Hematoma growth is a determinant of mortality and poor outcome after intracerebral hemorrhage

Intracerebral hemorrhage (ICH) is associated with a substantially worse prognosis than ischemic stroke, with a mortality rate approaching 50% and little effective treatment.¹ Early surgery for supratentorial ICH was not beneficial in a large recent trial of over 1,000 patients.² In contrast, improved outcomes using recombinant activated Factor VII (rFVIIa) indicate the potential of a medical therapy for patients treated within the first 4 hours of the onset of ICH.³

The rationale for rFVIIa was based on new insights into the pathophysiology of ICH. First, the volume of ICH, together with conscious state, were both recognized as critical prognostic determinants.⁴ Second, ICH is not a static phenomenon. It was erroneously believed that the volume of a cerebral hematoma was usually maximal at onset, then contained by hydrostatic forces and elevated intracranial pressure.⁵ This contrasted with understanding of the evolving pathophysiology in ischemic stroke, where the infarct core typically expands due to progressive failure of perfusion in the ischemic penumbra. This insight led to introduction of thrombolytic therapy.⁶

Although studies in ICH have not confirmed an ischemic penumbra,⁷ a significant proportion of patients demonstrate hematoma expansion. This was first shown in retrospective studies,^8,9 and then confirmed prospectively in the Cincinnati study, which showed that 38% of cases exhibited significant growth, defined as >33% volume increase, over 24 hours.¹⁰ This finding demonstrated that, like ischemic stroke, ICH is a dynamic process where the treatment goal is limitation of stroke volume.¹¹

Established prognostic factors in ICH include baseline hematoma volume, age, conscious state, infratentorial location, and intraventricular hemorrhage (IVH),¹² but previous reports have studied patients at heterogeneous times from onset, most have been retrospective, and it is unknown whether subsequent hematoma growth is an independent predictor of mortality and poor functional outcomes.

The imaging paradigm in the Cincinnati study was the basis for the three studies testing safety, dosage, and efficacy using rFVIIa. All involved CT within 3 hours of onset of ICH, with repeat imaging approximately 24 hours later to measure expansion.^3,13 By pooling these patients, we tested the hypothesis that hematoma growth is associated with increased mortality and poor functional outcomes, independent of baseline ICH volume and other established clinical factors. If hematoma growth is an independent determinant, it should be used in clinical decision-making and evaluated in studies of acute interventions, such as blood pressure reduction and other pharmacologic therapies.

Methods.

Results.

Discussion.

This pooled individual patient meta-analysis, involving 218 acute patients studied prospectively with a similar CT imaging protocol and outcome measures, has demonstrated that hematoma growth is a crucial, independent determinant of both mortality and functional outcome after ICH. Hematoma growth is very common in acute ICH, any degree of growth being seen in 73% of patients in this pooled analysis. For each 10% increase in hematoma growth, there was a 5% increased hazard of death, a 16% greater likelihood of worsening by 1 point on the mRS, or 18% of moving from independence to assisted independence or from assisted independence to poor outcome on the BI. The study has confirmed that other critical determinants of outcome include initial volume, IVH, and clinical factors (age, baseline conscious state). Other factors identified in previous studies, such as initial blood glucose, blood pressure, and infratentorial hemorrhage, were not significant.

Early hematoma growth was initially observed in case series of ICH, without coagulopathy, and correlated with clinical deterioration.^17,18 Growth and larger ICH volumes predicted worsening, rather than clinical indicators.¹⁹ Larger retrospective Japanese studies^8,20 indicated that a minority of patients with ICH exhibited hematoma growth, which predicted poor outcome. However, these studies were limited by considerable variations in imaging protocols and definitions of growth. In one study⁸ hematoma expansion occurred in 14% of cases studied by CT within 24 hours of onset, repeated 24 hours later. While this retrospective study suggested that growth predicted increased mortality, baseline ICH volume was not included. In contrast, another group⁹ reported hematoma growth in only 2% of ICH patients.

The Cincinnati study was the only prospective study of hematoma growth, prior to the rFVIIa studies.¹⁰ Significant growth was seen in 38% of patients, with two thirds evident within 1 hour after the first scan. Growth was associated with early neurologic deterioration, but a trend to higher mortality and poor functional outcomes was not significant. Hemorrhage growth probably reflects continued bleeding or rebleeding. Extravasation of angiographic contrast medium has been demonstrated in hypertensive ICH^21,22 and correlated with hematoma expansion²³ and poor outcome.²⁴

In the Japanese studies, predictors of growth included earlier time from onset, irregularly shaped hematoma, liver dysfunction, alcohol consumption, coagulation abnormalities, and depressed conscious state.^8,20 In contrast, the Cincinnati study found that baseline ICH volume, GCS, time from onset to baseline CT, blood pressure, age, vascular risk factors, and coagulation factors were not predictors of growth.¹⁰ Hematoma growth has been associated with elevated blood pressure,^25,26 suggesting that blood pressure lowering might attenuate this phenomenon. Warfarin therapy has also been associated with hematoma growth.²⁷ Recently, it was reported that molecular signatures of vascular injury and inflammatory markers in acute ICH (IL-6, TNF-α, MMP-9, and c-Fn) are associated with hematoma enlargement and poor outcomes.²⁸

We found that for each 1 mL increase in baseline ICH volume, the hazard ratio of dying increased by 1% and survivors were 6% more likely to worsen by 1 point on the outcome mRS. A number of studies have shown the importance of baseline ICH volume in the prediction of mortality and functional outcome.^12,29,30 A baseline hematoma volume of more than 30 mL is highly predictive of a poor functional outcome.⁴ For patients with an ICH volume of 60 cm³ or greater on the initial CT scan and a GCS of 8 or less, there is a 30-day mortality of 91%.⁴ Various ICH scores have been developed to facilitate more precise prognosis. One of these incorporated GCS, age (≥80 years), infratentorial location, ICH volume, and presence of IVH.¹² These clinical scores are useful in assessing ICH prognosis at baseline, but our study was designed to assess whether early hematoma growth further worsened outcome.

We found that age, baseline conscious state, and IVH were also independent contributors to prognosis, taking into account initial ICH volume and subsequent hematoma growth. Infratentorial location of ICH has been associated with a worse prognosis, particularly brainstem hemorrhage.^12,31 Some studies have indicated that patients with lobar ICH have a better outcome.^32,33 Although we did not find a statistically significant relationship between ICH location and outcome, our analysis was limited by the very small proportion of patients with infratentorial hemorrhage (7.3%). An increased mortality rate in these patients did not achieve statistical significance, likely because of the small sample size (see table 4). IVH is a well-recognized adverse prognostic factor.^12,34 This study showed that IVH doubled the risk of death.

Other clinical prognostic factors in various studies have included hyperglycemia^35,36 and gender.^8,37 Hyperglycemia is an important prognostic determinant in ischemic stroke, where it is associated with increased lesion lactate, impaired salvage of the ischemic penumbra, and worse outcomes.^38,39 There have been fewer studies in ICH, but recent reports have concluded that there is a similar adverse relationship.^35,36 However, unlike ischemic stroke, hyperglycemia in ICH is probably not independent of stroke severity and may reflect a stress response.³⁶ Therefore, by correcting for ICH volume and the GCS, hyperglycemia may not have been significant in our study. Previous studies have reached different conclusions concerning gender and outcome.^33,37 Modeling percentage growth, gender was not predictive of mortality or functional outcome after ICH. However, modeling absolute growth, female gender was associated with reduced hazard of death. Baseline hypertension has been linked to worse outcome after ICH^40,41 and also as an independent contributor to hematoma growth.^20,26 We did not show an independent effect of initial blood pressure on outcomes.

Despite some methodologic differences between the Cincinnati and rFVIIa studies, we considered that the pooled meta-analysis was appropriate to assess the influence of hematoma growth in the context of baseline prognostic factors. Baseline ages, ICH volumes, and proportions with IVH and glucose levels were comparable in the two larger studies. A sub-3 hour CT scan with repeat imaging approximately 24 hours later was the common protocol. Absolute and percentage hematoma growth were similar in the Cincinnati study and the rFVIIa trials.

The major methodologic difference between the studies was the timing of the clinical endpoints. Mortality, mRS, and the BI were calculated at 4 to 6 weeks in the Cincinnati study and at 90 days in the rFVIIa studies. Interestingly, mortality was higher in the Cincinnati study (39.8%) than the rFVIIa studies (26.1%) despite the earlier timing of the endpoints, yet the studies had similar baseline demographic data, ICH volumes, and growth. However, patients in the Cincinnati study had a worse conscious state at baseline (GCS 11.2 compared with 13.2 in the rFVIIa trials) due to exclusion of patients in deep coma from the rFVIIa trials. This higher mortality may have also reflected the difference in comorbidity factors between the community-based cohort and the highly screened trial patients (which also excluded those with previous myocardial infarction and ischemic stroke). The trial type (Cincinnati vs rFVIIa) was considered when constructing the proportional odds model and was not a significant predictor. Furthermore, subanalyses of each trial yielded similar results, supporting the validity of our approach in pooling the individual patient data.

There is renewed interest in acute therapy for ICH with the evidence that rFVIIa attenuated hematoma growth and that this biologic effect was associated with reduced mortality and enhanced functional outcomes in survivors.³ Other interventions currently under investigation include blood pressure reduction and neuroprotective compounds.¹⁶ These acute ICH trials will need to take into account the effects of any intervention on hematoma growth, as well as controlling for the other critical determinants of outcome. Finally, this study has emphasized the importance of hematoma growth as a therapeutic target in both clinical practice and research.

Note added in proof.

Mayer et al.⁴² and Zimmerman et al.⁴³ were accepted after this article was accepted for publication.

Acknowledgment

The authors thank Brett Skolnick, PhD, for valuable advice.