Predictive markers of hemorrhagic transformation in ischemic stroke

Full Text

List of abbreviations

HI, hemorrhagic infarction

HT, hemorrhagic transformation

BBB, blood-brain barrier

CT, computed tomography

HDL, high-density lipoprotein

LDL, low-density lipoprotein

TC, total cholesterol

PH, parenchymatous hematoma

TG, triglycerides

AUC, area under the curve

FAR, fibrinogen-to-albumin ratio

ICAM-1, intercellular adhesion molecule-1

IL, interleukin

MMP, matrix metalloproteinase

NLR, neutrophil-to-lymphocyte ratio

NSE, neuron-specific enolase

NT-proBNP, N-terminal pro B-type natriuretic peptide

PAI-1, plasminogen activator type 1 inhibitor

rt-PA, recombinant tissue plasminogen activator

TGF-β, transforming growth factor-β

TIMP, tissue inhibitor of metalloproteinases

VCAM-1, vascular cell adhesion molecule-1

VEGF, vascular endothelial growth factor

ZO-1, zonula occludens-1

INTRODUCTION

Despite recent decreases in cerebral stroke mortality (from 20.7% in 2020 to 17.6% in 2022), this condition remains a major socioeconomic challenge in Russia. More than 450,000 stroke cases are reported annually, with ischemic stroke accounting for 80%. Post-stroke disability can be caused by the lesion severity and site, as well as complications in the acute phase, which may promote neurological deficit. In 2019, 22% of post-stroke patients completely stopped working, with a confirmed disability in 60%; another 6% (27,500) of patients had to shift to part-time work. Inpatient treatment costs considerably contribute to the economic burden of stroke (up to 0.5 trillion rubles per year; ~70%). Rehabilitation costs account for up to 30%, with 84% of all costs associated with ischemic stroke [1].

Reperfusion therapy (drug-induced thrombolysis and/or endovascular thrombectomy) is an effective treatment option in ischemic stroke, reducing mortality and disability rates. In recent years, the number of reperfusion procedures in Russian specialty units for patients with stroke has increased considerably [2]. Therefore, it is relevant to investigate the mechanisms of complications in patients with ischemic stroke, including after reperfusion therapy. One such complication is hemorrhagic transformation (HT) of cerebral infarction, which worsens the prognosis and limits treatment and rehabilitation options. The incidence of spontaneous HT is 10%–43%, and the incidence of symptomatic HT after thrombolytic therapy is 2%–7% [3].

PATHOGENESIS OF HEMORRHAGIC TRANSFORMATION

The pathogenesis of hemorrhagic transformation is largely mediated by damage to the blood-brain barrier (BBB). The BBB is a physiological barrier between circulating blood and brain matter, consisting of endothelial cells, astrocytes, pericytes, neurons, and extracellular matrix [4]. The BBB protects the brain from toxic substances in the blood, and its selective permeability ensures that the brain receives needed nutrients and other functional molecules from the blood, as well as allows the transport of metabolic wastes into the blood [5]. Endothelial cells are essential for regulating ion transport and maintaining selective molecular permeability. The increased number of mitochondria in these cells produces energy necessary for nutrition and protection of the brain. Endothelial cells of cerebral microvessels that form the BBB have tight junctions, lack fenestras, and have minimal pinocytic activity, which considerably limits the paracellular pathway compared to peripheral circulation. Transmembrane tight junction proteins in cerebral endothelial cells, such as claudin-5, occludin, and zonula occludens (ZO) proteins, form complexes with cytoskeletal proteins and crosslink cell membranes, providing high transendothelial resistance and mediating selective permeability for any hydrophilic substances and water. The entry of immune cells into the central nervous system is limited by low expression of leukocyte adhesion molecules on endothelial cells [5]. The extracellular matrix, which is composed of structural proteins such as laminin, fibronectin, collagen type IV, elastin, trombospondin, and proteoglycanes, is essential for BBB stability and regulation of intercellular communication. Astrocytes are also important components of the BBB. They form perivascular sheaths, provide neurons with energy, maintain synapse growth, and regulate brain water content and electrolyte balance via aquaporin-4 [4].

Ischemic stroke is associated with neuroinflammatory processes, including microglia and astrocyte activation, as well as secretion of pro-inflammatory cytokines and growth factors such as tumor necrosis factor alpha (TNF-α), interleukins (IL-1β and IL-17), interferon gamma (INF-γ), vascular endothelial growth factor (VEGF), and matrix metalloproteinase(MMP). This activates cerebral endothelial cells, reducing the barrier function and increasing the expression of cell adhesion molecules [4]. Neutrophils and lymphocytes are recruited to the damage site in the brain, interacting with the reactive endothelium and attaching to endothelial cells. They then migrate into the parenchyma and release MMPs, specifically MMP-9, which directly degrade tight junction proteins and the basement membrane, destabilizing the BBB [6]. Lymphocytes infiltrate the brain after neutrophils, releasing a variety of pro- and anti-inflammatory cytokines. Monocytes, after transferring into the brain tissue, differentiate into macrophages, promoting the expression of certain factors protecting the BBB [4].

Increased BBB permeability disrupts ion transport. As a result, Cl^–, Na⁺, and water enter the cells, causing swelling and intracellular edema. As the BBB is further damaged, water and some plasma proteins enter the brain parenchyma due to increased permeability, resulting in vasogenic edema. Increased BBB permeability promotes red blood cell extravasation, leading to the deposition of neurotoxic hemoglobin products, such as free iron, in the brain parenchyma. Active bivalent iron promotes lipid peroxidation and oxidative damage, which causes neuronal death and worsens cerebral edema. The resulting reactive oxygen species further damage the BBB, promoting blood cell extravasation and leading to HT [5].

Recanalization therapy restores the normal blood flow and causes reactive hyperemia with loss of cerebral vasoregulation. This worsens cytotoxic edema and promotes reactive microvasculature obstruction that further aggravates BBB breakdown. Vasogenic edema increases intercellular permeability [7]. Recombinant tissue plasminogen activator (rt-PA) has been shown to disrupt the BBB through several mechanisms, including the expression of LDL-receptor-related protein-associated protein on endothelial cells, microglia, and astrocytes, increasing plasma kallikrein levels and platelet-derived growth factor C activation [8–10]. Shi et al. [11] found that rt-PA also activates immune cells, promoting HT after ischemic stroke. Mechanical thrombectomy also increases the risk of HT due to endothelial trauma, edema in the intimal and medial layers, basement membrane destruction, and rapid reperfusion [7].

Therefore, ischemic stroke is associated with a cascade of inflammatory cellular and biochemical reactions that cause damage to the BBB, which can be exacerbated by reperfusion therapy and result in HT. Tight junction proteins, matrix metalloproteinases, cytokines, and proteins of the activated endothelium and astroglia are the major contributors.

CLASSIFICATION OF HEMORRHAGIC TRANSFORMATION

According to neuroimaging findings (computed tomography [CT] or magnetic resonance imaging [MRI] of the brain performed after reperfusion therapy), HT is traditionally classified by the type of hemorrhagic infarction (HI). It is characterized by petechial hemorrhages in the infarction zone or at a distance from it and parenchymal hematomas (PHs) in cases where a more homogeneous, dense hematoma with mass effect forms in the ischemic zone or at a distance from it [12].

In 1999, Fiorelli et al. [13] updated the classification criteria for various types of HT, which are still in use today. Two HI subtypes (HI1 and HI2) and two PH subtypes (PH1 and PH2) were described based on neuroimaging findings:

1. HI type 1 (HI1) is characterized by small petechial hemorrhages in the infarction zone, less than 30% of the infarction zone;
2. HI type 2 (HI2) is characterized by confluent petechial hemorrhages in the ischemic zone without mass effect, more than 30% of the infarction zone;
3. PH type 1 (PH1) is a hematoma occupying less than 30% of the ischemic zone, with insignificant mass effect;
4. PH type 2 (PH2) is a hematoma occupying more than 30% of the ischemic zone, with significant mass effect, or any hematoma outside the ischemic zone [13].

PH2 are the most clinically relevant since they considerably worsen the condition and determine the prognosis in patients after ischemic stroke. Moreover, they have clear radiologic signs: these are dense, homogeneous hematomas measuring more than 30% of the ischemic zone, with a significant mass effect [14].

Hemorrhagic transformation is classified as symptomatic or asymptomatic based on a ≥4-point worsening of the neurological status on the National Institute of Health Stroke Scale (NIHSS) within the first 24–36 h of disease onset [15, 16]. The following risk factors for HT are identified based on neuroimaging findings: infarction volume; early signs of brain ischemic lesions (smoothing of the cerebral sulci, loss of definition of the gray-white interface, insular ribbon sign); hyperdense middle cerebral artery sign on CT; severe leukoaraiosis; collateral circulation; changes on perfusion CT in the form of a large-core infarction [17].

Thus, it is relevant to identify new accurate and easy-to-use markers for predicting HT. These markers may include laboratory parameters, both routine and specific.

ROUTINE LABORATORY PARAMETERS IN PREDICTING HEMORRHAGIC TRANSFORMATION

Liu et al. [18] examined 1207 patients and identified 17 routine HT prognostic biomarkers indicative of inflammation: leukocytes; monocytes, neutrophils, lymphocytes, and neutrophil-to-lymphocyte ratio (NLR); clotting/fibrinolysis disorder (platelets); vasoreactivity (creatinine and cystatin C); lipid metabolism (total cholesterol [TC], triglycerides [TG], high-density [HDL] and low-density [LDL] lipoproteins; liver function (alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, total bilirubin, glucose). The authors emphasize that a single biomarker is not sufficient to assess the risk of HT, and several parameters must be considered. The platelet count, NLR, and LDL levels at admission had the highest prognostic value.

Sun et al. [6] found that NLR is an independent risk factor for spontaneous HT after acute ischemic stroke, confirming the critical role of immune inflammatory responses in the pathogenesis of HT. The relative risk of spontaneous HT in patients with high NLR (≥3.14) was 4 times higher than in patients with low NLR (≤3.14). Xie et al. [19] also found that NLR plays a critical role in the prognosis of HT; however, the threshold value was considerably higher (10.59), and the relative risk of HT was more than 7 times higher. This marker is the simplest and quickest to use; however, some researchers question its prognostic value for stroke outcomes and treatment.

Several studies assessed the association between hyperglycemia at admission and HT, especially after reperfusion therapy. The multicenter randomized study DIRECT-MT (2018) [20] found that elevated glucose levels at admission were an independent predictor of symptomatic HT after thrombectomy. These findings were confirmed by Kuang et al. [21]; their predictive model included not only glucose levels but also the Alberta Stroke Program Early CT Score (ASPECTS) and collateral circulation. Comparable findings were previously reported for thrombolysis: hyperglycemia was independently associated with HT in patients treated with intravenous recombinant tissue-type plasminogen activator [22]. Elevated glucose levels can promote HT via increased formation of reactive oxygen species due to hyperglycemia, increased MMP activity, and overexpression of pro-inflammatory cytokines, which disrupts the BBB. Other potential mechanisms include increased severity of ischemic lesions due to impaired micro- and macrocirculation. Moreover, a long-term increase in blood glucose levels promotes abnormal changes around the walls of microvessels that form the BBB [23].

Urinary albumin, a widely used laboratory marker, can indicate endothelial dysfunction and damage. Micro- and macroalbuminuria are a well-established risk factor for cardiovascular morbidity and mortality. An assessment of urinary albumin levels in 154 patients found that micro- and macroalbuminuria were independently associated with severe HT after thrombolysis in patients with acute ischemic stroke. Moreover, the severity of albuminuria was associated with the severity of HT [24]. Micro- and macroalbuminuria after intravenous rt-PA administration can indicate systemic endothelial damage and increased risk of HT. However, further research is needed to determine whether albuminuria is caused by thrombolysis or represents an endothelial damage that precedes stroke [24].

Some researchers believe that the cerebrospinal fluid/serum albumin ratio is a reliable predictor of HT. Under normal physiological conditions, cerebrospinal fluid albumin levels are minor. However, following BBB damage in stroke, plasma albumin enters the cerebrospinal fluid. This parameter cannot be determined without an invasive lumbar puncture; therefore, this method is not widely used [25].

The fibrinogen-to-albumin ratio (FAR) is a novel systemic inflammatory marker associated with ischemic stroke and HT. FAR has been shown to predict HT after intravenous thrombolysis, with an AUC of 0.613 (95% CI: 0.530–0.695; р = 0.005) and an optimal threshold of 0.101. Higher FAR values are independently associated with HT after intravenous thrombolysis in patients with ischemic stroke [26].

The significance of lipid profile parameters was demonstrated by Luo et al. [27]. This study included 763 consecutive patients with ischemic stroke who received intravenous thrombolysis. LDL cholesterol, total cholesterol to HDL cholesterol ratio (TC/HDL-C), triglycerides to HDL cholesterol ratio (TG/HDL-C), and LDL cholesterol to HDL cholesterol ratio (LDL-C/HDL-C) were assessed in all patients. According to ROC curves, optimal threshold values for predicting HT and unfavorable outcomes were LDL 2.99 mmol/L and 2.01 mmol/L, TC/HDL-C 4.05 and 3.66, TG/HDL-C 0.82 and 1.02, and LDL-C/HDL-C 2.67 and 2.71, respectively. Low TC/HDL-C, TG/HDL-C, and LDL-C/HDL-C were associated with an increased HT risk after thrombolysis.

Routine hemostasis parameters of importance include thromboelastography indices, which can be obtained in a standard clinical diagnostic laboratory. In a study in 221 patients with ischemic stroke and thrombolysis, HT was detected in 40 cases. The R value (clotting time) was significantly higher in the HT group than in the non-HT group (6.65 min vs 5.50 min) (р < 0.001). Maximal amplitude was significantly lower in the HT group than in the non-HT group (61.28 vs 64.94) (p < 0.001). According to the authors, R can be used as a minimally invasive HT risk marker [28]. In a study that assessed the association between thromboelastography parameters and the functional outcome of stroke in 160 patients, it was found that a decrease in R, especially below 5 min, was associated with poor functional recovery. However, this study did not assess the risk of HT after endovascular therapy [29]. McDonald et al. [30] reported a longer clotting time in patients with HT after thrombolysis; however, there were no significant differences with patients without HT (р = 0.046).

Cardiac dysfunction caused by brain ischemia can result in increased N-terminal pro B-type natriuretic peptide (NT-proBNP) levels, which is common in patients with stroke. Stroke activates the hypothalamic–pituitary–adrenal axis, promoting the release of catecholamines. Overactivation of β-receptors due to catecholamine release may cause oxidative stress, calcium overload, and prolonged cardiomyocyte contraction, resulting in their damage and elevated NT-proBNP levels. Cardiac stroke is one potential cause for elevated NT-proBNP levels in HT. Another cause is that HT can worsen stroke-associated cardiac dysfunction. An assessment of NT-proBNP levels in 404 patients after intravenous thrombolysis revealed that NT-proBNP levels are independently associated with HT, with the most significant correlation observed in patients with atherothrombotic stroke. HT was detected in 72 patients, whose NT-proBNP levels were significantly higher than in non-HT patients (р < 0.05) [31].

Thus, routine laboratory parameters that can serve as prognostic markers for HT include NLR, glucose levels, albuminuria, FAR, lipid profile parameters (LDL, TG/HDL-C, TG/HDL-C, LDL-C/HDL-C), and NT-proBNP levels (Table 1).

 

Table 1

Routine and specific laboratory markers used to predict hemorrhagic transformation after ischemic stroke

Parameter	Role in HT pathogenesis, cause of elevated levels	Threshold value in HT prognosis	Source
Routine laboratory parameters used to predict HT
Neutrophil-to-lymphocyte ratio (NLR)	Role in immune inflammation Neutrophils and lymphocytes are recruited to the ischemia zone, migrate to the parenchyma, and activate MMPs, which directly degrade tight junction proteins and the basement membrane. Neutrophils secrete cytokines, chemokines, adhesion molecules, and various proteases, ultimately increasing BBB permeability and causing its dysfunction. Neutrophil migration to the brain parenchyma results in a compensatory increase in their blood levels	3.14	[6]
		10.59	[19]
Glucose	Hyperglycemia promotes the formation of reactive oxygen species, increases MMP activity and anti-inflammatory cytokine expression, and affects micro- and macrocirculation, increasing the severity of ischemic lesions. A long-term increase in blood glucose levels promotes abnormal changes around the walls of microvessels that form the BBB	No data	[23]
Urinary albumin	Micro- and macroalbuminuria can indicate systemic endothelial damage, which existed prior to thrombolysis or was caused by it	No data	[24]
Fibrinogen-to-albumin ratio (FAR)	A systemic marker of inflammation in ischemic stroke	0.101	[26]
LDL, mmol/L	Lipids influence the vascular wall and BBB integrity. The protective effect is primarily determined by ratios of various lipids. Lipid imbalance may result in BBB damage	2.99	[27]
TC/HDL-C, mmol/L		4.05
TG/HDL-C, mmol/L		0.82
LDL-C/HDL-C, mmol/L		2.67
Clotting time (R), thromboelastography	A longer clotting time due to hypocoagulation may be associated with the risk of HT	No data	[28, 30]
N-terminal pro B-type natriuretic peptide (NT-proBNP)	Stroke activates the hypothalamic–pituitary–adrenal axis, promoting the release of catecholamines. Overactivation of β-receptors due to catecholamine release may cause oxidative stress, calcium overload, and prolonged cardiomyocyte contraction, resulting in their damage and elevated NT-proBNP levels. HT can worsen stroke-associated cardiac dysfunction and increase NT-proBNP levels	No data	[31]
Specific laboratory markers of HT
Matrix metalloproteinase (MMP-9), ng/mL	MMPs destroy the extracellular matrix and degrade tight junction proteins, causing BBB damage, cerebral edema, and HT	140	[33]
MMP inhibitors (TIMPs)	TIMPs regulate intercellular matrix remodeling; imbalance may cause BBB damage due to MMP overactivation	No data	[35–37]
Fibronectin, μg/mL	An intercellular matrix protein secreted by endothelial cells; it is released to plasma when the basement membrane is damaged	3.6	[33]
S100B, μg/L	Is primarily expressed by mature astrocytes and released to blood in BBB damage	0.23	[25, 33]
Tight junction proteins; occludin, claudin-5, zonula occludens-1 (ZO-1), ng/mL	Promote tight junctions of endothelial cells and mediate selective permeability. Cerebral ischemia and MMP activation degrade tight junction proteins and increase BBB permeability	4.48 No data No data	[38, 41]
Intercellular adhesion molecule-1 (ICAM-1), ng/mL	Neuroinflammation activates cerebral endothelial cells, reducing the barrier function and increasing the expression of cell adhesion molecules, which causes BBB dysfunction	250.5	[4, 42]
Vascular endothelial growth factor (VEGF)	The expression increases by three times in brain edema. VEGF receptor antagonists inhibit HT	No data	[5, 43]
Neuron-specific enolase (NSE)	A neuron-specific enzyme that enters systemic circulation only when the nerve tissue is damaged; elevated levels are associated with the risk of HT	No data	[39]
Interleukin 33 (IL-33), ng/L	Is secreted by leukocytes recruited to the brain; its levels are elevated in neuroinflammation	67.66	[39]
Plasminogen activator type 1 inhibitor (PAI-1), ng/mL	The balance between coagulation and thrombolysis can contribute to HT. Low levels of PAI-1, an endogenous fibrinolysis inhibitor, can promote fibrinolysis and blood extravasation and are associated with HT	21.4	[43, 44]

Note. BBB, blood-brain barrier; HT, hemorrhagic transformation.

 

SPECIFIC LABORATORY MARKERS OF HEMORRHAGIC TRANSFORMATION

Metalloproteinases and Other Extracellular Matrix Proteins

Specific immunohistochemical laboratory markers that can predict HT have been investigated in numerous studies, including reviews and meta-analyses. A review by Yao et al. [32] examined new diagnostic markers and therapeutic targets in HT, including tight junction proteins, S100B protein, albumin, matrix metalloproteinases (MMP-9, MMP-2, MMP-12), interleukins (IL-1β, TNF-α, IL-6, IFN-γ, TGF-β, IL-10, IL-4, IL-13), adhesion molecules (ICAM-1, VCAM-1), selectins (P-selectin, E-selectin), VEGF, aquaporin-4, etc.

A meta-analysis of 2230 studies by Krishnamoorthy et al. [33] identified the most significant prognostic markers for HT, which included fibronectin, MMP-9, ferritin, S100B, and NLR. However, ferritin and fibronectin were found to have wide confidence intervals, reducing the reliability of these markers. MMPs are a family of endogenous proteolytic zinc-containing enzymes that destroy the extracellular matrix and degrade tight junction proteins, leading to BBB damage, cerebral edema, and HT. MMPs degrade collagens, elastin, fibrillin, laminin, decorin, and other proteoglycans and activate various growth factors. MMP-9 levels are independently associated with the degree of BBB damage. MMP-9 is activated by t-PA administration; therefore, it may be a potential biomarker for predicting the risk of HT after thrombolysis. Patients with blood MMP-9 levels ≥140 ng/mL had a 29.5 times higher risk of symptomatic HT after ischemic stroke, whereas MMP-9 levels in thrombectomy had no significance for HT prognosis [33]. An immunohistochemical study of cellular MMP-9 and MMP-2 expression in neutrophils yielded opposite results, with patients with HT having two times higher MMP-9 expression and proteolytic activity after endovascular treatment. MMP-9 release into stroke-affected brain regions increased the degree of intracerebral hemorrhages and clinical stroke severity after recanalization. Moreover, it independently increased the odds of space-occupying parenchymal hematomas by 1.54 times and the odds of severe disability or death (mRS ≥5 at discharge) by 2.33 times per 1000 ng/mL increase [34].

Tissue inhibitors of metalloproteinases (TIMPs) have a critical role in the pathogenesis of HT. The balance between MMPs and their inhibitors regulates intercellular matrix remodeling, whereas impairments in this balance due to ischemia may cause BBB damage. Experimental studies in mice have demonstrated the role of TIMPs in maintaining BBB integrity [35]. Several studies have found that the balance between MMPs and their inhibitors is associated with HT, and the risk of HT increases with this ratio [36, 37].

Fibronectin is a dimer that consists of two 250 kDa subunits linked by two disulfide bonds. Fibronectin is a critical basement membrane component that is synthesized and secreted by endothelial cells. After basement membrane destruction, fibronectin is released into plasma, resulting in polymorphonuclear leucocyte migration to the vascular injury site. Fibronectin is primarily found in vascular endothelial cells, and its elevated plasma levels can be a highly sensitive marker of BBB damage and HT. The threshold fibronectin level for HT prognosis was ≥3.6 μg/mL [33].

TIGHT JUNCTION PROTEINS AND CELL ADHESION MOLECULES

Occludin is an integral membrane protein. According to clinical studies, serum occludin levels in patients with HT are significantly higher than in those without HT. In early ischemic stroke, MMP-2 promotes fast occludin degradation in the brain microvasculature, resulting in BBB breakdown in vitro and in vivo. Recent findings obtained in animal models and pilot clinical studies indicate an increase in BBB permeability caused by occludin degradation in the microvascular endothelium caused by cerebral ischemia/reperfusion. Blood samples of animals after stroke showed a sharp increase in occludin levels 4.5 h after ischemic stroke, which remained elevated compared to baseline. These findings indicate that serum occludin can be a clinically relevant biomarker for early BBB damage after ischemic stroke. Importantly, blood occludin levels are a highly specific and sensitive marker of BBB damage [25].

An original study by W. Li et al. [38] in 78 patients with stroke and reperfusion therapy (asymptomatic HT in 12 patients and symptomatic HT in 2 patients) found that serum occludin levels in the HT group were significantly higher (5.34 ± 1.36 ng/mL) than in the non-HT group (4.16 ± 1.31 ng/mL) (p = 0.005). The threshold occludin level was 4.48 ng/mL.

Di Biase et al. [39] reported elevated occludin, claudin-5, and ZO-1 levels in patients with HT. Occludin showed 58.6% sensitivity, 67.5% specificity, 12.3% positive predictive value (PPV), and 95.5% negative predictive value (NPV), with an AUC of 0.622. Claudin-5 showed 64.3% sensitivity, 65.8% specificity, 16.8% PPV, and 94.5% NPV, with an AUC of 0.599. ZO-1 showed 56.7% sensitivity, 56.0% specificity, 9.1% PPV, and 94.3% NPV, with an AUC of 0.519. Of note is the high negative predictive value of all three tight junction proteins, i.e., the negative HT prognosis is more accurate.

Li et al. [40] demonstrated the significance of BBB damage markers for HT prognosis. The authors assessed occludin and MMP-9 levels in 86 patients with stroke at admission, after 24 h, and after 7 days. All patients received endovascular reperfusion therapy, and CT was performed after 24 h to assess contrast extravasation. Half of the patients received normobaric oxygen therapy for neuroprotection [40]. This technique effectively maintained the BBB integrity by inhibiting MMP-induced degradation of tight junction proteins, specifically occludin [41]. The group of normobaric oxygen therapy combined with endovascular treatment had significantly lower levels of BBB damage markers after 24 h than the group that received endovascular treatment alone. After 7 days, the combination therapy group also had lower occludin and MMP-9 levels. The incidence of contrast extravasation was lower in the combination therapy group than in the endovascular treatment group. Moreover, serum occludin and MMP-9 levels after 24 h were significantly higher in the contrast extravasation group, confirming the role of these proteins in the pathogenesis of HT [40].

Intercellular and vascular adhesion molecules and VEGF can also serve as potential markers of BBB damage and HT. The intercellular adhesion molecule-1 (ICAM-1) is an adhesion protein involved in inflammatory responses. It is a sensitive but not specific marker of HT [4].

Hypertension, diabetes mellitus, stroke, and cerebral microhemorrhages were more common in patients with high ICAM-1 levels (≥250.0 ng/mL) than in patients with low ICAM-1 levels. High serum ICAM-1 levels in patients with stroke and microhemorrhages were independently associated with a higher risk of HT [42].

GROWTH FACTORS, CYTOKINES, AND NERVE TISSUE PROTEINS

The vascular endothelial growth factor (VEGF) can also be associated with BBB damage after brain ischemia. In a rat thromboembolism model, delayed rt-PA administration 4 h after ischemia promoted VEGF in the BBB, MMP-9 activation, and degradation of BBB components, ultimately resulting in HT. The authors found that HT is inhibited by intravenous administration of a neutralizing anti-VEGF antibody/VEGF receptor antagonist [43].

Nerve tissue-specific S100 proteins belong to a family of acidic calcium-binding proteins with at least 21 S100 isoforms. S100B is primarily expressed by mature astrocytes. Under physiological conditions, S100B cannot cross the BBB; however, when the BBB is damaged, S100B can be released into circulation. Elevated median S100B levels have been reported in patients with symptomatic HT, with a threshold value of ≥0.23 μg/L. However, elevated S100B levels are also found in other brain disorders such as head injuries; therefore, S100B levels are not a specific BBB damage biomarker after ischemic stroke [25, 33].

Researchers have identified two other prognostic markers for HT: the neuron-specific enolase (NSE)—a neurospecific enzyme that only enters systemic circulation when the nerve tissue is damaged—and IL-33. According to a study in 83 patients with stroke, elevated NSE levels in peripheral blood indicate BBB damage and are associated with a higher risk of HT [39]. IL-33 is an independent HT predictor in patients with acute brain ischemia. A study in 151 patients with ischemic stroke found a threshold IL-33 level of 67.66 ng/L, with 81.3% sensitivity, 63% specificity, and an AUC of 0.739 [39].

ENDOGENOUS FIBRINOLYSIS INHIBITORS

Endogenous fibrinolysis inhibitors have a critical role in maintaining the balance between coagulation and thrombolysis; moreover, they can contribute to HT after thrombolytic therapy. A study that assessed plasminogen activator type 1 inhibitor (PAI-1) levels in 77 patients with stroke before rt-PA administration found that patients who developed symptomatic HT had 1.5 times lower baseline PAI-1 levels than patients without HT (р < 0.01). The threshold value was 21.4 ng/mL. In vitro studies confirm that fibrinolytic enzymes interact with microvascular endothelial cells in the brain, influencing BBB integrity via active cell contractions [43]. Low PAI-1 levels can increase fibrinolysis and blood extravasation, promoting the progression of small hemorrhagic lesions to large parenchymatous hematomas. Experiments in mice found that an optimal effect after thrombolysis requires combination therapy that increases endogenous fibrinolysis by inhibiting PAI-1 while reducing the BBB permeability [44].

Thus, MMP-9, fibronectin, S100B, occludin, ICAM-1, IL-33, and PAI-1 are promising specific markers of HT prognosis with well-established sensitivity, specificity, and threshold values (Table 1). These and other specific markers need to be validated as prognostic criteria for HT in large cohort studies.

CONCLUSION

The search is ongoing for an optimal predictor of hemorrhagic transformation after reperfusion therapy in patients with ischemic stroke. This predictor must have high sensitivity and specificity, with an established threshold value. Moreover, it must respond to BBB damage, allow for minimally invasive determination, and be easily available, simple, and quick to use and assess the results. Currently available biomarkers do not fully meet these criteria. According to many researchers, the prognostic value can be improved by using biomarker complexes. This complex assessment may include non-specific parameters, such as NLR, glucose levels, albuminuria, FAR, lipid profile parameters, and NT-proBNP levels, as well as specific BBB damage and neuroinflammation markers, such as MMP-9, fibronectin, S100B, occludin, ICAM-1, IL-33, and PAI-1. The pathogenesis of hemorrhagic transformation and involved proteins is still being investigated. Understanding the pathogenetic mechanisms underlying hemorrhagic transformation will facilitate identifying an early, accurate, and simple predictor. Moreover, it will help select protective therapy and improve stroke outcomes after reperfusion therapy.

ADDITIONAL INFORMATION

Author contributions: O.V. Liang, definition of the concept, visualization, data analysis, writing a draft of the manuscript; M.A. Soldatov, L.V. Klimov, L.B. Zavaliy, T.V. Kiseleva, D.N. Shamalova, I.L. Gubsky, N.A. Marskaya, data analysis, writing a draft of the manuscript; N.A. Shamalov, definition of the concept, revision and editing of the manuscript. Thereby, all authors provided approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding sources: The article was prepared as part of the research project “Creation of a protocol for a comprehensive multimodal examination of patients in the acute period of ischemic stroke, which allows for the identification of predictors of the development of hemorrhagic transformation of the brain lesion, for a personalized approach to reperfusion therapy”, №1024031500085-1-3.2.25, Federal State Budgetary Institution “Federal center of brain research and neurotechnologies” of the Federal Medical Biological Agency, Moscow, Russia.

Disclosure of interests: The authors declare that they have no competing interests.

Statement of originality: This article has not been previously published.

Data availability statement: The authors report that all data is presented in the article and/or its appendices.

Generative AI: No generative AI technology was used to create this article.

About the authors

Olga V. Lyang

Federal Center of Brain Research and Neurotechnologies; Peoples’ Friendship University of Russia

Author for correspondence.
Email: lyang@fccps.ru
ORCID iD: 0000-0002-1023-5490
SPIN-code: 9105-6218

МD, PhD, Assistant Professor

Russian Federation, Moscow; Moscow

Mikhail A. Soldatov

Federal Center of Brain Research and Neurotechnologies; Research Institute for Healthcare Organization and Medical Management

Email: soldatov1477@gmail.com
ORCID iD: 0000-0002-5294-5706
SPIN-code: 7208-6556
Russian Federation, Moscow; Moscow

Leonid V. Klimov

Federal Center of Brain Research and Neurotechnologies

Email: dr.klimov@mail.ru
ORCID iD: 0000-0003-1314-3388
SPIN-code: 5618-0734

MD, PhD

Russian Federation, Moscow

Lesya B. Zavaliy

Federal Center of Brain Research and Neurotechnologies; Sklifosovsky Research Institute for Emergency Medicine

Email: ZavaliyLB@sklif.mos.ru
ORCID iD: 0000-0002-8572-7094
SPIN-code: 6158-5433

MD, PhD

Russian Federation, Moscow; Moscow

Tatiana V. Kiseleva

Federal Center of Brain Research and Neurotechnologies; Research Institute for Healthcare Organization and Medical Management

Email: tatiana-kis17@yandex.ru
ORCID iD: 0000-0003-4913-351X
SPIN-code: 5606-8751
Russian Federation, Moscow; Moscow

Daria N. Shamalova

The Russian National Research Medical University named after N.I. Pirogov

Email: shamalova.daria@mail.ru
ORCID iD: 0009-0006-2906-5117
Russian Federation, Moscow

Ilya L. Gubskiy

The Russian National Research Medical University named after N.I. Pirogov

Email: gubskiy.ilya@gmail.com
ORCID iD: 0000-0003-1726-6801
SPIN-code: 9181-3091

MD, PhD

Russian Federation, Moscow

Nataliya A. Marskaya

Federal Center of Brain Research and Neurotechnologies

Email: marskayana@gmail.com
ORCID iD: 0000-0002-0789-4823
SPIN-code: 5578-2649
Russian Federation, Moscow

Nikolay A. Shamalov

Federal Center of Brain Research and Neurotechnologies; The Russian National Research Medical University named after N.I. Pirogov

Email: shamalovn@gmail.com
ORCID iD: 0000-0001-6250-0762
SPIN-code: 2865-9817

MD, PhD

Russian Federation, Moscow; Moscow

References

Supplementary files