Introduction 

   Spontaneous intracerebral hemorrhage (ICH) is usually a monophasic event, but bleeding can persist for up to 6 hours postictus, and reports have indicated continued bleeding even later during the treatment course.^1)11)21) Persistent bleeding of an ICH may lead to expansion of the hematoma, which is highly predictive of neurological deterioration and is an independent predictor of mortality and functional outcome.^{5)6)10)11)14)18)20)} Therefore, predicting the development of hematoma expansion is important in effectively targeting clinical treatment of ICH, especially with the emergence of recombinant factor VIIa as an investigational treatment for acute stages of ICH.⁶⁾¹³⁾ 
   Unfortunately, there is still limited knowledge of an established clinical or radiographical indicator to identify patients at risk for hematoma expansion. Prior reports have suggested contrast extravasation during cerebral angiography and intrahematomal gadolinium during magnetic resonance imaging (MRI) as an indicator of hematoma expansion, but both are of limited use in an acute setting.¹⁶⁾²¹⁾ However, recent studies have reported a growing interest in the use of computed tomographic angiography(CTA) as a marker of hematoma expansion.^1)8)9)15)20) The use of CTA provides a rapid noninvasive assessment of the cerebral vasculature, and the presence of contrast within the hematoma has been thought to represent contrast extravasation due to ongoing bleeding.^1)15)20)
   In this study, we attempted to identify the association between extravasation and hematoma expansion, and assessed the clinical and radiographic characteristics of patients who exhibited contrast extravasation on initial CTA. 

Material and Methods 

   Ninety six patients who were diagnosed with intracerebral hemorrhage and who received CTA within 12 hours from initial onset of symptoms and who received a follow up brain CT within 48 hours from the initial CTA between April 2004 and March 2007 were retrospectively assessed. All patients who were diagnosed with ICH secondary to vascular malformation, hemorrhagic transformation of ischemic stroke, head trauma, and tumor were excluded. The patients' age ranged from 34 to 90 years with a mean age of 56.6 years. There were 40 men and 56 women. 
   Image acquisition of CTA was performed as quickly as possible after arrival. First, a nonenhanced head CT was performed using a helical CT scanner (16-slice GE Lightspeed plus: GE Medical Systems, Milwaukee, WI). This was followed by helical scanning during the administration of 120 mL of nonionic contrast agent at 3-5 mL/second with a 20 to 40 second prep delay using standard scan parameters of 120 to 140 kVp and 170 to 220 mAs.¹⁹⁾ Section thickness was 5 mm for nonenhanced scans and 1.25 mm for CT angiography. The presence of contrast extravasation was defined as the presence of high-density material within the hematoma (Fig. 1).⁹⁾²⁰⁾ 
   Hematoma volume on CT scans was calculated on the initial and follow-up CTs using the previously validated ellipsoid formulation of Kothari et al.¹²⁾: hemorrhage volume (mL) = ABC/2, where A is the greatest hemorrhage diameter by CT, B is the diameter 90 to A, and C is the approximate number of CT slices with hemorrhage multiplied by the slice thickness. When the hematoma was irregular in shape, the volume of each portion was estimated separately. Hematoma expansion was defined as an increase in volume of > 30% or > 6 mL from baseline on follow up brain CT (Fig. 1).¹³⁾²⁰⁾ 
   Patients were classified into the extravasation and no extravasation groups. Patient age, sex, initial Glasgow Coma Scale (GCS) score, hematoma location, initial and blood pressure after 6 hours, initial international normalized ratio of prothrombin time (INR), time span from onset of symptoms to initial CTA, and hematoma expansion were assessed in both groups and were compared using the Student's t test, Chi square test and Fisher's exact test. A p value of less than 0.05 was considered significant. All values are expressed as mean±standard deviation. 

Results 

   Of the 96 patients who received CTA, 15 (19%) demonstrated a presence of extravasation. The mean age was 56.40±10.02 years (range, 43-79) in the extravasation group and 56.68±15.22 years(range, 34-90) in the no extravasation group. No significant correlation was seen between extravasation and age. The initial GCS did not a show a significant difference between the two groups (13.18±2.95 in the extravasation group, 13.80± 3.58 in the no extravasation group). 6 patients were male and 9 female in the extravasation, compared to 34 and 47 in the no extravasation group, but did not show any statistical significance. The initial INR were in the normal range and did not show any difference between the two groups (1.10±0.07 in the extravasation group, 1.07±0.03 in the no extravasation group). Also, no significant differences were seen with respect to blood pressure between the two groups (Table 1). 
   The main location of ICH were in the deep supratentorial areas but did not exhibit a significant correlation with extravasation (Table 2). Hematoma expansion was observed in 21 (22%) of all patients, and a significantly higher rate of hematoma expansion was seen in the extravasation group compared to the no extravasation group (47% vs 17%, p=0.027). There was a trend toward the extravasation group having greater ICH volume compared to the no extravasation group, and the mean increase was also significantly higher in the extravasation group (7.67±11.41 vs 1.73±4.93, p=0.012, Table 3). All patients in the extravasation group underwent an initial CTA within 6 hours from onset of symptoms, and the mean time from onset of symptoms to initial CTA was significantly shorter in the extravasation group (3.5±1.3 hours vs 7.6±2.5 hours, p<0.001). Also, detection of extravasation on CTA significantly correlated with time from symptom onset to intial CTA, especially when it was less than 4 hours (p>0.001) (Table 4). 

Discussion 

   Spontaneous intracerebral hemorrhage accounts for 10 to 15% of all strokes and is associated with a higher mortality rate than either ischemic stroke or SAH, with up to 50% mortality at 30 days and 1 year survival of only 38%.^2)7)20) These rates are expected to rise as result of an aging population who are more susceptible to hypertension and amyloid angiopathy, which are widely acknowledged risk factors of ICH.²⁾⁷⁾ Accordingly, numerous studies have been undertaken to assess the possible predictors of higher mortality rates and poor outcome.¹⁾³⁾⁶⁾ Initial size of hematoma has been shown to be one of the most important predictors of 30-day mortality in ICH patients.³⁾ However, it has been well established that initial hemorrhage volume is not stable and frequently progresses, and usually within the first 6 hours of symptom onset.¹⁾¹¹⁾ An association between hematoma expansion and increased mortality and diminished functional outcome has been reported.^1)6)10) 
   Several large prospective and retrospective studies have reported rates of 14 to 38% of hematoma expansion after initial presentation.^2)4)6)7)9) Our study also exhibited a similar rate of hematoma growth (22%). Numerous factors have been identified to be associated with enlargement.^1)4)8)15) Fujii, et al.⁸⁾ reported that chronic alcohol consumption, short interval from onset, consciousness disturbance, irregularly shaped hematomas, and low levels of fibrinogen were independently associated hematoma expansion. Other predictors that have been identified include hyperglycemia, liver disease, and high blood pressure.¹³⁾ However, recent observations have focused on radiographical predictors of hematoma expansion, and especially extravasation of radiographic contrast on CTA.^1)6)15)20) 
   The mechanism of hematoma expansion is still not well understood, but several possible etiologies have been suggested in the past. Parenchymal microaneurysms that occur in deep brain structures and are usually associated with old age and hypertension has been described as the cause of spontaneous ICH. These microaneurysms has been described as not true aneurysms, but adherent clots or pseudoaneurysms, or more recently, as vascular tortuosity and coiling which produce the appearance of an aneurysm.^5)14)15) Secondary hemorrhage into the perihematoma tissue has been suggested as another mechanism of hematoma expansion. Mechanical disruption and ischemia caused by compression by the hematoma have been proposed, but recent studies have implicated inflammatory mediators as the cause.^8)18)20) However, even though extravasation has repetitively been associated with hematoma expansion, these mechanisms have yet to be fully described on cerebral angiography or CTA. 
   Recent studies have shown extravasation on CTA as a predictor of persistent hemorrhage and resultant hematoma expansion as well as an independent predictor of in-hospital mortality.¹⁾⁹⁾ Goldstein et al.⁹⁾ found an association of an 18- fold increase in the odds of hematoma expansion in patients with evidence of extravasation on CTA. The incidence rate of extravasation on CTA has been reported as approximately 30 to 50%.^1)8)9)20) Our study showed that 19% of all patients exhibited extravasation, which is slightly lower than others. This difference may be due to the fact that we did not take into account the delayed post contrast images which may exhibit further extravasation not seen on the initial source image.⁹⁾ However, hematoma expansion was observed in 47% of the patients who initially exhibited extravasation on CTA, which was similar to prior studies which found hematoma expansion rates of 46 to 60% in patients with extravasation.⁹⁾²⁰⁾ 
   Other studies have shown a number of factors to be associated with contrast extravasation on CTA. A larger hematoma size at baseline has been frequently identified as being more likely to demonstrate this finding.¹⁾ We noted that patients in the extravasation group had a trend towards initially having larger sized hematoma. It may be that extravasation is more easily detected in larger hematomas. The association of contrast extravasation with hypertension has also been debated.⁸⁾¹⁷⁾ Elevated blood pressure have been repeatedly associated with the risk of continued hemorrhage.¹⁾¹⁷⁾ However, we, as well as others, did not find this correlation.⁹⁾²⁰⁾ We also did not appreciate previously noted effects of level of consciousness and initial INR on extravasation.⁷⁾⁸⁾ 
   Since hematoma enlargement occurs primarily in the early hours after hemorrhage, it is natural to expect that the risk of extravasation would be higher in patients undergoing early CTA. Intravenously administered contrast material cannot cross the normal blood-barrier (BBB), and contrast leakage seen on CT has been shown to originate from ruptured vessels or from vessels with ischemia changes, which may explain the presence of extravasation in the early hours after hemorrhage.^8)16)21) Previous studies have reported time of symptom onset to initial CTA in the extravasation group to be 4 to 10 hours, and we also found a median time of 3.5 hours in the extravasation group, which was statistically significant.^1)9)20) 
   Several differences between our study and previous reports may have been attributed to the limitations of this study. First, our study was conducted retrospectively, and since the indications for CTA in ICH are not well established, no definite protocol was used when selecting patients who received initial CTA. This limited patient selection may have altered the generalizability of these results. Also, a relatively large number of patients did not fit the inclusion criteria, which may have limited our statistical power. Finally, although the definition of contrast extravasation has been adapted