Introduction
After intracerebral hemorrhage (ICH), the mass effect from the hematoma and edema, among other factors, can raise intracranial pressure (ICP) [
1]. High ICP predicts poor outcome and death [
2,
3] and is linked to impaired consciousness [
4]. However, ICP and related compensatory changes in response to large hemorrhagic mass lesions have not been well studied in animal models.
The Monro-Kellie hypothesis predicts that mobile cranial fluids, specifically blood and cerebrospinal fluid (CSF), are displaced to maintain ICP in response to a change in the volume of the cranial contents (e.g., from hemorrhage) [
5,
6]. In contrast to the mutability of blood and CSF volumes, brain parenchymal volume, consisting of interstitial and intracellular fluids, is thought to be relatively unchangeable [
5]. The exhaustion of rapid compensatory mechanisms (displacement of blood and/or CSF) due to a large insult causes ICP to increase, resulting in brain displacement or herniation and impaired cerebral blood flow [
6‐
8].
We previously reported significantly (>20 mmHg) and persistently (at least 3 days) raised ICP following moderate and large collagenase-induced ICH in rats, with modest change after a lesion-size-matched whole blood-induced hemorrhage [
9]. Here, we examined changes in ICP following striatal infusion of 100, 130, and 160 μL of blood. Most studies infuse 100 μL of blood or less in rats, which is comparable to a large ICH in patients, relative to brain size [
9,
10]. We predicted that blood infusion would cause volume-dependent ICP elevations. In the collagenase model, ICP and edema peak 3 days after ICH [
9]. Thus, we measured ICP for 72 h after ICH and then assessed brain water content. Further, most preclinical research measures brain water content in tissue containing the hematoma, but this may not reflect edema in surrounding tissue. Accordingly, we assessed whether increased brain water content following blood infusion was due to true peri-hematoma edema or to water contained within the hematoma. In addition, since temperature can impact ICP elevations [
11] and large insults may impair temperature regulation, we measured whether it changed after large ICH. Also, we predicted that large hemorrhages would cause global ischemia as a result of high ICP [
6]. We assessed this by histological examination of cells in the hippocampal CA1 sector and primary somatosensory cortex.
Discussion
Raised ICP is a potentially deadly consequence of ICH. A thorough understanding of pressure rises and compensatory mechanisms is vital to developing novel treatments. However, studying these concepts requires appropriate animal models. Here, we show that large hemorrhages induced by striatal blood infusion are readily accommodated for, causing ICP to increase only modestly and transiently. Thus, at least with regard to ICP, this particular model does not accurately mimic severe ICH in patients. Interestingly, we found that very large mass lesions in both common rodent models of ICH caused cell density to increase and cell size to decrease in the hippocampus and cortex, and likely in other brain regions. These initial findings may challenge the long-held notion that the brain is largely incompressible because large mass lesions appear to compress parenchymal volume.
The transient nature of the ICP rises reflects compensatory changes in the volumes of cranial contents to preserve normal pressure [
5,
6]. The conventional view of cranial pressure-volume relationships describes three cranial volumes: the blood, CSF, and brain. However, despite occupying >80% of the cranial space, brain parenchyma has not been considered a mutable cranial volume, in contrast to the other rapidly displaced cranial fluids [
25]. Our findings that cell-packing density is increased bilaterally in two disparate brain regions after large ICH suggest that interstitial fluid is displaced or neuropil volume is reduced. Similarly, decreased cellular size indicates a decrease in intracellular fluid. Thus, large mass lesions can seemingly displace parenchymal volume, an unexpected finding of potential therapeutic interest.
We show evidence of decreased brain volume in two ICH models. The whole blood model produces a much larger hematoma than a lesion size-matched collagenase insult; however, the collagenase model causes much more edema [
11,
15]. Therefore, the mass effect of each is significant. Notably, effects were bilateral, consistent with our previous findings that there are no easily detectable pressure gradients after rodent ICH [
9]. We also note that decreased cell size and increased density were observed only after the most severe hemorrhages, indicating that easily displaced fluids are moved first. This implies that parenchymal volume can be diminished outside the hemorrhagic zone. Furthermore, this decreased volume is not attributable to cell death or sub-lethal injury because we found no histological evidence of global ischemia (e.g., cellular debris and greater microglia numbers [
26]) and we assessed regions outside of the pathologically active peri-hematoma zone. Moreover, these changes are not likely due to the inflammatory response, which tends to be localized to the hemorrhage site and is resolving by day 7 [
27]. As well, infiltration of inflammatory cells would presumably increase parenchymal volume, which is the opposite of what we observed in the cortex and hippocampus. We assessed changes in cell size and density 7 days after ICH, which means the observed changes in cell size and density are persistent. However, it is unclear when this shrinkage begins, and it may have occurred after smaller insults but resolved before 7 days. We speculate that mechanical force due to the mass lesion (i.e., the mass effect) drives an efflux of interstitial fluid from the brain along paravascular routes [
28], though this should be tested directly. It remains unclear if cellular shrinkage is an active (regulated [
29]) or passive (mechanical [
28]) process.
Direct assessment of the implications of decreased cell size and increased density was beyond the scope of this study. Nonetheless, it should be noted that stroke impairs metabolism and function of surviving tissue [
30] and high ICP depresses consciousness [
4]. Further, changes in interstitial volume are associated with large-scale state-dependent activity changes [
29], and changes in cellular volume can alter excitability [
31]. Notably, seizures occur after collagenase-induced ICH, but not after a 100-μL blood infusion [
16]. Seizure activity may occur in very large blood-induced hemorrhages that cause changes in cell size and density. Taken together, these findings suggest that changes in intracellular and interstitial volume may contribute to altered neural activity after stroke, though this should be tested directly.
We previously reported ICP elevations above 20 mmHg following collagenase-induced ICH [
9,
11]. Here, we found severity-dependent, but modest (<20 mmHg), increases in ICP after large ICH in the whole blood model. Further, the temporal pattern of the pressure response differs between models. The collagenase model raises ICP for at least 3 days after hemorrhage, whereas blood infusion causes immediate, but short-lived, increases. It is possible that ICP increases over time in some models due to the progression and expansion of the lesion and edema, which occurs to a much greater extent in the collagenase model [
15]. Notably, delayed ICP peaks are also observed in models of focal ischemia [
13,
32].
Edema is a common endpoint in preclinical ICH research [
33]. While edema is thought to be a major contributor to raised ICP [
1], recent studies show a complex relationship [
11,
32,
34]. Here, we found that much of the increase in brain water content after ICH was due to water contained within the hematoma. This suggests that there is relatively little peri-hematoma edema in the whole blood model (vs. collagenase model), even with massive hematoma size. However, it is possible that the 160-μL infusion caused more peri-hematoma edema than the smaller infusions. In addition, we found no significant relationship between brain water content and peak ICP in ICH animals. Taken together, these findings suggest a limited role of edema in ICP rises in this model; the hematoma itself is the primary source of the mass effect in some cases. This hypothesis is consistent with clinical findings that peri-hematoma edema appears to be significantly derived from hematoma-constrained water [
35]. Furthermore, the relationship between edema and ICP is highlighted by treatment effects. We [
11] and others [
32,
34] have reduced post-stroke ICP elevations using hypothermia. In these studies, ICP was lowered without affecting edema. While these findings conflict with evidence that hypothermia reduces edema [
36], they support the notion that edema is not alone in causing ICP elevations after brain injury. While edema may reflect outcome, it may not predict ICP changes as previously thought and appears to be commonly overestimated due to including hematoma in samples. Of course, edema measurement with the wet-dry weight method is dependent on the tissue sample size (e.g., extent of normal and abnormal tissue), making cross-study and cross-technique comparisons challenging [
37]. Thus, the use of ICP measurement, although technically challenging in rodents, is recommended when done with an established and consistent method. Additional measures, such as tissue oxygenation, may better predict outcome when used alone or in conjunction with ICP recordings [
38].
Preclinical investigation of ICP has produced considerable inter-study variability in the absolute magnitude of ICP measurements, even in uninjured animals. For example, one study reported resting ICP in rats from 17 studies as ranging from 0 to 47 mmHg [
39]. Most reports indicate that normal rat ICP is 4 to 8 mmHg, compared to ~10 mmHg in humans [
6,
39]. In the present study, we observed ICP in uninjured animals to be 3.4 ± 0.5 mmHg, which is slightly lower than our past studies likely due to normal variation within the population (i.e., sampling chance) or some unknown technical issue. We previously found ICP in awake control animals to be ~4 to 5 mmHg using this technique [
9,
11,
13] and with non-telemetric, fluid-filled transducers in anesthetized animals (unpublished data). Nonetheless, the lack of consistent measures between groups and studies is concerning. Differences in technique and location may underlie the variability, though the most common recording sites (intraventricular, intraparenchymal, and epidural) seem to produce the same measurements within studies [
9,
39,
40], but measurement from the cisterna magna may underestimate ICP [
40]. We chose epidural measurement due to the ease of access and lack of tissue damage compared to other common sites. Greater inter-study consistency in methodology would ease interpretation and comparison of findings.
There are several limitations with our methodology. First, there is a delay between ICH and the beginning of ICP recordings. Since we tend to observe the highest ICP soon after ICH in the whole blood model, we may have missed important early spikes. However, it takes time for ICP to normalize following opening of the cranium for blood infusion. Nonetheless, ICP does not increase enough to significantly reduce cerebral blood flow because there was no evidence of widespread global ischemia (i.e., no CA1 sector cell death, no spikes in ICP to levels which would significantly impair cerebral blood flow), though we did not assess whether localized peri-hematoma hypoperfusion or hypoxia occurred [
6]. Second, we were not able to correlate behavioral impairment and lesion volume with ICP because these outcomes were measured in separate animals. However, we observed infusion-volume-dependent changes in mean ICP, lesion volume, and impairment between experiments. Future studies should directly assess the relationship between ICP, related mechanisms (e.g., tissue oxygenation), and histological and behavioral outcomes in preclinical models. Also, it is unlikely that fixation artifacts confound our findings on cell size and density because all tissues were treated identically. Nissl stains tend to underestimate cell volume due to restriction to the Nissl body and tissue shrinkage, but underestimates are linear [
41,
42]. Thus, our comparisons between identically processed samples are seemingly valid, but we acknowledge that our measurements of soma size may overrepresent larger cells [
20]. Lastly, we did not assess total parenchymal volume changes because that was not our initial aim, and changes were assessed at only one time post-ICH. Future studies could chronically assess brain volume using MRI, but mass effects may confound these measures. Alternatively, interstitial and cellular volumes could be assessed by other methods, such as iontophoresis and stereology, respectively. We did not assess total or regional parenchymal volumes because our cell density and size findings were post hoc findings and we only took enough sections to assess lesion volume. More sophisticated stereological assessment may provide more accurate measures of cell size and density [
17,
20]. Thus, future studies are essential to test and extend our initial findings and interpretation.
The threshold at which ICP rises become dangerous should be clarified. Well-defined thresholds for harmful ICP would aid clinical management [
43]. Currently, a thorough understanding of ICP thresholds is lacking, especially for brain hemorrhage. Indeed, current guidelines suggesting a threshold of 20 mmHg are largely based upon management principles for traumatic brain injury [
1]. Notably, we observed ICP >20 mmHg to be potentially fatal in rodent stroke models [
9,
13]. Alternatively, future studies could directly assess cranial compliance after brain injury. The ability of the cranial space to compensate for pressure changes may better relate to outcome than ICP alone because of the predictive potential of compliance measures (e.g., pressure-volume index) for dangerous pressure spikes [
43]. However, it is unclear how predictive rat ICP changes are of the human case or even how best to analyze ICP [
44]. We also note that studies on the benefits of ICP monitoring after acute brain injury are inconclusive [
45].
In summary, we show that severe ICH induced by large striatal infusion of blood does not cause significant and persistent ICP rises, indicating that the whole blood model of ICH does not adequately mimic the human condition in this regard. In addition, we provide initial evidence that brain parenchymal volume markedly decreases in response to large hemorrhagic lesions in the autologous blood infusion and collagenase models. Diminished parenchymal volume coupled with little peri-hematoma edema is perhaps the most parsimonious explanation as to why the blood infusion model does not cause large ICP rises, but this should be confirmed. These findings may question the doctrine that CSF and blood are the only sources of compensatory reserve to maintain normal ICP. The functional impact of reduced parenchymal volume should be assessed. We encourage further study on the mechanisms underlying ICP rises and compensatory responses after brain injury so that more effective treatments can be developed.