Intracerebral Hemorrhage with Intraventricular Extension Associated with Loss of Consciousness at Symptom Onset

Spontaneous intracerebral hemorrhage (ICH) is a devastating neurological disease. One-year mortality is close to 50% [1], and the incidence of ICH has nearly tripled over the past 70 years to most recently 73 cases per 100,000 person-years [2]. New treatment targets and effective therapies are urgently needed.

Intraventricular hemorrhage extension (IVH) of ICH is independently associated with unfavorable outcome [3‐5]. Factors that are thought to mediate the worsening effect on outcome are meningitis, induced by blood breakdown products, development of hydrocephalus, and the excess intracranial volume itself leading to increased intracranial pressure [6,7]. An analysis of the INTERACT-2 study cohort showed that IVH was independently associated with both early (within 24 h) and late (up to 7 days) neurological deterioration after ICH [8]. Thus, early recognition of IVH might constitute the first step toward establishing new therapies aiming at containment or removal of intraventricular blood, which might then avert such secondary worsening.

IVH is time-sensitive and associated with hematoma expansion (HE) [4,5,9‐11], which in turn can be restricted through blood pressure control, manipulation of coagulation parameters, or surgical intervention [12,13]. In 48%–54% of cases, IVH develops in the first few hours after symptom onset and is already present on hospital admission computed tomography (CT) scan [4,5]. Thus, determination of IVH risk is more likely to impact patient triage, acute management, and potential surgical prearrangements if it happens before arrival at the receiving hospital. Pre-hospital IVH markers may become clinically relevant in the context of mobile stroke units where a CT diagnosis of a developing intracranial hemorrhage can be made before hospital arrival [14]. Alternatively—in the absence of a diagnostic CT scan before hospital arrival—pre-hospital markers may serve as part of diagnostic panels, in conjunction with acute serum biomarkers aiming to distinguish between ischemic and hemorrhagic stroke [15‐17].

In patients with spontaneous subarachnoid hemorrhage, loss of consciousness (LOC) at ictus was found to be associated with greater amounts of subarachnoid blood [18]. We speculated that intraventricular extension of ICH might be associated with LOC, too, possibly through a sudden increase in intracranial pressure from blood pouring into the ventricular space rather than being retained by brain tissue. Here, we tested whether LOC at the onset of ICH might serve as an acute clinical marker of IVH on admission or follow-up CT.

Among 3000 ICH cases enrolled in the ERICH study, we excluded 144 cases because of missing CT volumetric measurements and 134 cases because they had a seizure at symptom onset. Baseline characteristics of included patients (n = 2724) are shown in Table 1. Those with LOC (n = 352) compared to those without LOC (n = 2372) were younger (mean 60 vs. 62 years, p = 0.005) and had a greater proportion of women among them (49 vs. 40%, p = 0.04). Patients with initial LOC had a lower median GCS score upon ED arrival (median 6 vs. 15, p < 0.001), greater admission systolic blood pressure (200 vs. 184 mmHg, p < 0.001), and serum glucose (158 vs. 127 mg/dl, p < 0.001). Two hundred and seventy-one (77%) patients with LOC had IVH on either admission or follow-up CT scan compared to 978 (41%) patients without LOC (p < 0.001). Both initial ICH volume (26 vs. 10 ml, p < 0.001) and peak ICH volume (28 vs. 11 ml, p < 0.001) were significantly greater among patients with LOC compared to those without LOC. Characteristics of patients with the presence of IVH on either admission or follow-up CT scan compared to those without IVH on any CT scan are shown in Table 2. Table 3 summarizes hospital interventions, in-hospital mortality, 3-month functional (mRS), and quality of life outcomes (EQ5D) of patients stratified by the presence or absence of LOC at ICH onset. In-hospital mortality in patients with LOC was 38% compared to 7% in those without LOC (p < 0.001). 27% with LOC had unfavorable functional outcome at 3 months, compared to 23% in patients without initial LOC (p < 0.001). EQ5D at 3 months indicated significantly worse quality of life in patients with initial LOC in most categories, with the exceptions of pain, anxiety, and depression, which were similar between both groups (p = 0.09 in both categories).

Among ICH cases from the multicenter case–control study ERICH, we found that loss of consciousness at the onset of ICH was independently associated with the presence of IVH on either admission or follow-up CT. LOC was also associated with 90-day unfavorable outcome when adjusting for components of the ICH score and HE. These findings held up in sensitivity analyses excluding patients with positive drug screening on hospital admission and those whose ICH was located in the brainstem. When controlling for GCS upon ED admission, the association between LOC and outcome was partially preserved depending on which GCS cutoff was used, suggesting overlapping prognostic information between LOC and admission GCS. A mediation analysis revealed that the association between LOC and outcome was significantly mediated through IVH. Because brainstem bleeds are in close proximity to the reticular activating system and the ventricular system, we conducted a sensitivity mediation analysis in which we excluded patients with brainstem bleeds. This analysis produced similar results as the primary analysis suggesting that the mediated effect of LOC on outcome through IVH was not driven or significantly attenuated by hematoma location in the brainstem. Overall, these findings suggest that in patients with ICH, LOC at symptom onset is a marker of IVH which in turn worsens outcome.

Prior studies have established that intraventricular extension of ICH is a time-dependent process which can occur with delay after a primarily parenchymal bleed and that in 7%–10% of all ICH cases IVH extension happens after the hospital admission CT scan [4,5,9]. With regard to its effect on outcome, IVH is not an “all or nothing” event. Increasing degrees of IVH correlate with an increasing risk of worse outcome [27‐30]. These findings suggest that there might be a subgroup of patients with ICH in whom ICH is diagnosed early enough to potentially prevent IVH or contain an early stage IVH through blood pressure control or surgical intervention [12,31]. Certainly, such interventions would still need to be tested in prospective trials with regard to their efficacy. Despite the established relationship between IVH and clinical outcome, dedicated studies focusing on prediction of IVH were not available so far. Associations between greater ICH volume and hyperglycemia upon hospital arrival on the one hand and occurrence of IVH on the other have been reported. An inverse relationship between lobar hemorrhage location and IVH has been shown as well [4,5,9,32]. We reproduced the inverse relationship between lobar location of the ICH and IVH in our study. However, systolic BP and admission serum glucose were not associated with the presence of IVH in multivariate analysis (Table 4). Of note, several of the baseline characteristics of patients with LOC differed from those without LOC. Patients without LOC were older, more frequently male, more often had hyperlipidemia, more frequently were taking aspirin as home medication, and had greater PTT values on admission. Patients without IVH also had a greater median admission PTT. These differences might be driven by a greater prevalence of lobar ICH location in those without LOC (compared to those with LOC) and in those without IVH (compared to those with IVH). Patients with lobar ICH may have included a greater proportion of patients with amyloid angiopathy, a demographic that is typically older and has more comorbidities.

The novel finding our study adds is the identification of LOC as an acute clinical marker of IVH. LOC had the strongest independent association with IVH, and—being an early data point, collected before hospital arrival—LOC might be clinically more meaningful for early IVH prediction than the GCS score in the emergency department, or CT and laboratory findings which typically do not become available until hospital arrival. Our head-to-head comparison of the pre-hospital LOC variable and the dichotomized GCS in the ED shows that LOC overpowers the GCS at lower cutoff values (less than 8) that tend to align more with “loss of consciousness.” This suggests that the early level of consciousness after ICH is more meaningful for outcome prognostication than the assessment of consciousness in the ED, possibly because the pre-hospital level of consciousness is less likely confounded by sedating medications.

In our regression analysis (Table 4), we controlled for ICH volume and thus established that IVH is not simply a surrogate marker of a large parenchymal hematoma more likely to cause IVH and coma at symptom onset through mass effect [5]. However, we were not able to control for midline shift, hydrocephalus, or sudden increase in intracranial pressure (data not available), conditions that may co-occur with IVH and then lead to compression of brain structures associated with maintaining consciousness. We controlled for ICH location, but a greater anatomical resolution than large cohort studies typically provide would be needed to assess for hematoma locations as a primary cause for LOC. Studies investigating coma in patients with ICH have, among others, implicated brain structures with the presence of coma that are close to the ventricular system, for example lesions of the head of the caudate or the paramedian tegmentum [33,34]. In accordance with these findings from prior studies, patients with LOC had a greater proportion of deep and a smaller proportion of lobar hemorrhages in our study (p = 0.02). Thus, proximity of the hematoma to the ventricles may have influenced the association between impaired consciousness and IVH extension. Lastly, one may speculate that IVH may be an initial event with a cascade of downstream effects leading to LOC rather than being the direct cause of LOC. For example, IVH may cause hydrocephalus, which in turn may lead to ICP increase, which then might interfere with either cortical function or lead to different degrees and types of herniation. All of these mechanisms are well-described causes of coma [35,36].

Although our article reports associations between clinical variables, which are relevant for diagnosis rather than therapy of IVH, it implicitly raises the question what the optimal intervention is in response to IVH detection. The literature reports conflicting findings about whether removal of blood and blood breakdown products or reversal of hydrocephalus should be the primary goal of surgical intervention [37‐39]. The question is difficult to address in research studies as insertion of an external ventricular drain will typically treat both aspects of the pathology. The results of the CLEAR III trial suggest that intraventricular clot removal through insertion of an external ventricular drain in combination with alteplase irrigation does not generally improve clinical outcome in all patients with ICH. However, it may be beneficial in selected subgroups or in combination with a lumbar drain [40‐42]. More research is needed to define patient subgroups and interventions that target IVH.

The results of our study should be seen as preliminary and need confirmation in future prospective trials. If the association between LOC at ICH onset and IVH can be reproduced, its use in conjunction with other pre-hospital tests might allow efficient triage. Possibly, this might lead to earlier EVD insertion in those who need it.

Our study has limitations. First, although the study data were collected prospectively, our analysis was retrospective, which has conceptual limitations. We cannot exclude a degree of inaccuracy regarding assessment and documentation of the initial symptom (loss of consciousness); however, data in ERICH were collected carefully by dedicated study investigators. Second, the etiology and dynamics of the initial LOC were not addressed in this study. Importantly, data on the presence of hydrocephalus on CT were not available. Hyperacute onset hydrocephalus may contribute to the association between LOC and IVH, and it certainly contributes to the association between IVH and clinical outcome [37,39]. Many different trajectories after initial LOC are possible, ranging from recovery to progression to brain death. Future studies will hopefully clarify these different trajectories as well as how initial LOC relates to early and delayed neurological deterioration. Third, we did not differentiate between patients with initial and delayed IVH in our analysis. The distinction between the two depends on the time interval between symptom onset and imaging [5], a variable that was not available with hour/minute resolution for all cases in ERICH. To account for that, we decided to analyze patients with initial and delayed IVH together, and to use, wherever available, peak ICH volume instead of ICH volume on admission CT. Peak ICH volume can be assumed to be a more robust measure in the context of variable symptom-to-imaging time intervals. Of note, around 30% of ERICH cases did not have follow-up CT imaging. Thus, in these cases only one ICH volume measurement is available which precludes assessment of HE. Dedicated studies would be needed to assess whether LOC or decline of mental status after hospital arrival is associated with delayed IVH.

Fourth, there are limitations inherent to mediation analysis as a statistical technique; mediation analysis may help to identify statistical associations, but it cannot prove causality between mediator and dependent variable. Fifth, this is a clinical study and correlations between clinical and imaging findings cannot be made with high accuracy. Small prospective studies with high-resolution imaging technology might be more appropriate study designs for this purpose. Lastly, the ERICH study includes only people of White or Black race (both non-Hispanic) or Hispanic ethnicity which may limit the generalizability to other populations.

In this analysis of ICH cases from the prospective multicenter ERICH cohort, pre-hospital LOC was strongly associated with the presence of IVH on admission or follow-up CT brain scan. This information may become clinically useful for patient triage and risk stratification in future trials. LOC was also associated with long-term functional outcome after ICH, but a significant portion of this effect was mediated through IVH. The findings of this report emphasize the importance of IVH as a treatment target in patients with ICH. Future studies are needed to characterize how LOC can be integrated into clinical tools that predict IVH, and how IVH and LOC may connect mechanistically.