Abstract
Background: Patients with acute intracerebral hemorrhage (ICH) presenting within 3 hours of symptom onset are known to be at increased risk of expansion. However, only a minority arrive within this time frame. Therefore, alternative markers for expansion risk are needed.
Objective: To examine whether contrast extravasation on CT angiography (CTA) at presentation predicts subsequent hematoma expansion.
Methods: Consecutive patients with primary ICH presenting to an urban tertiary care hospital were prospectively captured in a database. We retrospectively reviewed images for all patients receiving a CTA and at least one further CT scan within 48 hours.
Results: Complete data were available for 104 patients. Contrast extravasation at the time of CTA was present in 56% of patients, and associated with an increased risk of hematoma expansion (22% vs 2%, p = 0.003). Patients who received a baseline CTA within 3 hours were more likely to have subsequent expansion (27%, vs 13% for those presenting later, p = 0.1). However, after multivariable analysis, contrast extravasation was the only significant predictor of hematoma expansion (OR 18, 95% CI 2.1 to 162). This effect was independent of time to presentation.
Conclusions: Contrast extravasation is independently associated with hematoma expansion. Patients presenting within the first few hours after symptom onset have traditionally been considered those at highest risk of expansion. However, for those presenting later, the presence of contrast may be a useful marker to guide therapies aimed at decreasing this risk.
In order to effectively target clinical treatment of acute intracerebral hemorrhage (ICH), it is important to predict which patients will develop hematoma expansion.1–8 Known risk factors for expansion include both antecedent warfarin use9 and ultra-early presentation.1–3,8 Unfortunately, most patients present in a delayed fashion,10 and there are no established markers for identifying patients with ongoing bleeding. Therefore, there is a need for a time-independent marker of those patients at risk for hematoma expansion.
One candidate for such a marker is the extravasation of radiographic contrast. Case reports have suggested that contrast extravasation during cerebral angiography is associated with continued bleeding.11–13 However, angiography is limited to selected centers, and can be invasive and time consuming. While the presence of intrahematomal gadolinium during MRI is associated with hematoma expansion,14 MRI is often not available in the acute setting. The use of CT angiography (CTA) provides 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, and is associated with increased mortality.15–17
We hypothesized that contrast extravasation within the hematoma on CTA represents ongoing bleeding and therefore identifies those patients at high risk for subsequent hematoma expansion. If true, this finding could offer a method for selecting patients most likely to benefit from interventions aimed at arresting ongoing bleeding.8
Methods.
Patient selection and data collection.
We reviewed data collected as part of an ongoing prospective cohort study of ICH outcome.18,19 Since 1994, consecutive patients with ICH who present to Massachusetts General Hospital are registered in a database. Patients are aged ≥18 years, and excluded if ICH is secondary to head trauma, ischemic stroke with hemorrhagic transformation, tumor, vascular malformation, or vasculitis. All aspects of the study have been approved by the Institutional Review Board.
Study protocol.
We reviewed CT scans and CTAs for all patients with ICH from January 1, 2002, to March 2006. Patients were included if they received a CTA (performed at the discretion of the treating physicians) and if they received at least one more CT scan within 48 hours after their CTA.
ICH volumes were determined from baseline CT scans as previously described.9 Images were electronically transferred in DICOM format to a dedicated workstation for analysis using Alice software (Parexel Corporation, Waltham, MA) and reviewed by study staff blinded to the patient's clinical status and other radiographic studies according to a method with high inter-rater reliability.9 Hematoma location was categorized as lobar vs nonlobar on the basis of the CT scan by one of the study neurologists based on previously established criteria.19 The presence of intraventricular hemorrhage (IVH) was scored as a dichotomous variable.
All subjects or their family informants were prospectively interviewed for clinical data including history of diabetes (DM), coronary artery disease (CAD), and medications. Laboratory studies including serum glucose, platelet count, and international normalized ratio of the prothrombin time (INR) were taken as the first levels drawn; for patients transferred from outside facilities, the outside facility value was captured. When these data were not available, the laboratory value on arrival at our hospital was substituted. The first documented systolic blood pressure (SBP) and diastolic blood pressure (DBP) were captured as admission SBP and DBP. Admission Glasgow Coma Scale (GCS) score was captured from the emergency department (ED) note or admission note. Time of symptom onset was defined as the last time the patient was known to be symptom free, and was captured retrospectively by review of each patient's admission note and discharge summary.
Of 494 patients with ICH presenting during the study period, a total of 294 patients had at least two CT scans performed and available for review. Of these, 108 did not receive a corresponding CTA and 28 did not have a subsequent CT scan. Nineteen patients underwent surgical procedures and were excluded. For 28 patients, either the CTA or one of the CT scans was unavailable for viewing and in 4 patients, the second CT was performed more than 48 hours after the index CT scan. Finally, one patient received gadolinium instead of nonionic contrast, and two patients were enrolled in a clinical trial of activated factor VII, leaving 104 patients for analysis.
CT angiography.
First, nonenhanced head CT was performed using a helical CT scanner (High-Speed Advantage; GE Medical Systems, Milwaukee, WI). This was followed by helical scanning during the administration of 90 to 120 mL of nonionic contrast agent at 3 to 5 mL/second with a 25 to 40 second prep delay using standard scan parameters of 120 to 140 kVp and 170 to 220 mAs.20 Delayed images are not routinely obtained in our institution and were not analyzed. Section thickness was 5 mm for nonenhanced scans and 0.6 mm for most CTAs (except for 1.25 mm for 4 patients and 2.5 mm for 6 patients; there was no apparent effect on detection of extravasation). CT scans were reviewed independently by two reviewers blinded to all clinical data. Interrater agreement was 90% (kappa 0.80), and all differences were resolved by consensus. The presence of contrast extravasation was operationally defined as the presence of high-density material within the hematoma (figures 1 and 2).
Figure 1. CT scan of an 83-year-old woman with acute onset of left sided weakness. (A–D) Nonenhanced CT (contiguous 5 mm slices shown) with a right fronto-parietal hemorrhage. (E–L) CT angiogram (contiguous 0.625 mm slices shown) with normal cerebral vascular enhancement and a punctuate focus of intrahematomal contrast (arrow). (M–P) Nonenhanced CT (contiguous 5 mm slices shown) performed 1 hour later with evidence of hematoma expansion.
Figure 2. Contrast within the hematoma, historically interpreted and operationally defined here as extravasation. (A) Contrast pooled within the hematoma. (B) Contrast speckling, with areas of contrast distributed throughout the hematoma. (C) Contrast blush, with two regions of contrast in the superior portion of the hematoma.
Outcome measures.
The primary outcome was hematoma expansion within 48 hours of CTA, as defined by an increase in volume of >33% from baseline.2,9 ICH volume was compared between the CT performed at the time of CTA and the last CT performed up to 48 hours after CTA. Time to repeat CT was measured as the time from CTA to latest CT within 48 hours. In-hospital mortality was measured as a secondary outcome.17
Data analysis.
Statistical tests used are listed in the tables. Dichotomous variables were compared between groups using Fisher exact test for significance. The Wilcoxon rank sum test was used for continuous and ordinal variables. Multivariable analysis was performed with a logistic regression model. Based on the sample size of 104 patients, our study had 80% power to detect a 20% absolute difference in hematoma expansion rate at the alpha = 0.05 level. All analyses were performed with Stata software (Stata Corp, College Station, TX).
Results.
Our primary hypothesis was that contrast extravasation, operationally defined as the presence of intrahematomal contrast on CTA, is an independent predictor of hematoma expansion in patients with primary ICH. During the study period 104 (35%) of 294 eligible patients had a CTA available for review and were included. Patients who underwent CTA did not significantly differ in time to presentation, demographic features, radiographic features, or outcome compared to those who did not. Localization of contrast within the hematoma was central in 38% of patients, peripheral in 38%, and mixed (both central and peripheral) in 13% of patients. Overall, hematoma expansion greater than 33% from baseline occurred in 13% of patients, and in those with expansion, hematoma volume increased by median 12 mL (IQR 4 to 21), and the median percent change in volume was 51% (33 to 102%).
We found that by univariate analysis, contrast extravasation on CTA was the most powerful predictor of subsequent hematoma expansion (table 1). Contrast extravasation was present in 92% of patients who developed subsequent hematoma expansion, compared with 51% of those who did not (p = 0.006). The sensitivity and specificity of extravasation for predicting hematoma expansion were 93% and 50%, yielding a low positive predictive value (24%) but a striking negative predictive value (98%). There was no effect of patient transfer status, alcohol use, or tobacco use (not shown). There was a trend toward earlier time to CTA in patients with hematoma expansion (27% of those presenting within 3 hours vs 13% of those presenting after 3 hours, p = 0.1, figure 3). Patients with more severe disease tended to present earlier; those with a GCS score less than 8 presented within a median of 8 (3 to 17) hours, while those with a GCS score of at least 8 presented within a median of 13 (5 to 28) hours (p = 0.007 by Spearman correlation). However, there was no significant effect of initial disease severity on risk of subsequent expansion (table 1).
Table 1 Predictors of hematoma expansion
Figure 3. Histogram showing the number of patients with hematoma expansion as a function of time from symptom onset to CT angiography.
We then analyzed whether any clinical factor was associated with contrast extravasation. Overall, extravasation was present in 58 patients (table 2). Extravasation was associated with larger hematoma size; it is not clear whether this reflects increased ability to detect this finding. Unlike in previous reports,17 we found no effect of time to CTA on presence of contrast extravasation (p = 0.5; figure 4). We did confirm, however, an increased rate of in-hospital mortality in those with extravasation (33% vs 15%, p = 0.04).
Table 2 Clinical factors associated with contrast extravasation
Figure 4. Histogram showing the number of patients with contrast extravasation as a function of time from symptom onset to CT angiography.
Multivariable analysis demonstrated an independent effect of contrast extravasation on hematoma expansion (table 3). Variables were included in the model for p < 0.1 in univariate analyses (age, time to CTA, admission SBP, and hematoma volume). The presence of extravasation increased the odds of hematoma expansion by 18-fold (95% CI, 2.1 to 162).
Table 3 Multivariable analysis of predictors of hematoma expansion
Discussion.
We found an independent association of contrast extravasation with hematoma expansion. Approximately half of all patients with primary ICH who received a CTA showed evidence of contrast extravasation, and this finding was associated with an 18-fold increase in the odds of hematoma expansion. While time to presentation has previously been used to guide therapy to those at highest risk of expansion,8 our findings suggest that the presence of extravasation may be a more direct marker of ongoing bleeding.
Other studies have shown a number of factors to be associated with contrast extravasation on CTA. We noted that patients with increased hematoma size at baseline17 were more likely to demonstrate this finding. It may be that extravasation is more easily detected in larger hematomas; alternatively, large hematomas may mark those patients with more severe underlying vasculopathy or coagulopathy. While it might be expected that warfarin use would increase the likelihood of detecting contrast extravasation, neither we nor others14,17 could demonstrate this. While we confirmed a previously noted trend toward increased systolic blood pressure in those with extravasation,17 this effect was not confirmed in multivariable analysis. Other studies have failed to demonstrate an effect of initial BP as well.14,16 We also did not appreciate previously noted effects of level of consciousness or time to presentation17 on extravasation. Some of these differences may be methodologic. For example, one group17 categorized contrast extravasation into three categories: none, minimal, or extensive. In our analysis, we found poor interrater reliability between the minimal and extensive categories; therefore, we analyzed extravasation as a binary variable. It is worth noting that we saw few patients with extensive extravasation, and this was likely due to the fact that delayed images (which are rarely obtained in our center) are required to appreciate this finding. As we noted a similar overall frequency of extravasation,17 and a higher frequency than did an MRI-based protocol,14 we do not believe that the initial source images are meaningfully less sensitive for detection of extravasation. It may be, however, that if extravasation becomes used in routine clinical practice, delayed images will have value in presenting more clear and less subtle images for the emergency practitioner. Further studies involving delayed images should help determine whether they can further clarify those at risk for expansion.
Our study is limited by its retrospective design. A number of patients with primary ICH did not receive a CTA, and the decision to perform this study was made by the treating clinicians, rather than a standard protocol. Patient selection for CTA may therefore limit the generalizability of these results to other ICH populations. Some patients did not receive a second and subsequent CT scan; whether this was due to early death, DNR status, or lack of clinical deterioration is not clear. In addition, a relatively small number of patients achieved the primary endpoint, limiting our statistical power. Finally, the term contrast extravasation has been operationally defined by our group and others14–17 as the presence of high-density material within the hematoma that is not seen without contrast. It is important to note, however, that in the absence of histologic examination, or angiography with continuous imaging, it is difficult to establish what this material represents. The presence of contrast may reflect breakdown of the blood–brain barrier, ongoing bleeding from a single vascular source, delayed bleeding from a secondary source, or even pooling of contrast within a collapsed vein that has been trapped within the hematoma. While this finding is operationally valuable as a radiographic marker of expansion, further studies are necessary to determine the underlying pathophysiology.
From a clinical perspective, a radiographic finding that predicts hematoma expansion regardless of time to presentation has important implications for treatment stratification. Those presenting early are already known to be at high risk for expansion.2 The major strength of our cohort, however, lies in the fact that the majority of our patients presented later than 3 hours, allowing us to investigate a marker of expansion in these delayed presenters. We propose that those presenting later, but with evidence of intrahematomal contrast, can potentially be targeted for acute hemostatic therapy. For example, anticoagulation reversal is considered critical, but there is wide variation in choice of agents and timing of reversal21–24; patients with evidence of extravasation may be particularly strong candidates for more expensive, rapidly acting agents.22,25–27 In addition, a recent trial suggested that acute delivery of a hemostatic agent to patients with no known coagulopathy could decrease the incidence of hematoma expansion and improve outcome.8 However, this agent was only provided to those presenting within 3 hours of symptom onset, and an ongoing phase III trial includes a similar time cutoff. Clinical trials in ischemic stroke have begun using specific findings on neuroimaging to select patients most likely to benefit from thrombolytic therapy,28,29 and future clinical trials in ICH targeting hemostatic therapy on the basis of contrast findings may demonstrate efficacy even for those with a delayed presentation.
Footnotes
-
Commentary, see page 883
Supported by The National Institute of Neurological Disorders and Stroke (NIH 1 K23 NS42695-01, R01 NS04217), and the Jerome Lyle Rappaport Charitable Foundation.
Disclosure: Dr. Rosand has received research support, consulting, and speaking fees from NovoNordisk A/S. Dr. Goldstein has received consulting and speaking fees from NovoNordisk A/S. Dr. Lev has received educational support, consulting, and speaking fees from GE Healthcare, consulting and speaking fees from Bracco Diagnostics, and consulting fees from Coaxia. No compensation or fees were in excess of $10,000.
Received August 2, 2006. Accepted in final form November 21, 2006.
References
Letters: Rapid online correspondence
- Contrast extravasation on CT angiography predicts hematoma expansion in intracerebral hemorrhage
- Reply from the authors
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan