Traumatic brain injury (TBI) is the most disabling

Transcription

1 J Neurosurg 120: , 2014 AANS, 2014 Patients with brain contusions: predictors of outcome and relationship between radiological and clinical evolution Clinical article Corrado Iaccarino, M.D., 1 Paolo Schiavi, M.D., 1 Edoardo Picetti, M.D., 2 Matteo Goldoni, Ph.D., 3 Davide Cerasti, M.D., 4 Marialuisa Caspani, M.D., 2 and Franco Servadei, M.D. 1 1 Arcispedale Santa Maria Nuova Istituto di Ricovero e Cura a Carattere Scientifico, Reggio Emilia; and Neurosurgery-Neurotraumatology Unit, 2 Intensive Care Unit, 3 Department of Biostatistics, and 4 Department of Neuroradiology, University Hospital of Parma, Parma, Italy Object. Traumatic parenchymal mass lesions are common sequelae of traumatic brain injuries (TBIs). They occur in up to 8.2% of all TBI cases and 13% 35% of severe TBI cases, and they account for up to 20% of surgical intracranial lesions. Controversy exists concerning the association between radiological and clinical evolution of brain contusions. The aim of this study was to identify predictors of unfavorable outcome, analyze the evolution of brain contusions, and evaluate specific indications for surgery. Methods. In a retrospective, multicenter study, patients with brain contusions were identified in separate patient cohorts from 11 hospitals over a 4-year period ( ). Data on clinical parameters and course of the contusion were collected. Radiological parameters were registered by using CT images taken at the time of hospital admission and at subsequent follow-up times. Patients who underwent surgical procedures were identified. Outcomes were evaluated 6 months after trauma by using the Glasgow Outcome Scale-Extended. Results. Multivariate analysis revealed the following reliable predictors of unfavorable outcome: 1) increased patient age, 2) lower Glasgow Coma Scale score at first evaluation, 3) clinical deterioration in the first hours after trauma, and 4) onset or increase of midline shift on follow-up CT images. Further multivariate analysis identified the following as statistically significant predictors of clinical deterioration during the first hours after trauma: 1) onset of or increase in midline shift on follow-up CT images (p < 0.001) and 2) increased effacement of basal cisterns on follow-up CT images (p < 0.001). Conclusions. In TBI patients with cerebral contusion, the onset of clinical deterioration is predictably associated with the onset or increase of midline shift and worsened status of basal cisterns but not with hematoma or edema volume increase. A combination of clinical deterioration and increased midline shift/basal cistern compression is the most reasonable indicator for surgery. ( Key Words cerebral contusion radiological evolution outcome clinical deterioration traumatic brain injury Traumatic brain injury (TBI) is the most disabling of traumatic injuries, often leading to lifelong physical, cognitive, behavioral, and emotional impairment. 17 The incidence of TBI throughout Europe ranges from 91 to 546 cases per 100,000 population per year. If the extreme rates are deleted, the overall rate becomes about 235 cases per 100,000 population per year. 34 In North America, the incidence ranges between 47 and 618 cases per 100,000 population per year. 12 Common sequelae of TBIs are traumatic parenchymal mass lesions, Abbreviations used in this paper: GCS = Glasgow Coma Scale; GOSE = Glasgow Outcome Scale-Extended; ICP = intracranial pressure; IQR = interquartile range; ROC = receiver operating characteristic; TBI = traumatic brain injury. which occur in up to 8.2% of all TBI cases, in 13% 35% of severe TBI cases, and which account for up to 20% of all surgical intracranial lesions in a representative series. 2 Although the frequency of posttraumatic mass lesions is higher among comatose patients, these lesions also occur in patients with mild or moderate head injury. Most of these patients will recover without deleterious sequelae, but a few will progressively deteriorate, even to death (talk-and-die cases). 24 Repeated imaging of cerebral contusions typically shows a progressive increase in mass lesions. 4,6,21,22,32 These events may represent the effects of hematoma expansion, of perihematoma edema, or even of the development of new lesions in previously uninjured brain areas. 27,32 The current scientific literature lacks uniformity in 908 J Neurosurg / Volume 120 / April 2014

2 Brain contusions: radiological and clinical evolution defining the term radiological evolution, and the lack of agreement regarding this clinical definition might be one reason for the wide range of reported hemorrhagic progressions of contusions (from 16.4% to 51.0%) reported by different clinical studies. 1,4,6,21,22,35,37 Considering the natural history of cerebral contusion, a major clinical question remains as to when and how to perform surgery for a patient harboring a brain contusion. More than 20 years ago, Marshall et al. 18 proposed a classification of head injury based on cerebral CT images. They stratified head injury according to the status of the mesencephalic cisterns, the presence of midline shift (> 5 mm), and the volume of the main intracranial lesion (> 25 ml). Patients with a lesion volume greater than 25 ml who had not undergone surgical intervention were classified as a nonoperated mass lesion cohort, indirectly suggesting that surgery was indicated for such patients. According to a recent review by the Brain Trauma Foundation, 2 current clinical indications for surgery for traumatic brain contusions comprise an amalgam of clinical and radiographic criteria, including lower Glasgow Coma Scale (GCS) score, presence of neurological deterioration, location of contusion, increased lesion volume, CT image appearance (increased midline shift and/or basal cistern compression), and increased intracranial pressure (ICP). Among these parameters, the most frequent factors used by attending neurosurgeons as criteria for surgical intervention for posttraumatic parenchymal damage are radiological and clinical deterioration. 25 Our aim was to evaluate specific clinical factors that might be accurately predictive of outcome and to investigate the association between clinical and radiological deterioration. We also analyzed indications for surgery in patients with cerebral contusion. Methods Inclusion and Exclusion Criteria We identified and retrospectively analyzed data from a prospectively registered database. We selected the records of all patients with a history of TBI and a CT diagnosis of cerebral contusion who had been treated during from January 2008 to December 2011 in 11 hospitals of the northwestern Emilia region of Italy. In this area there are 2 central hospitals with neurosurgical departments (Parma and Reggio Emilia) and 9 peripheral hospitals that are connected by a continuous teleradiology service that enables decision making with regard to the suitability of transferring a patient on the basis of imaging and clinical information. The teleradiology system is combined with area guidelines for management of head injuries. 10 Patients with the following characteristics were considered eligible for inclusion in the study: a cerebral contusion as the main posttraumatic intracranial lesion; a diagnosis of cerebral contusion with hemorrhagic volume greater than 1 ml, as also reported by Chang et al.; 4 at least 3 CT scans acquired during hospitalization; complete and available clinical data with particular attention to any alterations in neurological examination findings during the first hours after trauma; and hospitalization on the 1st day of TBI. J Neurosurg / Volume 120 / April 2014 We initially identified 629 patients but subsequently excluded 277 patients from the study because of 1 or more of the following: cerebral contusion volume less than 1 ml, unsatisfactory and incomplete clinical report, lack of available follow-up CT scans, and impossibility of assessing patients outcome after 6 months. The remaining 352 patients were included in the data analyses associated with this study. Clinical Data Collection The following clinical data were collected and analyzed for all patients: age, sex, mechanism of injury, results of first GCS evaluation, history of concurrent conditions (hypertension and/or cardiopathy and/or diabetes), treatment with an antiaggregant or anticoagulant, onset of neurological deterioration during the first 12 hours after trauma, and neurosurgical intervention. Patients were classified as neurologically deteriorating if the GCS score decreased by at least 2 points or if onset of pupillary abnormalities (as defined by Morris et al. 20 ) was registered. The mechanism of injury was classified as either highenergy trauma or low-energy trauma. 31 All other data were identified from clinical reports. The neurosurgical procedures were identified from an online register from the neurosurgical hospitals. Radiological Assessment For each patient, we collected the first 3 CT scans, including the one taken at the time of hospital admission. The CT images were reread by a central reader (D.C., a neuroradiologist), who was mostly blinded as to the time of the scan. Other radiological data collected and analyzed included the number of cerebral contusions, the cerebral contusion location, hematoma volume, edema volume, the presence of midline shift (> 5 mm), the presence of basal cistern effacement, the presence of other posttraumatic intracranial lesions (subarachnoid hemorrhage, subdural hematoma, extradural hematoma, intraventricular hemorrhage, cranial fracture), and the number of intracranial lesions identified in association with the cerebral contusion. As many as 5 cerebral contusions per patient were recorded. The location of the cerebral contusion was distinguished regionally as frontal, temporal, parietal, occipital, posterior fossa (cerebellum or brainstem), or basal nucleus. Hematoma volume was calculated by using the following formula: volume = (ABC)/2 (cm 3 ). 16 For patients with more than 1 cerebral contusion, the volume of each contusion was calculated and then added to obtain the total volume of contusion. For most patients, the edema component had a hypodense circumferential, not regular, aspect. According to the attending neuroradiologists, we registered 2 measurements: one including only the hyperdense (hemorrhagic) component and the other including the hypodense (pericontusion edema) and hyperdense (hemorrhagic) components of the lesion. By subtracting the first measurement from the second, we obtained a volume that was considered a reliable estimate of the hypodense component constituted by edematous tissue in the first few hours after trauma. 24 The midline shift was measured by 2 physicians, and the cases were divided ac- 909

3 C. Iaccarino et al. cording to a midline shift of more or less than 5 mm. The status of basal cisterns was categorized as normal versus abnormal (compressed or absent). On the second and third set of CT scans, we assessed the following specific parameters: percentage increase of hematoma size, expansion of the edema, new onset or an increase of at least 2 mm in the midline shift, worsening of the status of the basal cisterns, expansion of at least one other posttraumatic intracranial lesion. The CT scans were reread by a central reader (D.C.), who was mostly blinded as to the time of the scan. Evolution of hematoma size was defined as significant if enlargement of 30% of the original size was noted on CT scans, according to a recently published study. 1 Before choosing this expansion cutoff, we searched the literature for a recognized cutoff. We identified a cutoff of 30% volume increase in accordance with the more recent articles published on this issue, which considered increases of 25%, 30%, and 33%. 1,22,35 The percentage increase of hematoma size was also inserted into the database as a linear variable. Unlike hemorrhage, a cutoff value for expansion of the edema component of a contusion has not been established in the medical literature. In agreement with the previously defined cutoff, 1 an increase of more than 30% of edema volume was considered significant. A worsening in the status of the basal cisterns was registered if a normal parameter became abnormal (compressed or absent cisterns). Expansion of other posttraumatic intracranial lesions was evaluated by the attending neuroradiologist; all patients were stratified into 2 categories: 1) those with stable lesions or 2) those with an increase of at least 1 intracranial lesion. Outcome Measurements Initially, we carefully searched for death certificates in the archives to identify those patients who died during the first 6 months after TBI. Subsequently, after obtaining informed consent, we administered a Glasgow Outcome Scale-Extended (GOSE) by telephone to all patients 6 months after trauma. If it was not possible to evaluate the patient s outcome, we excluded the patient from the collective database. According to published studies, 36 the outcome was classified as either favorable (GOSE Scores 5 8) or unfavorable (GOSE Scores 1 4). Statistical Analyses Prognostic factors were assessed by means of binary logistic regression models as follows: 1) univariate p value was calculated by using a logistic regression with single variables as predictors; and 2) multivariate p value was calculated by using an adjusted logistic regression model, including all predictors with a p value of at least 0.2 in the univariate model. Furthermore, multicollinearity was excluded by performing a correlation matrix analysis (Pearson and Spearman correlation tests), considering the most clinically significant prognostic factor when the coefficient of determination (R2) is greater than 0.5. Residual output was also assessed to check residual values above or below 5 SDs, a cutoff based on the sample size of our patient group. For assessment of the diagnostic power of a given prognostic factor, a receiver operating characteristic (ROC) curve was used to calculate the area under the curve with 95% CIs and the cutoff value (for example, the flex point of the curve, where the sum of sensitivity and specificity was the maximum). Results Patient Demographics and Cerebral Contusion Characteristics A total of 352 patients met the inclusion criteria for the study; their demographic data are listed in Table 1. The GCS scores at the time of admission were distributed as follows: 41.5% mild head injury (GCS score of 14 15), 29.8% moderate head injury (GCS score of 9 13), and 28.7% severe head injury (GCS score of 3 8). Most patients (n = 281 [80%]) were treated in hospitals with a neurosurgical unit, whereas only 71 (20%) were admitted to peripheral hospitals under neurosurgical supervision, where all inclusion criteria were observed. The mean time from injury to initial CT scan was 2 hours (range 47 minutes 11 hours, interquartile range [IQR] minutes). The second set of CT images was obtained at an average of 9 hours (range 2 18 hours, IQR minutes) after the initial scan. The third set of CT scans was obtained an average of 38 hours (range hours, IQR hours) after the initial scan. Computed tomography parameters at the time of admission (admission CT) are shown in Table 2. Of the 352 patients, 91 (25.9%) had more than 1 cerebral contusion. The average volume of intraparenchymal bleeding was 12.2 ± 10.7 ml (range ml). For 307 patients, the volume of the single contusion was 1 10 ml; for 28 patients, the volume was ml; and for 17 patients, the volume was more than 25 ml. The average (± SD) edema volume detected on admission CT images was 3.7 ± 4.9 ml. Most cerebral contusions were located in the frontal (53.2%) and temporal (29.8%) lobes; the rest were scattered throughout the parietal (5.9%) and occipital (0.6%) lobes, posterior fossa (5.3%), and basal nucleus (5.2%). The cerebral contusion was in the right hemisphere for 41% of patients and in the left hemisphere for 51.9%. Most patients had at least 1 other intracranial posttraumatic lesion. Frequency of associated lesions is shown in Table 2. An urgent neurosurgical procedure was performed for 15 patients (4.3%) after admission CT scan, and a delayed neurosurgical intervention was performed for 46 (13.1%) patients (41 after the second CT scan and 5 after the third CT scan). All patients surgically treated after the first CT scan had had a severe TBI and more than 20 ml of intraparenchymal bleeding; of these 15 patients, 14 had a midline shift, 12 had basal cistern effacement, and 11 had a single cerebral contusion. For 13 of these 15 patients, a decompressive craniectomy was performed in association with lesion evacuation. Of the 46 patients who underwent delayed neurosurgical intervention, 35 underwent a craniotomy with evacuation of cerebral contusion and 11 also underwent a decompressive craniectomy. Characteristics of patients who underwent surgical treatment after the second CT scan are shown in Table 3. The mean ± SD age for all 61 patients undergoing surgery was 47.8 ± 29.4 years. 910 J Neurosurg / Volume 120 / April 2014

4 Brain contusions: radiological and clinical evolution TABLE 1: Description of population: epidemiological data Variable No. of Patients (%)* treatment site neurosurgical hospital (67.1) neurosurgical hospital 2 45 (12.8) peripheral hospital 71 (20.1) patient sex M 256 (72.7) F 96 (27.3) comorbidities absent 217 (64.5) present 125 (35.5) antiaggregant therapy absent 273 (77.6) present 79 (22.4) anticoagulant therapy absent 330 (93.8) present 22 (6.2) mechanism of injury high-energy trauma 206 (41.3) low-energy trauma 145 (58.7) GCS score at admission (28.7) (29.8) (41.5) surgical procedure 45 (12.7) GOSE score patient age 3 98 yrs unfavorable 121 (34.4) favorable 231 (65.6) patient age yr unfavorable 65 (18.4) favorable 287 (81.6) * Mean patient age ± SD was 59.1 ± 23.4 years. Parma, Italy. Reggio Emilia, Italy. Evolution of Neurological and Radiological Parameters on CT Images The evolution of clinical and radiological parameters is shown in Table 4. During the first 12 hours after trauma, neurological examinations revealed the following: clinical deterioration for 111 (31.5%) patients, clinical improvement for 22 (6.3%), and stable neurological function for 219 (62.2%). Concerning CT parameters assessed in this study, lesion volume decreased for 43 (12.2%) patients and remained unchanged between the first and second CT scans for 103 (29.3%). All lesions that showed reduced volume contained less than 2 ml. No lesions showing a reduction between the first and the second CT scan showed an evolution between the second and the third CT scan. Of the 206 patients in whom hematoma volume increased J Neurosurg / Volume 120 / April 2014 TABLE 2: Parameters measured on CT images taken at time of admission Variable No. of Patients (%) no. of cerebral contusions (74.1) 2 91 (25.9) total intraparenchymal bleeding (ml) (87.2) (8.1) >25 17 (4.7) midline shift >5 mm on admission CT image present 76 (21.6) absent 276 (78.4) basal cisterns status on admission CT image absent or compressed 69 (19.6) normal 283 (80.4) other associated lesions subdural hematoma 169 (48) subarachnoid hemorrhage 242 (68.8) epidural hematoma 40 (11.4) intraventricular hemorrhage 32 (9.1) cranial fracture 145 (41.2) between the first and the second CT scans, for 135 (65.5%) volume increased by 30% or more. Moreover, between the second and the third CT scan, for only 14 patients (36.8%), lesion volume increased by more than 30%. Overall, a significant (> 30%) evolution of total intraparenchymal bleeding was observed for 149 patients (42.3%). Increased cerebral edema was noted for 162 patients; among those, the increase was noted between the first and second CT scans for 59 patients (36.4%) and between the second and third CT scans for 103 patients (63.6%). A verifiable midline shift was documented on the admission CT scans for 76 (18.1%) patients. Overall, for 97 patients, a worsening of midline shift was observed on follow-up CT scans. For 70 patients, a new midline shift appeared on the second CT scan in the absence of any previous signs of shift (Fig. 1). For 27 patients, a midline shift was document on the admission CT scans and increased on follow-up CT scans. For 107 patients, worsening of the status of the basal cisterns was observed on follow-up CT scans. For 76 patients, although the appearance of the basal cisterns was normal on the admission CT scans, on the follow-up scans, the basal cisterns had become compressed or absent. Of the 31 patients in whom basal cisterns were absent or compressed on the admission CT scans, the status of the cisterns had worsened at the time of follow-up CT. Analysis of the variables associated with a single contusion (coexistence of other cerebral contusion, localization, and volume at first CT scan) revealed that the volume of the single contusion is predictive for the evolution of the hematoma (ROC curve documented that a value of 4 ml or less has a sensitivity of 95% and specificity of 75% for predicting absence of evolution). 911

5 TABLE 3: Association between clinical and radiological parameters and need for surgery after second CT scan C. Iaccarino et al. Variable Surgery (n = 46) No Surgery (n = 291) Univariate p Value Multivariate p Value* GCS score at admission < < mean patient age (yrs) ± SD 55.6 ± ± 25.1 < < worsening clinical condition < radiological appearance increase or onset of midline shift < worsening of basal cistern status < evolution of hematoma increased edema patient outcome favorable 21 (45.6%) 204 (70.1%) severe disability 18 (39.1%) 44 (15.1%) death 7 (15.2%) 43 (14.7%) * Boldface indicates statistical significance. Outcome Prognostic Factors The results of univariate and multivariate analysis to identify factors predicting a favorable outcome are shown in Table 5. The analyzed data excluded an association between outcome and intraventricular hemorrhage and expansion of other intracranial lesions. There was no relevant difference in the risk for unfavorable outcome among patients treated in hospitals with or without a neurosurgical department (p = 0.315), indirectly suggesting that patients admitted to and treated in peripheral hospitals after a selection is made via telemedicine do not TABLE 4: Clinical and radiological evolution for 352 patients Variable No. of Cases (%) neurological deterioration absent 241 (68.5) present 111 (31.5) evolution of hematoma absent 203 (57.7) present 149 (42.3) increased edema vol absent 190 (54) present 162 (46) midline shift on 1st CT absent 276 (21.6) present 76 (78.4) appearance or increase of midline shift absent 255 (72.4) present 97 (27.6) basal cistern status on admission CT absent or compressed 69 (19.6) normal 283 (80.4) onset of or increase in basal cistern effacement absent 245 (69.6) present 107 (30.4) Fig. 1. Axial CT scans of TBI patient. A and B: Scan obtained at time of admission, showing a small amount of temporobasal bleeding in the absence of documented midline shift. C and D: Scan acquired 5 hours after admission, showing a temporal cerebral contusion and the onset of midline shift. 912 J Neurosurg / Volume 120 / April 2014

6 Brain contusions: radiological and clinical evolution TABLE 5: Predictors of favorable/unfavorable outcome* Predictor Univariate p Value Multivariate p Value Exp(B) 95% CI per EXP(B) age < sex hypertension cardiopathy < diabetes antiaggregant therapy < anticoagulant therapy INR at admission < mechanism of injury < GCS score at admission 3 8 < < < SAH (1st scan) SDH (1st scan) EDH (1st scan) cranial fracture total hematoma vol < midline shift on admission CT < clinical deterioration < surgery hematoma evolution edema vol evolution < midline shift on follow-up CT < basal cistern status on admission CT < basal cistern status on follow-up CT < increased hematoma vol < * EDH = epidural hematoma; INR = international normalized ratio; SAH = subarachnoid hemorrhage; SDH = subdural hematoma. have a worse outcome than do patients directly admitted to neurosurgery. For all other variables, univariate analysis documented statistical significance. To document the predictive power for unfavorable outcome, we created a logistic regression model with 26 parameters. The following factors showed statistical significance: 1) older patient age (p < 0.001); 2) low GCS scores at first evaluation (p < 0.001); 3) clinical deterioration in the first hours after trauma (p = 0.003); and 4) onset or increase of midline shift at follow-up CT (p < 0.001). To evaluate whether there was a reliable cutoff after which prognosis rapidly worsens, we created an ROC curve for prognostic value of patient age. The identified cutoff age was 69 years, but the sensitivity (72.7%) and the specificity (75.3%) of the analysis were too low to authorize use of this cutoff in medical practice. No association was found between evolution of parenchymal bleeding and outcome. Similarly, no predictive prognostic power was documented for edema expansion. To investigate whether a dichotomization for hematoma expansion with a higher cutoff would give different results, we performed 2 univariate analyses. The first analysis compared patients with a hematoma expansion cutoff of 50% (288 patients without and 64 patients with J Neurosurg / Volume 120 / April 2014 hematoma expansion), and the second analysis compared patients with a doubling in hematoma size (313 patients without and 39 patients with doubling). No statistical significance was documented by either analysis (p = and p = 169, respectively). Correlation Between Radiological Progression and Clinical Deterioration To prove the association between clinical deterioration and specific radiological parameters, we performed univariate and logistic regression analyses (Table 6). For all patients with a midline shift greater than 5 mm, total intraparenchymal bleeding greater than 12 ml was documented. The mean total intraparenchymal bleeding among patients with midline shift was higher (16.3 ml) than that among patients without midline shift (8.1 ml). Moreover, the mean edema volume among patients with midline shift was higher (5.7 ml) than that among patients without midline shift (2.1 ml). Using multivariate analysis, we identified that the ability of onset or increase in midline shift at follow-up CT to predict clinical deterioration was statistically significant (p < 0.001) as was increased 913

7 C. Iaccarino et al. TABLE 6: Predictors of clinical deterioration Predictor Univariate p Value Multivariate p Value Exp(B) 95% CI per EXP(B) midline shift on admission CT total hematoma vol edema vol evolution < midline shift on follow-up CT < basal cisterns on admission CT < basal cisterns on follow-up CT < hematoma evolution < increased hematoma vol < basal cistern effacement at follow-up CT (p < 0.001) (Fig. 2). These same parameters as measured on admission CT scans were not significantly associated with clinical deterioration. The distribution of the different radiological parameters among patients with and without worsening of neurological function is shown in Table 7. Discussion Brain contusions are the most frequent type of posttraumatic intracerebral lesions. Because they usually occur in combination with other types of hematoma, it is difficult to collect a pure series of patients with isolated brain contusions. Unlike previously published studies, 1 our study included only those patients for whom the brain contusion was the main lesion and/or the primary reason for surgery. Because many of these patients are in hospitals that do not have a neurosurgical unit but that are linked to a neurosurgery department via image transmission, 14 we included patients who received treatment at both the primary hospital neurosurgical unit and the peripheral hospitals under supervision. Unlike the United States, where only 10% of patients are transferred to neurosurgical units of major hospitals from peripheral hospitals, 13 in Europe, more than 40% are transferred to a major trauma center from other smaller hospitals; some countries like Italy, France, or the United Kingdom report well above 50% referrals from other hospitals. 13 Our data demonstrate that among patients with brain contusions, the management of a minority of cases (71 patients [20.1%]) outside a major neurosurgical center, with the aid of teleradiology and area guidelines, 10 was not detrimental to patient outcome. In our study, older age was associated with worse outcomes in a continuous way (linear relationship), as previously reported. 13 When patient age exceeded 70 years, clinical prognosis worsened but with very low specificity and sensitivity (sensitivity 72.7%; specificity 75.3%). Age was found to be so prognostic that all medical risk factors (heart disease, diabetes, and hypertension) that are significant on univariate analysis are included by age on multivariate analysis. In previous studies, we have shown how the use of antiplatelet agents in elderly patients increases the risk for intracranial lesions (mainly contusions) after mild traumatic head injury. 11 These data were only partially confirmed because, in our study, the use of antiplatelet agents clearly interacted with age in the multivariate analysis. Other studies have shown that a low number of platelets (< 100,000) is associated with hematoma progression and a 9-fold increased risk for death. 1 Glasgow Coma Scale scores have also been shown to be highly predictive of prognosis. 6,32,35 One of the most interesting observations of our study is the high percentage (32% [111/352]) of patients showing clinical deterioration during hospitalization. A worsening of neurological outcome occurred overall for 1 patient out of 3 and, more specifically, for 5% of patients with mild TBI, 39% of patients with moderate TBI, and 62% of patients with severe TBI. As previously suggested, 8 these data confirm that neurosurgeons on duty await clinical deterioration before deciding in favor of surgery. Another interesting finding is the high risk for clinical deterioration in patients with moderate head injury (39%). In a previous multicenter study of patients with moderate TBI, harboring all types of intracranial hematomas, clinical worsening occurred for 28%, 7 thus confirming the high risk of deterioration for this population. Multivariate analyses revealed that none of the prognostic parameters of admission CT scans (hematoma volume, presence of associated lesions, midline shift, and basal cistern status) were associated with clinical outcome. Admission CT scans did not predict outcome because contusions by definition are lesions prone to evolve and outcomes are related to the evolved CT scan as previously reported. 26 We defined an evolved hematoma as one that increased by more than 30%. This limit was arbitrarily chosen and is in agreement with previously published criteria. 1 Increases in cerebral contusion volume of 25% 22 and 50% 6 have also been considered. When TBI results in a contusion, the hemorrhagic lesion often expands or a new hemorrhagic lesion might develop remotely (noncontiguously) from the original contusion during the first several hours after impact. Hemorrhagic progression of a contusion might be detected on CT scans even in patients with mild head injury. 17 For a delayed progression of hemorrhage to occur, microvessels must have been ruptured at the time of impact or microvessels unaffected at the time of impact must show subsequent rupture, hours after the impact, causing extravasation of blood. Recent findings suggest that the latter scenario prevails. 23,29 Severe cerebral contusion is often associated with nonhemorrhagic mass effects that progress rapidly within 914 J Neurosurg / Volume 120 / April 2014

8 Brain contusions: radiological and clinical evolution Fig. 2. Upper: Patient with 2 cerebral contusions; intracranial hypertension and midline shift are present. Lower: Patient with 2 large frontal bilateral cerebral contusions; common signs of intracranial hypertension and midline shift are absent hours after trauma. The mechanisms underlying such a rapid progression of mass effect cannot be fully explained by classic concepts of vasogenic and cytotoxic brain edema. Data from previous clinical trials, including diffusion-weighted MR imaging studies, have indicated that cells in the central (core) area of the contusion undergo shrinkage, disintegration, and homogenization, whereas cellular swelling is located predominantly in the peripheral (rim) area during this period. 17 A recent study J Neurosurg / Volume 120 / April 2014 by Kawamata et al. 15 demonstrated that cerebral contusion induces a rapid increase in tissue osmolality without contribution from inorganic ion fluxes. Furthermore, in this study, the authors suggest that the increase in colloid osmotic pressure through the metabolic production of osmoles or the release of idiogenic osmoles is probably the main cause of contusion edema. According to the literature, factors predicting hematoma evolution include the presence of subarachnoid hemorrhage, 6,22 presence of an associated subdural hematoma, 1,22 and high volume of contusion at admission. 1,4 Contusions of less than 10 ml were never associated with hematoma evolution. 1 In our study, we found a similar observation for single lesions of less than 4 ml, confirming a link between hematoma evolution and contusion size at admission. Regarding the existing issue of multiple contusions (98/352 patients [27.8%]), we believe that the risk for clinical deterioration is associated with the sum of the volumes of the different hematomas, so we considered the additive volume of the different intraparenchymal hematomas as a risk factor for every patient. Since publication of the Marshall classification 18 for CT scans of TBI patients, the limit of 25 ml has been used as the clinical standard and guideline for surgical evacuation of hematomas, including those showing minor changes. 2 We suggest that when 25 ml is used as a limit for surgical intervention, the fact that patients with multiple contusions differ from patients with a single contusion must be considered. We believe that the 25-ml cumulative index might be more a sensitive and predictive prognostic index of outcome and is therefore more useful. Somewhat surprisingly, according to multivariate analysis, the most reliable CT parameter related to outcome was the appearance or increase of midline shift. Midline shift in trauma patients has long been considered an influential parameter in CT classification, but it has never been shown to be superior to other CT data. In our study, we attempted to correlate clinical worsening with deterioration on CT scans. Our univariate analysis data (Table 7) demonstrated that worsening of the clinical neurological function is associated with hematoma evolution, edema increase, onset or increase of midline shift, and onset or increase of basal cistern effacement; our multivariate analysis data demonstrated that only onset and increase of midline shift and basal cistern effacement were significant. Our observations confirm those of Alahmadi et al., 1 who reported that not all patients with hematoma progression subsequently showed clinical deterioration. Conversely, Narayan et al. 21 documented a link between hematoma increase and clinical deterioration, although in a limited case series. These observations have a clear implication for surgical indications for patients with contusion. The factors currently used as guidelines typically include clinical deterioration, hematoma progression, and increased ICP (in monitored comatose patients). But what can we decide when hematoma progression does not correspond to clinical worsening (86/149 cases [57.7%]) or, to the contrary, when clinical worsening does not correspond to hematoma progression (48/203 cases [23.6%])? The correlation between radiological and clinical evolution remains a controversial issue. In a study of 116 patients, Smith et al. 30 reported that 5% of patients re- 915

9 C. Iaccarino et al. TABLE 7: Patients with and without neurological deterioration: distribution of different neuroradiological parameters Neurological Deterioration (No. [%]) Present (n = 111) Absent (n = 241) Parameter Univariate p Value Multivariate p Value hematoma evolution yes (n = 149) 63 (56.7) 86 (35.7) no (n = 203) 48 (43.3) 155 (64.3) edema increase < yes (n = 162) 77 (69.4) 85 (35.3) no (n =190) 34 (30.6) 156 (64.7) onset or increase in midline shift < < yes (n = 97) 89 (80.2) 8 (3.3) no (n = 255) 22 (19.8) 233 (96.7) onset or increase in effacement of basal cisterns < < yes (n = 107) 74 (66.7) 33 (13.7) no (n = 245) 37 (33.3) 208 (86.3) quired neurosurgical intervention after routine follow-up head CT indicating delayed cerebral lesions. However, in the above-cited study, the patients who required delayed surgery always exhibited associated clinical changes. Furthermore Alahmadi et al. 1 analyzed 98 patients with brain contusions and reported that about half of the contusions managed conservatively would progress radiologically over time in hospitalized patients. However, not all patients with radiological progression showed clinical deterioration or required surgical intervention. Alahmadi et al. reported that of the 17 patients who underwent delayed surgery, only 4 required a craniotomy for evacuation of cerebral contusions and the others underwent surgery for extracerebral posttraumatic hematoma or for insertion of an ICP monitor. Conversely, in our study, we restricted the definition of surgically treated patients to those requiring surgery for brain contusion. It seems that in many patients with brain contusions a correlation between clinical and radiological parameters (hematoma and/or edema increase) is lacking. For our patient cohort, surgery to remove contusions was both early (4.3%) and delayed (11.6%). Of the 61 patients who underwent surgery, for only 15 was it performed early. In another recently published study, Compagnone et al. 8 evaluated surgical management in a group of 729 patients with intradural lesions; surgery was performed early for 404 patients (12.6% for contusions) and was delayed for 154 patients (31.2% for contusions). Taken together, the results of our study and those of the above-mentioned study suggest that patients with cerebral contusions seem to require delayed surgery on a more frequent basis. Our data suggest that a combination of clinical parameters (neurological deterioration) together with selected radiological parameters (including appearance or increase of midline shift and/or appearance of cisternal compression) are better predictors of outcome than simple hematoma evolution alone, as previously advocated. 3,19 Future studies are warranted to investigate whether these radiological parameters can predict clinical deterioration and therefore can be useful as guidelines for surgical indications before the onset of GCS score worsening. The reasons underlying why hematoma increase and/or pericontusion edema increase do not influence clinical deterioration are probably associated with the complexity of the anatomic lesions in these TBI patients. Many patients have multiple contusions and/or small extracerebral hematomas that contribute to brain compliance and genesis of mass effect. Only midline shift and cisternal compression are able to accurately depict brain compression leading to clinical deterioration. Another reason for the lack of correlation between hematoma evolution and clinical evolution could be a low (30%) cutoff for considering significant hematoma evolution. This cutoff was identified in accordance with the more recent articles published on this issue. 1,22,35 Additional analyses performed with a 50% increase and hematoma doubling confirmed the absence of a link between expansion of a single hematoma and patient outcome. We monitored ICP in 45 comatose patients after the second follow-up CT scan. In 18 patients (40%), pressure increased over 25 mm Hg despite CSF drainage and reinforced medical therapy. 33 Of these 18 patients, 16 underwent surgical intervention (10 also received an associated decompressive craniectomy). A recent prospective study demonstrated the lack of significant differences in outcome between comatose patients with or without ICP monitoring. 5 Additional prospective 9 and large retrospective studies 28 further demonstrated worsened outcomes in aggressively monitored and treated patients. Our study results suggest that ICP is the most useful guideline for surgical intervention in comatose patients with potentially evolving brain contusions. When measurements of clinical outcome are unavailable, ICP monitoring has clinical utility for surgical decision making for patients with posttraumatic intracranial mass lesions. This study has some limitations. These limitations are the retrospective design, the total number of participants (which remains small overall but larger than that of many other published studies), the potential for subjective determination of clinical and radiological measures among 916 J Neurosurg / Volume 120 / April 2014

10 Brain contusions: radiological and clinical evolution multiple centers, and potential selection bias for treatment regimens among patients. Conclusions Among patients with intracranial contusions, the following features are heterogeneous: clinical status (mild moderate and severe TBI), radiological findings (single or multiple lesions, association with other hematomas, hematoma evolution), and outcomes. Establishing surgical indications among this patient population is difficult; the most reasonable indicator is a combination of clinical deterioration and increased midline shift/basal cistern compression. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Servadei, Iaccarino, Schi avi, Caspani. Acquisition of data: Iaccarino, Schiavi, Picetti, Cer asti. Analysis and interpretation of data: Servadei, Picetti, Goldo ni, Cerasti. Drafting the article: Iaccarino, Schiavi, Picetti. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manu script on behalf of all authors: Servadei. Statistical analysis: Goldo ni. Administrative/technical/material support: Iaccarino, Schiavi, Goldo ni, Cerasti. Study supervision: Servadei, Caspani. References 1. Alahmadi H, Vachhrajani S, Cusimano MD: The natural history of brain contusion: an analysis of radiological and clinical progression. Clinical article. J Neurosurg 112: , Bullock MR, Chesnut R, Ghajar J, Gordon D, Hartl R, Newell DW, et al: Surgical management of traumatic parenchymal lesions. Neurosurgery 58 (3 Suppl):S25 S46, Si Siv, Bullock R, Golek J, Blake G: Traumatic intracerebral hematoma which patients should undergo surgical evacuation? CT scan features and ICP monitoring as a basis for decision making. Surg Neurol 32: , Chang EF, Meeker M, Holland MC: Acute traumatic intraparenchymal hemorrhage: risk factors for progression in the early post-injury period. Neurosurgery 58: , Chesnut RM, Temkin N, Carney N, Dikmen S, Rondina C, Videtta W, et al: A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med 367: , Chieregato A, Fainardi E, Morselli-Labate AM, Antonelli V, Compagnone C, Targa L, et al: Factors associated with neurological outcome and lesion progression in traumatic subarachnoid hemorrhage patients. Neurosurgery 56: , Compagnone C, d Avella D, Servadei F, Angileri FF, Brambilla G, Conti C, et al: Patients with moderate head injury: a prospective multicenter study of 315 patients. Neurosurgery 64: , Compagnone C, Murray GD, Teasdale GM, Maas AI, Esposito D, Princi P, et al: The management of patients with intradural post-traumatic mass lesions: a multicenter survey of current approaches to surgical management in 729 patients coordinated by the European Brain Injury Consortium. Neurosurgery 57: , Cooper DJ, Rosenfeld JV, Murray L, Arabi YM, Davies AR, D Urso P, et al: Decompressive craniectomy in diffuse traumatic brain injury. N Engl J Med 364: , 2011 (Erratum in N Engl J Med 365:2040, 2011) J Neurosurg / Volume 120 / April Fabbri A: Approccio nel DEA e gestione del trauma cranico lieve-moderato. Torino: Italian Society of Emergency Medicine, Fabbri A, Servadei F, Marchesini G, Stein SC, Vandelli A: Predicting intracranial lesions by antiplatelet agents in subjects with mild head injury. J Neurol Neurosurg Psychiatry 81: , Feigin VL, Theadom A, Barker-Collo S, Starkey NJ, McPherson K, Kahan M, et al: Incidence of traumatic brain injury in New Zealand: a population-based study. Lancet Neurol 12: 53 64, Hukkelhoven CW, Steyerberg EW, Rampen AJ, Farace E, Habbema JD, Marshall LF, et al: Patient age and outcome following severe traumatic brain injury: an analysis of 5600 patients. J Neurosurg 99: , Iaccarino C, Rapanà A, Compagnone C, Tagliaferri F, Servadei F: Teleradiology for traumatic brain injury, in Kumar S, Krupinski E (eds): Teleradiology. New York: Springer, 2008, pp Kawamata T, Mori T, Sato S, Katayama Y: Tissue hyperosmolality and brain edema in cerebral contusion. Neurosurg Focus 22(5):E5, Kothari RU, Brott T, Broderick JP, Barsan WG, Sauerbeck LR, Zuccarello M, et al: The ABCs of measuring intracerebral hemorrhage volumes. Stroke 27: , Kurland D, Hong C, Aarabi B, Gerzanich V, Simard JM: Hemorrhagic progression of a contusion after traumatic brain injury: a review. J Neurotrauma 29:19 31, Marshall LF, Marshall SB, Klauber MR, van Berkum Clark M, Eisenberg HM, Jane JA, et al: A new classification of head injury based on computerized tomography. J Neurosurg 75 (1S):S14 S20, Mathiesen T, Kakarieka A, Edner G: Traumatic intracerebral lesions without extracerebral haematoma in 218 patients. Acta Neurochir (Wien) 137: , Morris GF, Juul N, Marshall SB, Benedict B, Marshall LF: Neurological deterioration as a potential alternative endpoint in human clinical trials of experimental pharmacological agents for treatment of severe traumatic brain injuries. Neurosurgery 43: , Narayan RK, Maas AI, Servadei F, Skolnick BE, Tillinger MN, Marshall LF: Progression of traumatic intracerebral hemorrhage: a prospective observational study. J Neurotrauma 25: , Oertel M, Kelly DF, McArthur D, Boscardin WJ, Glenn TC, Lee JH, et al: Progressive hemorrhage after head trauma: predictors and consequences of the evolving injury. J Neurosurg 96: , Patel AD, Gerzanich V, Geng Z, Simard JM: Glibenclamide reduces hippocampal injury and preserves rapid spatial learning in a model of traumatic brain injury. J Neuropathol Exp Neurol 69: , Peterson EC, Chesnut RM: Talk and die revisited: bifrontal contusions and late deterioration. J Trauma 71: , Servadei F, Compagnone C, Sahuquillo J: The role of surgery in traumatic brain injury. Curr Opin Crit Care 13: , Servadei F, Murray GD, Penny K, Teasdale GM, Dearden M, Iannotti F, et al: The value of the worst computed tomographic scan in clinical studies of moderate and severe head injury. Neurosurgery 46:70 77, Servadei F, Nanni A, Nasi MT, Zappi D, Vergoni G, Giuliani G, et al: Evolving brain lesions in the first 12 hours after head injury: analysis of 37 comatose patients. Neurosurgery 37: , Shafi S, Diaz-Arrastia R, Madden C, Gentilello L: Intracranial pressure monitoring in brain-injured patients is associated with worsening of survival. J Trauma 64: , Simard JM, Kilbourne M, Tsymbalyuk O, Tosun C, Caridi J, 917

11 C. Iaccarino et al. Ivanova S, et al: Key role of sulfonylurea receptor 1 in progressive secondary hemorrhage after brain contusion. J Neurotrauma 26: , Smith JS, Chang EF, Rosenthal G, Meeker M, von Koch C, Manley GT, et al: The role of early follow-up computed tomography imaging in the management of traumatic brain injury patients with intracranial hemorrhage. J Trauma 63:75 82, Snoek A, Dekker M, Lagrand T, Epema A, van der Ploeg T, van den Brand JG: A clinical decision model identifies patients at risk for delayed diagnosed injuries after high-energy trauma. Eur J Emerg Med 20: , Soloniuk D, Pitts LH, Lovely M, Bartkowski H: Traumatic intracerebral hematomas: timing of appearance and indications for operative removal. J Trauma 26: , Stocchetti N, Zanaboni C, Colombo A, Citerio G, Beretta L, Ghisoni L, et al: Refractory intracranial hypertension and second-tier therapies in traumatic brain injury. Intensive Care Med 34: , Tagliaferri F, Compagnone C, Korsic M, Servadei F, Kraus J: A systematic review of brain injury epidemiology in Europe. Acta Neurochir (Wien) 148: , White CL, Griffith S, Caron JL: Early progression of traumatic cerebral contusions: characterization and risk factors. J Trauma 67: , Wilson JT, Pettigrew LE, Teasdale GM: Structured interviews for the Glasgow Outcome Scale and the extended Glasgow Outcome Scale: guidelines for their use. J Neurotrauma 15: , Yadav YR, Basoor A, Jain G, Nelson A: Expanding traumatic intracerebral contusion/hematoma. Neurol India 54: , 2006 Manuscript submitted May 26, Accepted December 11, Portions of this work were presented orally at the XV World Congress of Neurosurgery, Seoul, South Korea, September 8 13, Please include this information when citing this paper: published online February 7, 2014; DOI: / JNS Address correspondence to: Franco Servadei, M.D., Department of Emergency Medicine, Neurosurgery-Neurotraumatology Unit, University Hospital of Parma and ASMN-IRCCS Reggio Emilia, Viale Gramsci 14, Parma 43100, Italy. fservadei@ao.pr.it. 918 J Neurosurg / Volume 120 / April 2014