 J Neurosurg Volume 131 • August 2019

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC Puffer et al.

BRIT) trial: design and methods. J Neurotrauma 26:2207– Author Contributions 2216, 2009 Conception and design: Puffer, Yue. Acquisition of data: Puffer. 35. Zafonte RD, Bagiella E, Ansel BM, Novack TA, Friedewald Analysis and interpretation of data: Puffer, Yue. Drafting the arti- WT, Hesdorffer DC, et al: Effect of citicoline on functional cle: Puffer, Yue. Critically revising the article: Okonkwo, Puffer, and cognitive status among patients with traumatic brain Mesley, Sharpless, Puccio, Diaz-Arrastia. Reviewed submitted injury: Citicoline Brain Injury Treatment Trial (COBRIT). version of manuscript: Okonkwo, Puffer, Mesley, Sharpless, Puc- JAMA 308:1993–2000, 2012 cio, Diaz-Arrastia. Approved the final version of the manuscript 36. Zumkeller M, Behrmann R, Heissler HE, Dietz H: Computed on behalf of all authors: Okonkwo. Statistical analysis: Puffer. tomographic criteria and survival rate for patients with acute Administrative/technical/material support: Mesley, Sharpless, . Neurosurgery 39:708–713, 1996 Fetzick, Puccio. Study supervision: Okonkwo, Billigen. Institu- tional business officer: Billigen. Disclosures The authors report no conflict of interest concerning the materi- Correspondence als or methods used in this study or the findings specified in this David O. Okonkwo: University of Pittsburgh Medical Center, paper. Pittsburgh, PA. [email protected].

J Neurosurg Volume 131 • August 2019 603

Unauthenticated | Downloaded 10/02/21 01:47 PM UTC