Perihematomal Edema After Spontaneous Intracerebral Hemorrhage.

Abstract

HomeStrokeVol. 50, No. 6Perihematomal Edema After Spontaneous Intracerebral Hemorrhage Free AccessReview ArticlePDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissions ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toFree AccessReview ArticlePDF/EPUBPerihematomal Edema After Spontaneous Intracerebral Hemorrhage Natasha Ironside, MBChB, Ching-Jen Chen, MD, Dale Ding, MD, Stephan A. Mayer, MD and Edward Sander Connolly Jr, MD Natasha IronsideNatasha Ironside From the Department of Neurological Surgery, Columbia University Medical Center, New York, NY (N.I., E.S.C.) Search for more papers by this author , Ching-Jen ChenChing-Jen Chen Department of Neurological Surgery, University of Virginia, Charlottesville (C.-J.C.) Search for more papers by this author , Dale DingDale Ding Department of Neurological Surgery, University of Louisville School of Medicine, KY (D.D.) Search for more papers by this author , Stephan A. MayerStephan A. Mayer Department of Neurology, Henry Ford Health System, Detroit, MI (S.A.M.). Search for more papers by this author and Edward Sander Connolly JrEdward Sander Connolly Jr Correspondence to Edward Sander Connolly, Jr, MD, Department of Neurological Surgery, Columbia University Medical Center, 710 W, 168th St, New York, NY 10032. Email E-mail Address: [email protected] From the Department of Neurological Surgery, Columbia University Medical Center, New York, NY (N.I., E.S.C.) Search for more papers by this author Originally published2 May 2019https://doi.org/10.1161/STROKEAHA.119.024965Stroke. 2019;50:1626–1633Other version(s) of this articleYou are viewing the most recent version of this article. Previous versions: May 2, 2019: Ahead of Print Mechanical disruption of adjacent neurons and glia after intracerebral hemorrhage (ICH) can result in both immediate and delayed neurological injury, with the eventual consequences of long-term functional impairment and mortality.1 In patients who survive the initial injury, inflammatory and cytotoxic responses to the hematoma and its breakdown components cause further damage to the surrounding parenchyma in the ensuing days to weeks, which represents a potential window for therapeutic intervention.2 In addition to attempts at surgically reducing the immediate mass effect of the hematoma, efforts to identify novel therapies that target secondary injury pathways are ongoing.3–7As the common end point for thrombin accumulation, influx of inflammatory mediators, and erythrocyte lysis, perihematomal edema (PHE) represents a promising surrogate marker of secondary brain injury after ICH.8–10 The temporal course of PHE development correlates with the manifestation of secondary injury, and PHE can be readily quantified on neuroimaging studies.11–13 Although some studies have found an association between PHE and clinical outcomes, others have reported no such relationship.13–23 These inconsistent findings may be influenced by heterogeneous PHE quantification methods, an incomplete understanding of the natural history of PHE, and selection bias from retrospective studies.13,19,20 To serve as a reliable indicator of secondary brain injury in ICH clinical trials, an improved understanding of PHE is necessary. Here, we critically assess the pathogenesis, quantification, and natural history of PHE, as well as its impact on clinical outcome in ICH patients.Pathophysiology of PHE FormationThe evolution of PHE occurs over 3 distinct phases (Figure 1). First, in the early phase (ie, within 1–4 hours of hematoma formation), pressure gradients between the surrounding parenchyma and capillaries may contribute to early PHE formation.24–26 Activation of the coagulation cascade results in clot retraction and serum protein extrusion from the hematoma, which elevates the interstitial osmotic pressure and promotes fluid movement from capillaries into the surrounding parenchyma.27 Elevated concentrations of serum protein in the surrounding parenchyma can be seen within 1 hour after the onset of ICH.28,29 This osmotic edema can even be observed in large-animal ICH models with an intact blood-brain barrier, underscoring the role of an osmotic gradient in early PHE formation.28 Protein concentration in the PHE fluid has also been inversely correlated with PHE clearance, and the osmotic gradient mechanism has been implicated in the rate of PHE expansion.30 Therefore, PHE evolution during the first few hours after ICH may be primarily mediated by clot retraction and serum protein accumulation in the surrounding parenchyma.Download figureDownload PowerPointFigure 1. Pathophysiology of perihematomal edema after spontaneous intracerebral hemorrhage. BBB indicates blood-brain barrier; Pc, capillary hydrostatic pressure; Pi, interstitial hydrostatic pressure; πc, capillary colloid osmotic pressure; and πi, interstitial colloid osmotic pressure.The role of cytotoxic edema in the acute phase after ICH remains unclear.31,32 One hypothesis is that microvascular compression causes hypoperfusion and failure of the cell membrane’s ion pumps in the perihematomal region, thereby further contributing to an osmotic gradient.26,33–35 In a study by Li et al36 comprising 21 ICH patients, cytotoxic edema in the perihematomal parenchyma was identified by diffusion-weighted magnetic resonance imaging in 10 patients from between 24 and 72 hours. Cytotoxic edema was associated with a higher 24-hour PHE growth rate (P=0.036) and a larger 72-hour PHE volume (P=0.020). The temporal pattern of these findings is consistent with histological observations of mitochondrial dysfunction in the perihematomal region among patients undergoing surgical ICH evacuation within 72 hours after onset.37 In contrast, Butcher et al33 reported an association between higher apparent diffusion coefficient values and higher PHE volumes (r=0.54; P=0.012) at a mean time of 21 hours after ICH. In a subsequent study, Li et al38 observed an association between greater 72-hour PHE formation and worse 90-day functional outcome (P<0.001), but they found no association between 72-hour cytotoxic edema formation and 90-day functional outcome. Therefore, despite the potential contributions of cytotoxic edema to PHE formation, in addition to that of plasma-derived osmotic edema, its clinical importance has yet to be determined.Next, in the intermediate phase of PHE evolution (ie, between 4–72 hours after hematoma formation), the thrombin mediates the conversion of fibrinogen to fibrin in the final common pathway of the coagulation cascade.39 Thrombin can also induce PHE formation independent of fibrinogen, which suggests that a pathophysiological mechanism exists that is distinct from the effects mediated by clot retraction.8 Direct thrombin infusion has been associated with pronounced inflammatory responses in experimental ICH models, including mitosis induction, leukocyte chemotaxis, platelet aggregation, and cytokine release.40,41 Furthermore, thrombin alters endothelial cell-cell and cell-matrix interactions, and therefore, it could mediate opening of the blood-brain barrier.26,35,42 Following blood-brain barrier disruption, activation of the complement cascade in the perihematomal parenchyma has been independently associated with PHE formation.9,43 Anaphylotoxins C3a and C5a serve as chemoattractants by activating endothelial cells and enhancing microglia infiltration, thereby amplifying the inflammatory response.9 Complement inhibition by N-acetylheparin attenuates thrombin-induced PHE and improves neurological function in rat ICH models.44 Furthermore, thrombin-specific inhibition is associated with a reduction in cerebral edema that is independent of hematoma volume.8,45 Therefore, thrombin-induced PHE formation is likely to result from the culmination of multiple coordinated mechanisms that are responsible for secondary brain injury.Finally, in the late phase of PHE evolution (ie, >72 hours after hematoma formation), hematoma resolution occurs by a combination of erythrocyte lysis and erythophagocytosis.2,46 Although erythrophagocytosis is believed to accelerate ICH recovery, erythrocyte lysis is cytotoxic.9,46,47 Depletion of intracellular energy reserve and formation of the complement membrane attack complex in erythrocytes results in subsequent cell lysis, leading to hemoglobin accumulation in the surrounding parenchyma.2,39,46,48 Hemoglobin inhibits Na+/K+ adenosine triphosphatase activity, generates hydroxyl radicals, stimulates lipid peroxidation, and causes neuronal death.27,48,49 Elevated hemoglobin levels can be detected in the cerebrospinal fluid during the first several days after ICH.50 In a rat ICH model, Xi et al10 observed marked brain edema 24 hours after intracerebral infusion with lysed autologous erythrocytes. In contrast, brain edema was delayed in rats infused intracerebrally with packed erythrocytes, implicating hemoglobin in edema formation.10 In addition, intraparenchymal infusion of lysed erythrocytes increases blood-brain barrier permeability without affecting cerebral blood flow, which suggests an underlying vasogenic mechanism.51,52 In preclinical ICH models, administration of deferoxamine, an iron chelator, is associated with reductions in edema formation, white matter injury, and neuronal cell death.53 Deferoxamine is also associated with decreased erythrocyte lysis and erythrophagocytosis in porcine ICH models.47 Therefore, the precise relationships among hematoma clearance, PHE formation, and ICH recovery remain incompletely defined.Natural History of PHEIn experimental ICH models, PHE typically develops within 2 hours, peaks on day 3, and persists for up to 7 days.29 However, edema peaks at 24 to 48 hours after intraparenchymal thrombin infusion and at 3 to 5 days after packed erythrocyte infusion.10 Therefore, different pathophysiological mechanisms may influence the temporal patterns of PHE formation. To date, human studies have marked the natural history of PHE by 2 distinct phases (Figure 2).11–13,23,54 An initial phase of rapid PHE growth has been observed during the first 24 hours after ICH.11,13,19 Subsequently, an inverse correlation between the PHE growth rate (cm/d) and time from onset (days; r2=0.82) has been demonstrated.13 Although peak PHE volume has been suggested to occur between 2 to 3 weeks after ICH, PHE has been observed to progress up to 21 days after ICH.12,13,22,55–57 However, characterizing the natural history of PHE has been limited by inconsistent durations of radiological follow-up, heterogeneous methods of PHE quantification, and small cohort sizes. In addition, the relationship between the pathophysiological mechanisms of secondary brain injury and PHE formation is not well understood in humans. As such, further natural history studies that detail the evolution of PHE in ICH patients are necessary.Download figureDownload PowerPointFigure 2. Perihematomal edema (PHE) evolution in a patient with spontaneous intracerebral hemorrhage (ICH). Segmentation of the ICH (red) and PHE (blue) was performed manually.Quantification of PHEA multitude of quantification methods for PHE have been developed to explore its potential association with clinical outcomes.13,18–21,23,58 Although the T2-weighted hyperintensity of PHE on magnetic resonance imaging effectively delineates it from the surrounding parenchyma, magnetic resonance imaging is not routinely used as a neuroimaging modality in ICH patients.12,59 Because of its accessibility and rapid acquisition, computed tomography (CT) may be a more suitable neuroimaging modality for PHE quantification.1,12 However, PHE manifests as a hypodensity on CT that poses challenges to threshold-based or edge-detection algorithms, due to its similarity in Hounsfield unit density to both cerebrospinal fluid and microangiopathy.13,60,61 Manual segmentation methods are time-consuming and cumbersome, and they have reportedly high rates of intra and interobserver variability.58,60 To mitigate the shortcomings of the various quantification methods, Volbers et al60 developed a semiautomated segmentation method, in which a generous region of interest, including the hematoma and surrounding hypodensities, was manually traced on each CT slice before application of a preset Hounsfield unit threshold range. With an optimized range of 5 to 33 Hounsfield unit, the semiautomated segmentation method achieved higher interobserver (0.96 versus 0.89) and intraobserver (0.96 versus 0.90) reliabilities than manual segmentation methods.60 Subsequently, a similar semiautomated segmentation method that permitted user discretion in Hounsfield unit threshold selection demonstrated a strong correlation between CT-based PHE volume measurements and those derived from the closest available T2-weighted magnetic resonance imaging (r2=0.98; P<0.0001).61 However, semiautomated segmentation methods are limited by a lack of external validation, small derivation sample sizes, preferential selection of early (0–5 days after ICH) CT scans, and omission of segmentation time comparisons. Accurate and reliable quantification of PHE will be crucial to future ICH clinical trials that utilize PHE as a surrogate indicator of clinical outcome.13,20 Therefore, optimization and standardization of PHE volumetric analysis is needed.PHE as a Marker of Clinical OutcomeDespite significant advancements in understanding the mechanisms that drive PHE formation, the implications of PHE on clinical outcomes are unclear (Table). Because hematoma volume is associated with clinical outcomes and PHE is correlated with hematoma volume, hematoma volume may confound the potential association between PHE and clinical outcome.13,15,16,18,19,62 In patients with hematoma volume <30 cm3, studies have reported associations between admission PHE volume and discharge modified Rankin Scale ≥3 and between 72-hour PHE growth and 90-day modified Rankin Scale ≥3.20,58 To control for hematoma volume, several PHE studies have investigated the association between relative PHE (defined as PHE volume/ICH volume) and clinical outcome.15 Although an early study by Gebel et al found an inverse correlation between larger initial relative PHE and worse 90-day functional outcomes, subsequent studies have failed to identify a relationship between relative PHE and outcomes.12,15,18,22,56,58 This has been attributed to disproportionately large relative PHE values in patients with small ICH volumes.20,63Table. Summary of Pertinent Studies Investigating the Relationship Between Perihematomal Edema and Clinical Outcomes After Spontaneous ICHStudyDesignPatientsNeuroimaging MethodEdema Measurement TimeEdema Evaluation MetricOutcomeFindingsMcCarron et al62Prospective, single-centerN=102, ICHCTAt presentationAbsolute PHEIn-hospital mortalityPHE was associated with mortality.Gebel et al15Prospective, single-centerN=86, supratentorial ICHCT<3 h after ICH onsetAbsolute PHE, relative PHE12-wk mRS score ≥3, 12-wk Barthel Index <85, 30-d mortalityRelative PHE was associated with worse functional outcome, but not mortality. Absolute EV did not predict outcomes.Sansing et al16Retrospective, single-centerN=80, supratentorial ICHCTNot specifiedAbsolute PHEDischarge to skilled nursing facility or death, 90-d mRS score >3PHE was associated with worse discharge and 90-d outcomes.Levine et al14Retrospective, single-centern=49 warfarin-associated ICH, n=49 matched controlsCTAt presentationAbsolute PHE90-d mortalityPHE was associated with reduced mortality.Zubkov et al17Retrospective, single-centerN=88 warfarin-associated ICHCTNot specifiedAbsolute PHE increase7-d mortality, discharge mRS score >3PHE increase was not associated with mortality or functional outcome.Arima et al18Retrospective, multicenterN=270, ICHCTAt presentation, 24±3 h, 72±3 hAbsolute and relative PHE increase.90-d mRS score ≥3 or mortalityNeither absolute nor relative PHE increase was associated with outcome.Venkatasubramanian et al12Prospective, single-centerN=27, supratentorial ICH ≥5cm3MRI48±12 h, 7±1 d, 14±2 d, 21±3 d after presentation48 h absolute and relative PHE, absolute and relative PHE increase, peak relative PHE≥2 point NIHSS score increase at 48 h, 90-d mRS score, Barthel Index, eGOS, and mortality.48 h absolute PHE was associated with neurological deterioration. 48 h relative PHE, 48 h absolute, and relative PHE increase were not associated with neurological deterioration. Peak relative PHE was not associated with 90-d outcomes.Sansing et al64Prospective, single-centerN=303, ICHMRI72 h after presentation.Absolute PHE90-day mRS scorePHE was associated with worse outcome.Staykov et al56Retrospective, single-centerN=219, supratentorial ICHCT1, 2, 3, 4–6, 7–11, 12–16, 17–21, >22 d after presentationPeak absolute and relative PHE. Absolute and relative PHE increase between day 1–3In-hospital mortalityAbsolute PHE increase from day 1–3 was associated with mortality. Peak PHE and relative PHE increase were not associated with mortality.Li et al36Prospective, single-centerN=59, ICHMRIAt presentation and 72 hAbsolute PHE, % cytotoxic edema determined using ADC90-d mRS score ≤372 h absolute PHE was associated with worse outcome. PHE at presentation and % cytotoxic edema were not associated with outcome.Li et al38Prospective, single-centerN=21, ICHMRI≤24 h, 72±12 h, 7±1 d after ICH onsetAbsolute PHE, % cytotoxic edema determined using ADC90-d mRS score ≤372 h absolute PHE was associated with worse outcome. % cytotoxic edema was associated with a trend towards worse outcome.Appelboom et al58Retrospective, single-centerN=133, ICHCTAt presentationAbsolute and relative PHEDischarge mRS score ≤3Absolute PHE was associated with worse outcome only in patients with hematoma volumes ≤30 mL. Relative PHE was not associated with outcome.Bakhshayesh et al65Prospective, single-centerN=63, ICHCT≤24 h, 72 h after ICH onsetAbsolute and relative PHE, absolute PHE increaseIn-hospital mortality, 90-d mRS score ≤372 h PHE increase was associated with in-hospital mortality. Neither absolute nor relative PHE was not associated with functional outcome.Yang et al19Retrospective, multicenterN=1,138, ICHCTAt presentation and 24±3 hAbsolute PHE increase90-d mRS score ≥3 or mortalityPHE increase was associated with worse outcome.Murthy et al20Retrospective, multicenterN=596, ICHCTAt presentation and 72 hAbsolute PHE increase90-d mRS score ≥3PHE increase was associated with worse outcome. This effect was most apparent in basal ganglia ICH and in ICH <30 mL.Volbers et al22Retrospective, single-centerN=220, supratentorial ICHCT1, 2–3, 4–6, 7–9, 10–12 d after presentationPeak absolute and relative PHE. Absolute and relative PHE increase between day 1–3.Discharge mRS score ≤3Peak absolute PHE volume was associated with worse outcome. Peak relative PHE, absolute PHE increase, and relative PHE increase were not associated with outcome.Urday et al54Retrospective, single-center.N=139, supratentorial ICHCTAt presentation, 24 h and 72 hAbsolute PHE. Absolute PHE increase.90-d mortality, 90-d mRS score ≥372 h PHE, 24 h PHE increase, and 72 h PHE increase were associated with mortality and worse functional outcomes.Murthy et al21Retrospective, multicenter.N=596, ICH.CTAt presentation and 72 hAbsolute PHE expansion rate90-d mortality, 90-d mRS score ≥3PHE expansion was associated with mortality and worse functional outcomes.Wu et al13Retrospective, single-centerN=861, ICHCTAt presentation and 72 hEdema extension distance6-mo mortality72 h edema extension distance was associated with mortality.Grunwald et al23Retrospective, single-centerN=115, ICHCTAt presentation, 24 h and 72 hAbsolute PHE expansion rate90-d mortality, 90-d mRS score ≥324 h PHE expansion was associated with mortality for deep and lobar ICH. 72 h PHE expansion was associated with worse functional outcomes exclusively in deep ICH.Volbers et al, 201866Retrospective, single-centerN=292, supratentorial ICHCT1, 2–3, 4–6, 7–9, 10–12 d after presentationPeak absolute PHE. Absolute PHE increase between day 1–3.90-d mRS score ≤3Peak PHE was associated with worse outcome. Absolute PHE increase was not associated with outcome.ADC indicates apparent diffusion coefficient; CT, computed tomography; eGOS, extended Glasgow Outcome Scale; ICH, intracerebral hemorrhage; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; and PHE, perihematomal edema.Recently, there has been an intensified focus on the relationship between PHE growth and clinical outcomes. In numerous studies, 72-hour PHE growth has been independently associated with 90-day functional outcomes, as well as with in-hospital, 90-day and 6-month mortality.13,19–21,23,36,38,54,56,64,65 These observations may reflect the role of secondary injury activation as an effector of PHE evolution.2,9,66 Edema extension distance is a novel PHE metric, defined as the mean thickness of PHE beyond the boundary of the hematoma, that arose from the hypothesis that the intensity of post-ICH inflammatory injury would manifest as a linear extension of PHE from the ICH border.63 Wu et al13 reported that a higher than projected edema extension distance at 72 hours predicted 6-month mortality (odds ratio, 1.60 [1.04–2.46]). However, because multiple coordinated injury mechanisms are thought to influence PHE formation, the use of edema extension distance as a clinical surrogate requires further validation.2,10 Furthermore, the implications of delayed edema formation on clinical outcomes have not been thoroughly characterized. Therefore, the temporal associations between PHE formation and clinical outcomes warrant comprehensive evaluation in a large, prospectively derived ICH patient cohort using standardized measurement techniques.Predictors of PHE FormationA greater understanding of factors that influence the extent of PHE formation may aid in the identification of potential therapies. ICH volume has been suggested to drive PHE growth, and significant correlations between baseline ICH and PHE volumes have been observed.18,58 Baseline ICH volume has also been associated with 72-hour PHE growth, which could represent the influence of hematoma blood components on the pathophysiology of PHE.8,10,41,47,49,50,52 PHE formation may be further influenced by the cause of the ICH.14,17 Levine et al14 reported warfarin-associated ICH to be associated with a lower relative PHE volume at admission (P<0.05), which is consistent with the facilitatory effects of coagulation cascade activation on PHE formation. Similarly, higher platelet count has been associated with increased PHE growth during the first 5 days after ICH (P=0.0013), and higher admission hematocrit has been associated with a greater delay in peak PHE formation (P=0.06).12,16 These findings support the time-dependent effects of coagulation on PHE formation.A history of hypertension was associated with greater 72-hour relative PHE growth (P=0.04) in patients enrolled in the INTERACT (Intensive Blood Pressure Reduction in Acute Cerebral Hemorrhage Trial).18 In INTERACT, admission systolic blood pressure negatively correlated with 72-hour absolute PHE growth (P=0.02). Decreased baroreflex sensitivity and cerebral perfusion abnormalities during the acute phase of ICH have been linked to the development of PHE.67 Although underpowered, an association between aggressive inpatient systolic blood pressure control and a reduction in 24-hour relative PHE growth (relative risk, 0.74 [0.43–1.29]) was not observed in the ATACH trial (Antihypertensive Treatment of Acute Cerebral Hemorrhage).68 As such, there is insufficient evidence to suggest that intensive systolic blood pressure management impairs PHE formation.Emerging Medical Therapies for PHEHyperosmolar agents, which have been effectively used to manage elevated intracranial pressure associated with PHE formation, are thought to act primarily via generation of an intravascular osmotic gradient.1,69 However, a correlation between osmotic gradient formation and clinical improvement has not been consistently demonstrated.70 Alternative mechanisms of hyperosmolar therapy, including modulation of free radical scavengers and reduction of serum viscosity, have been posited.70 Among 303 ICH patients enrolled in the placebo arm of the CHANT trial (Cerebral Hemorrhage and NXY-059), in-hospital administration of antiadrenergic agents was associated with a reduction in the 72-hour PHE volume, after adjusting for medication effects on ICH volume and blood pressure.64 Although β-agonist administration was not associated with a difference in 90-day functional outcomes, a larger 72-hour PHE volume was associated with worse outcome (odds ratio, 1.04 [1.02–1.05]).64In a cohort of 125 ICH patients, prior statin exposure, which has demonstrated a neuroprotective effect in preclinical studies, was independently associated with a reduction in both absolute (P=0.035) and relative (P=0.021) PHE volumes at admission.71 However, the effects of statin therapy on PHE evolution have not been investigated. Although numerous clinical trials targeting secondary brain injury pathways after ICH have utilized PHE volume as an outcome measure, its clinical significance is contingent upon the potential association between PHE and clinical outcomes.3–7 This association is complicated by many confounders, including ICH volume, PHE quantification method, and timing of PHE measurement.ConclusionsPHE is a promising surrogate indicator of secondary brain injury after ICH. In preclinical ICH models, PHE represents the common end point of several pathophysiological pathways induced by ICH. Current PHE quantification methods include manual and semiautomated segmentation, with inherent variations within each technique. Natural history studies in humans have suggested rapid PHE growth within the first 24 hours following ICH, with subsequent progression of PHE for up to 2 to 3 weeks thereafter. However, an association between PHE and clinical outcomes has not been consistently demonstrated. Human studies investigating PHE have been limited by their retrospective designs, lack of standardized PHE quantification methods, and insufficient radiological follow-ups. Future studies that address these limitations are essential to the validation of PHE as a viable end point in ongoing and forthcoming ICH clinical trials.DisclosuresNone.FootnotesCorrespondence to Edward Sander Connolly, Jr, MD, Department of Neurological Surgery, Columbia University Medical Center, 710 W, 168th St, New York, NY 10032. Email [email protected]columbia.edu