Enlargement of Spontaneous Intracerebral Hemorrhage
Incidence and Time Course
Background and Purpose Standard radiographic criteria for hematoma enlargement have not been established. We undertook this investigation to assess the incidence and time course of hematoma growth using objective cutoff values.
Methods We reviewed the clinical records of 204 patients with spontaneous intracerebral hemorrhage treated nonsurgically who underwent initial computed tomography (CT) within 48 hours and repeat CT within 120 hours of the onset of symptoms. The consensus of five observers reading the CT films was considered the “gold standard” for hematoma enlargement. The discriminant values of the difference (V2−V1) or the ratio (V2/V1) of the hematoma volume on the initial (V1) and second (V2) CT scans were determined by use of receiver operating characteristic curves. We chose the cutpoint that had the highest sensitivity and specificity for identifying hematoma expansion.
Results The cutpoint for hematoma enlargement was determined as V2−V1=12.5 cm3 or V2/V1=1.4 (sensitivity=94.4%, specificity=95.8%). Forty-one patients (20%) had changes that exceeded these criteria. Frequency of hematoma expansion was greatest among those who underwent the initial CT scan early (27 [36%] of 74 patients at ≤3 hours) and progressively declined as the time to initial scan was prolonged (7 [16%] of 45 patients at 3 to 6 hours; 5 [15%] of 33 patients at 6 to 12 hours; 2 [6%] of 34 patients at 12 to 24 hours; and 0 [0%] of 18 patients at 24 to 48 hours).
Conclusions The enlargement of hematoma was defined radiographically as the increase of its volume by ≥12.5 cm3 or by ≥1.4 times. Although expansion of intracerebral hemorrhage on CT scan was common in the hyperacute stage, 17% of hematoma expansion occurred even after 6 hours of onset. Enlargement after 24 hours of onset seems extremely rare. Early CT scanning appears to increase the rate of detection of enlarging hematomas.
Intracerebral hemorrhage is one of the most devastating forms of stroke.1 Hematomas can enlarge in the acute phase.2 3 4 5 6 7 An increase in the volume of ICH itself is a main cause of clinical deterioration. Therefore, it is important to clarify the time course of hematoma evolution for clinicians who are searching for better ways to manage patients with ICH. The criteria for hematoma enlargement, however, have been arbitrary in previous studies.
The decision rule, defined by choosing a cutpoint on a given criterion, is often evaluated by computing sensitivity (true-positive rate) and specificity (1−false-positive rate). The ROC curve is the graph of the true-positive rate as a function of the false-positive rate, obtained by varying the decision thresholds repeatedly.8 Using ROC curve analyses, we studied changes in hematoma volume related to the interval between the onset of ICH and the timing of the initial CT scan to determine the frequency and time course of hematoma enlargement.
Subjects and Methods
We reviewed the records of 399 patients with ICH who were admitted to our stroke care unit from January 1, 1985, through December 31, 1994. We then selected 298 patients in whom CT had been performed within 48 hours of ictus. In general, a CT scan was repeated within a few days after admission even when clinical course was uneventful in our institute. There were some cases for which this rule did not apply because the scheduled day fell on a holiday. A second or additional CT scan was also performed when a clinical deterioration was noticed. Excluded were patients with hemorrhage due to aneurysmal rupture (1), arteriovenous malformations (6), moyamoya disease (4), or infective endocarditis (5), as well as those receiving anticoagulants (9) or antiplatelet agents (13). Patients who had died (14) or received neurosurgical treatment (6) before a second CT scan and those who had undergone a second CT scan more than 120 hours after the onset of ICH (36) were also excluded. We thus obtained 204 patients with spontaneous ICH who were medically treated and who had undergone an initial CT scan within 48 hours and a second CT scan within 120 hours of onset of symptoms (Fig 1⇓). These 204 patients (77 women and 127 men; mean age, 64.1±12.7 years) served as subjects for the present study. Hypertension was judged to be present if the patient fulfilled one or both of the following criteria: (1) a history of antihypertensive medication or (2) a systolic blood pressure >160 mm Hg or diastolic blood pressure >90 mm Hg on at least two occasions before the onset of ICH. The presence or absence of clinical deterioration, implying an aggravation of consciousness disturbance or neurological deficits, was judged from medical records. Clinical outcome was also obtained from medical records.
The interval from the onset of ICH to the first and second CT scans was recorded in each patient. For patients in whom a third and fourth CT scan was obtained within 120 hours, the intervals to the subsequent CT scans were also recorded. In cases of onset during sleep, the recognized time of onset was on awakening. All CT scans were reviewed by two of the authors (S.K. and H.Y.) and three of their associates until a consensus was reached as to the location and size of the hemorrhage as well as the expansion or nonexpansion of hematoma. ICHs were classified as putaminal, thalamic, lobar, pontine, cerebellar, or other. If the origin of a massive ICH could not be categorized as either thalamic or putaminal, such a case was designated as mixed. ICH volume was determined in the following manner (Fig 2⇓)9 10 11 : On the CT slice with the largest area of ICH, the largest diameter (A) of the hematoma was measured by use of the centimeter scale on the CT film. The diameter of the hemorrhage perpendicular to the largest diameter represented the second diameter (B). The height of the hematoma was calculated by multiplying the number of slices involved by the slice thickness, providing the third diameter (C). The three diameters were multiplied and then divided by 2 (A×B×C/2) to obtain the volume of ICH. Although the formula of an ellipsoid is 4/3(π)(a×b×c), where a, b, and c represent the respective radii of ICH in three dimensions, Kwak and coauthors9 validated the formula A×B×C/2 by computer measurements of hematoma volume, and Lisk and colleagues10 demonstrated the usefulness of this method. We preferred A×B×C/2 because of its simplicity for bedside use. Each diameter was measured to a half centimeter.12 The results were expressed in cubic centimeters. Hemorrhage within the ventricular system was not measured.
Considering the consensus of the observers reading the CT films as the “gold standard” criterion for hematoma enlargement, the discriminant values of the difference (V2−V1) or the ratio (V2/V1) of hematoma volume on the initial (V1) and second (V2) CT scans were determined by means of ROC curves. Statistical analysis was performed with the use of the SAS package. We chose the cutpoint as that which had the highest sensitivity and specificity for detecting hematoma enlargement.
Classification of hemorrhage location was as follows: 70 patients (34%) had putaminal hemorrhage; 67 (33%), thalamic; 5 (2%), mixed; 34 (17%), lobar; 11 (5%), pontine; 10 (5%), cerebellar; and 7 (3%), other sites. A history of hypertension was documented in 171 patients (84%). The mean interval from the stroke onset to the first CT scan was 9.0±10.2 hours and that from onset to the second CT scan was 35.8±31.1 hours (Table 1⇓).
The mean volume of ICH on the initial CT scan and volume change between the initial and the second CT scans (in parentheses) according to the time interval between the onset and the initial CT scan were as follows: 20.1±18.0 cm3 (8.1±16.6 cm3) at ≤3 hours, 15.9±18.2 cm3 (3.8±12.9 cm3) at 3 to 6 hours, 24.5±24.5 cm3 (5.0±12.1 cm3) at 6 to 12 hours, 18.2±22.9 cm3 (−0.2±3.4 cm3) at 12 to 24 hours, and 11.9±12.8 cm3 (−0.4±1.4 cm3) at 24 to 48 hours (Table 2⇓). The mean hematoma volume on the initial CT scan and volume change between the initial and the second CT scans (in parentheses) according to the location of ICH were as follows: putaminal, 18.3±15.0 cm3 (4.0±13.3 cm3); thalamic, 12.7±17.3 cm3 (3.7±11.4 cm3); mixed, 44.4±14.8 cm3 (17.5±27.7 cm3); lobar, 36.9±17.2 cm3 (6.3±15.5 cm3); pontine, 4.1±3.6 cm3 (3.8±7.6 cm3); cerebellar, 12.7±9.1 cm3 (5.2±11.0 cm3); and other, 9.3±14.7 cm3 (−0.1±1.0 cm3) (Table 3⇓).
On the basis of observer consensus, 36 patients (18%) had enlargement of hematoma. The cutpoint for hematoma enlargement was determined by ROC curve analyses to be V2−V1=12.5 cm3 or V2/V1=1.4 (false-positives, 7 patients; false-negatives, 2 patients; sensitivity=94.4%; specificity=95.8%). Radiographic criteria for hematoma enlargement on CT scan were therefore defined as an increase in hematoma volume by ≥12.5 cm3 or by ≥1.4 times.
Forty-one patients (20%) fulfilled the above-mentioned criteria: 10 patients had putaminal hemorrhage; 13, thalamic; 2, mixed; 9, lobar; 4, pontine; 2, cerebellar; and 1, another site. Of the 74 patients who underwent the initial CT scan within 3 hours from onset of ICH, 27 (36%) showed expansion of the hematoma. Of the 45 patients with an interval of 3 to 6 hours from onset of ICH to initial CT scan, 7 (16%) had hematoma expansion. Five (15%) of 33 patients with an interval of 6 to 12 hours displayed enlarged hematoma. Two (6%) of 34 patients with an interval of 12 to 24 hours had hematoma expansion. None of the 18 patients with an interval of 24 to 48 hours demonstrated hematoma enlargement (Fig 3⇓).
Clinical deterioration was observed in 50 patients, consisting of 27 of 41 patients with hematoma enlargement and 23 of 163 patients with unchanged hematoma size. The odds ratio was 11.7, with 95% confidence intervals of 5.0 to 27.8 (Table 4⇓). Twelve patients (29%) in the group with hematoma expansion died as a direct consequence of ICH, whereas 5 patients (3%) in the nonexpansion group died after a short time. The difference in mortality between groups was significant (P<.0001, χ2 test). Surgical treatments were performed subsequently in 14 (7%) of 204 patients. They consisted of 10 patients in the nonexpansion group and 4 in the expansion group.
A third CT scan was performed within 120 hours of ictus in 102 patients (50%), comprising 76 patients with unchanged hematoma size and 26 with hematoma expansion between the initial and second CT scans. In the 76 patients with stable hematomas (V1 to V2), there were no cases of hematoma enlargement on comparison of V2 to V3, the volume on the third CT scan. In the 26 patients with prior hematoma expansion (V1 to V2), there were 4 cases of hematoma enlargement between the second and third CT scans. The interval between the onset and the second CT scan in these 4 patients was 3, 3, 5.5, and 13 hours, respectively. Among these patients, 2 underwent a fourth CT scan within 48 hours after onset. Both of them showed no change in hematoma size. The interval between the onset and the third CT scan in these 2 patients was 22 and 30 hours, respectively.
Symptomatic aggravation in patients with ICH depends on various factors: progression of hematoma,2 3 4 5 6 7 brain edema,13 hydrocephalus, and intraventricular extension.14 15 16 Hematoma expansion was associated with a fivefold increase in the risk of clinical deterioration, and it was a significant cause of ICH-related death in the present study. However, systemic disturbances such as organ failure, infection, imbalance of water or serum electrolytes, low or high blood pressure, and malnutrition can also contribute to clinical deterioration. Even when the degrees of hematoma enlargement are the same, those of symptomatic aggravation can vary according to the site of bleeding. This heterogeneity of the factors that influence the clinical symptoms and signs may confound the analysis of the relationship between symptomatic worsening and enlargement of hematoma. In the present study, hematoma enlargement was not necessarily related to clinical deterioration. Approximately one third of patients with hematoma enlargement remained clinically stable. Conversely, 46% of patients with clinical deterioration had no change in hematoma size. Therefore, we focused on the problem of enlargement of ICH on the basis of CT findings, rather than symptoms and signs.
Previous studies concerning enlargement of ICH dealt only with patients with markedly enlarged hematomas on CT.2 3 Other investigators adopted arbitrary criteria for hematoma enlargement, such as “substantially” increased volume,4 “significant” increase in hematoma size,6 or volume increase by >50% and by >2 mL or by >20 mL,7 none of which had any statistical basis. Insofar as we know, there have been no objective, radiographic criteria for hematoma enlargement. In the present report, we propose a cutoff value for the diagnosis of increased hematoma size on CT based on ROC curve analyses: V2−V1=12.5 cm3 or V2/V1=1.4. This criterion can be used for further evaluation of hematoma enlargement.
In our study, hematoma enlargement occurred in 20% of the 204 patients presenting with ICH within 48 hours of onset. Fujii and associates7 reported that 14.3% of patients who underwent initial CT within 24 hours of onset showed an increase in hematoma size. Other investigators have estimated the incidence of hematoma enlargement to be 3%,6 or 7%5 in clinical studies in which the interval from the onset to the initial CT was not described. These differences in frequency were probably due to variability in (1) the time from ICH onset to the first CT, (2) criteria for hematoma enlargement, and (3) other aspects of study design. The true incidence of hematoma expansion may be much higher than we reported, since the patients who died of rapid expansion of the hemorrhage before the second CT scan were excluded from our study.
The time course for the progression of ICH has been controversial. Active bleeding in ICH is commonly believed to last only for a fraction of an hour.17 Herbstein and Schaumburg18 injected 51Cr-labeled erythrocytes into 11 patients with hypertensive ICH between 1 to 2 and 4 to 5 hours after onset. Postmortem examination revealed no significant activity in the primary hematoma, suggesting that the bleeding had ceased within at least 2 to 5 hours after onset. In contrast, Mizukami and coworkers19 reported 7 patients with ICH in whom cerebral angiography was performed during the time ranging from 1.5 to 7 hours after onset, demonstrating extravasation of the contrast medium from the lateral lenticulostriate artery. They concluded that bleeding from the ruptured artery ensued over several hours, at least in those patients with rapid progression to death. Broderick and associates4 evaluated 8 patients with ICH who underwent CT within 2.5 hours of symptom onset and several hours later. In 6 patients (75%), the hemorrhage volume had increased substantially between CT scans (43% or more), and the patients exhibited neurological deterioration. Fujii and coauthors7 reviewed 419 patients with ICH in whom the first CT scan was performed within 24 hours of onset and the second within 24 hours of admission. In 60 patients (14.3%), the second CT scan showed an enlarged hematoma. They found that the incidence of hematoma growth decreased with time. There have been several case descriptions in which active bleeding seemed to last for more than 6 hours after the onset of ICH.3 5 17
We presented three important facts concerning the time course of ICH. First, a majority (83%) of patients with hematoma enlargement underwent the initial CT scan within 6 hours of onset, suggesting that early CT scanning may increase the frequency for detection of expanding hematomas. Second, although the frequency of expansion is highest in the hyperacute phase, approximately one sixth (17%) of patients with increased hematomas underwent the first CT scan between 6 and 24 hours, demonstrating that hematoma progression occurred even after 6 hours of onset. Therefore, although infrequent, hematoma may expand after 6 hours of onset. Third, the most intriguing finding in the present study is the fact that no patient in the group with the first CT scan from 24 to 48 hours had hematoma enlargement. This observation held true even when considering the third and fourth CT scans. There have been several case reports of neurological deterioration 2 to 14 days after the onset of ICH with demonstrated hematoma enlargement.3 5 6 20 However, these cases are probably exceptional and outside the spectrum of active bleeding in spontaneous ICH.
It remains to be clarified whether enlargement of hematoma is due to continuous bleeding or rebleeding, or whether hematoma expansion is due to bleeding from single19 or multiple21 vessels. Accumulation of case descriptions and prospective studies of hematoma growth are needed.
Selected Abbreviations and Acronyms
|ROC||=||receiver operating characteristic|
|V1||=||hematoma volume on initial computed tomography scan|
|V2||=||hematoma volume on second computed tomography scan|
The authors thank H. Oe, MD, W. Kakuda, MD, and R. Ohtani, MD, for their evaluation of CT images.
- Received March 18, 1996.
- Revision received May 23, 1996.
- Accepted June 7, 1996.
- Copyright © 1996 by American Heart Association
Kase CS, Crowell RM. Prognosis and treatment of patients with intracerebral hemorrhage. In: Kase CS, Caplan LR, eds. Intracerebral Hemorrhage. Boston, Mass: Butterworth-Heinemann; 1994:467-489.
Kelly RE, Berger JR, Scheinberg P, Stokes N. Active bleeding in hypertensive intracerebral hemorrhage: computed tomography. Neurology. 1982;32:852-856.
Chen ST, Chen SD, Hsu CY, Hogan EL. Progression of hypertensive intracerebral hemorrhage. Neurology. 1989;39:1509-1514.
Fehr MA, Anderson DC. Incidence of progression or rebleeding in hypertensive intracerebral hemorrhage. J Stroke Cerebrovasc Dis. 1991;1:111-116.
Kwak R, Kadoya S, Suzuki T. Factors affecting the prognosis in thalamic hemorrhage. Stroke. 1983;14:493-500.
Lisk DR, Pasteur W, Rhoades H, Putnam RD, Grotta JC. Early presentation of hemispheric intracerebral hemorrhage: prediction of outcome and guidelines for treatment allocation. Neurology. 1994;44:133-139.
Broderick JP, Brott TG, Grotta JC. Intracerebral hemorrhage volume measurement. Stroke. 1994;25:1081. Letter.
Broderick JP, Brott TG, Duldner JE, Tomsick T, Huster G. Volume of intracerebral hemorrhage: a powerful and easy-to-use predictor of 30-day mortality. Stroke. 1993;24:987-993.
Mayer SA, Sacco RL, Shi T, Mohr JP. Neurological deterioration in noncomatose patients with supratentorial intracerebral hemorrhage. Neurology. 1994;44:1379-1384.
Daverat P, Castel JP, Dartigues JF, Orgogozo JM. Death and functional outcome after spontaneous intracerebral hemorrhage: a prospective study of 166 cases using multivariate analysis. Stroke. 1991;22:1-6.
Ropper AH, Gress DR. Computerized tomography and clinical features of large cerebral hemorrhages. Cerebrovasc Dis. 1991;1:38-42.
Kase CS, Mohr JP, Caplan LR. Intracerebral hemorrhage. In: Barnett HJM, Mohr JP, Stein BM, Yatsu FM, eds. Stroke. 2nd ed. New York, NY: Churchill Livingstone; 1992:561-616.
Mizukami M, Araki G, Mihara H, Tomita T, Fujinaga R. Arteriographically visualized extravasation in hypertensive intracerebral hemorrhage. Stroke. 1972;3:527-537.
Wijdicks EFM, Fulgham JR. Acute fatal deterioration in putaminal hemorrhage. Stroke. 1995;26:1953-1955.