Hematology

Direct bilirubin: A predictor of hematoma expansion after intracerebral hemorrhage

a b s t r a c t

Background: Previous evidence demonstrated that several biomarkers involved in the pathological process of co- agulation/hemostasis dysfunction, impairment of brain vascular integrity and inflammation are associated with Hematoma expansion (HE) after intracerebral hemorrhage (ICH). We aimed to explore whether there were un- reported laboratory biomarkers associated with HE that were readily and commonly available in clinical practice. Methods: We retrospectively analyzed consecutive acute ICH patients from 2012 to 2020 with admission labora- tory tests and baseline and follow-up computed tomography (CT) scans. Univariate and multivariate regression analyses were used to evaluate associations between conventional laboratory indicators and HE. The results were verified in a prospective validation cohort. The relationship of candidate biomarker and 3-month outcomes was also investigated and mediation analysis was undertaken to determine causal associations among candidate bio- marker, HE and outcome.

Results: Of 734 ICH patients, 163 (22.2%) presented HE. Among the included laboratory indicators, higher direct bilirubin (DBil) was associated with HE (adjusted odds ratio [OR] of per 1.0 umol/L change 1.082; 95% confidence interval [CI] 1.011-1.158). DBil >5.65 umol/L was a predictor of HE in validation cohort. Higher DBil was also as- sociated with poor 3-month outcomes. The mediation analysis indicated that the association of higher DBil and poor outcomes was partially mediated by HE.

Conclusions: DBil is a predictor of HE and poor 3-month outcomes after ICH. DBil’s metabolic process and involve- ment in the pathological mechanism of HE are likely to contribute to the association between DBil and HE. Inter- ventions targeting DBil to improve post-ICH prognosis may be meaningful and worthy of further exploration.

(C) 2023 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://

creativecommons.org/licenses/by-nc-nd/4.0/).

  1. Introduction

Intracerebral hemorrhage (ICH) accounts for 10%-20% of all strokes [1,2] and is associated with high risks of mortality and disability [3]. He- matoma expansion (HE) occurs in approximately 1/3 of patients after ICH and is a strong predictor of poor post-ICH outcomes [4,5]. Predicting HE in a timely manner contributes to the early identification of HE and the implementation of targeted interventions, with the possibility of improving prognosis. Therefore, there is a crucial need to identify feasi- ble HE-related biomarkers to achieve better management of ICH.

Abbreviations: DBil, direct bilirubin; GCS, Glasgow Coma Scale; ICH, intracerebral hemorrhage; IVH, intraventricular hemorrhage; mRS, Modified Rankin scale; NIHSS, National Institutes of Health Stroke Scale.

* Corresponding authors at: Department of Neurology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, No 1095 Jiefang Avenue, Qiaok’ou District, Wuhan 430030, Hubei, China.

E-mail addresses: [email protected] (S. Huang), [email protected] (S. Zhu).

Quite a few studies have been dedicated to exploring potential predic- tors of HE to develop possible intervention targets. Among them, the CT angiography (CTA) spot sign is regarded as a robust predictor of HE [6], al- though its availability may be limited to some extent, and individual judg- ment may lead to subjective bias. Biomarkers evaluated by laboratory tests have the advantages of simplicity, convenience and accessibility and may provide beneficial clues to unravel the underlying mechanisms. Some proteins, such as hemoglobin, matrix metalloproteinase-9 (MMP- 9) and C-reactive protein (CRP), involved in the pathways of coagula- tion/hemostasis dysfunction, impairment of brain vascular integrity and inflammation were reported to be relevant to HE [7-13]. In addition, some electrolytes, namely, magnesium and calcium, have been asso- ciated with impaired coagulation status, which could affect HE [14,15]. Whether there are other biomarkers readily and commonly available from laboratory tests that are able to predict HE remains attractive to further study.

The purpose of our study was to investigate the optimal unreported laboratory biomarkers associated with HE that are quick and easy to

https://doi.org/10.1016/j.ajem.2023.06.042

0735-6757/(C) 2023 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

obtain to identify HE in a timely manner and provide evidence for devel- oping new interventions for improving prognosis.

  1. Methods
    1. Patients

We retrospectively extracted relevant data from our prospective co- hort of consecutive patients with ICH treated within 24 h after symptom onset from January 2012 to November 2020 in the neurology depart- ment of the hospital. The diagnosis of ICH was based on the World Health Organization definition of stroke [16] combined with imaging findings of ICH on the computed tomography (CT) scan. Patients who were younger than 18 years, had primary IVH, had missing laboratory tests on admission, and had ICH due to trauma, tumors, or Hemorrhagic transformation of cerebral infarction with or without thrombolytic ther- apy were excluded. To evaluate hematoma expansion, we also excluded patients without baseline CT scans within 6 h of symptom onset and follow-up CT scans within 48 h of baseline.

We also prospectively enrolled consecutive patients with ICH treated from December 2020 to November 2021 in the same depart- ment with baseline CT and laboratory tests within 6 h after symptom onset as the validation cohort. All patients underwent a follow-up CT within 24-48 h from baseline or earlier if the patient clinically deterio- rated or was ready for surgery. Patients who were younger than 18 years, had primary IVH, had mechanical ventilation and were unable to undergo follow-up CT, and had ICH due to trauma, tumors, or hemor- rhagic transformation of cerebral infarction with or without thrombo- lytic therapy were excluded.

The study was approved by the institutional ethics board of Huazhong University of Science and Technology, and informed consent was ob- tained from all participants (Clinical Trial Registration-URL: http://www. chictr.org.cn. Unique identifier: ChiCTR-ROC-2000039365).

    1. Data collection

The following baseline data were obtained for the enrolled patients:

(1) demographic data, including age and sex; (2) past medical and med- ication history, including hypertension, diabetes mellitus, coronary heart disease, atrial fibrillation, anticoagulant agents, and previous stroke, as well as Chronic liver disease, Antiplatelet agents, and statin use in the validation cohort; (3) imaging data, including the location and volume of the hematoma (measured by the ABC/2 method [17]), the presence of IVH or subarachnoid space extension, cerebral angi- ography [CTA, magnetic resonance angiography (MRA), or digital subtraction angiography (DSA)]; and (4) laboratory tests, including biochemistry, routine blood, coagulation function, and infection bio- markers (CRP) (considering clinical practicality, we used well- known conventional laboratory indicators); (5) other parameters, including blood pressure on admission, the pre-ICH modified Rankin Scale (mRS), the National Institutes of Health Stroke Scale score and the Glasgow Coma Scale score on admission; and

(6) therapy, including conservative treatment, ventricular drainage or minimally invasive Hematoma evacuation. Laboratory values were obtained immediately on admission and conducted in the de- partment of laboratory medicine in our hospital. All imaging was conducted in our hospital except for the baseline CT scans of 13 pa- tients in the validation cohort (conducted in other institutions), and all the images were reviewed by two neurologists referring to reports from an expert radiologist. CT image acquisition was per- formed on a GE Discovery CT750 HD by scanning from the base of the skull to the vertex using an axial technique. The CT protocols were as follows: 120 kV, automatic tube current modulation (300 mAs), 5 mm section interval, and 5 mm section thickness. HE was de- fined as absolute growth >12.5 mL or relative growth >33% from baseline to follow-up CT.

The follow-up was conducted by a telephone interview. The evalua- tion of patients’ prognoses was blinded to their clinical data. The clinical outcome was measured by mRS at the 3-month follow-up, and an unfa- vorable outcome was defined as mRS >3.

    1. Statistical analysis

Data are reported as the mean +- SD, median (IQR) for continuous variables, or n (%) for categorical variables. Pearson, ?2 and Fisher exact-tests were used to compare group data for categorical variables. Student’s t-test or the Mann-Whitney U test was applied to analyze con- tinuous variables between two groups. Several continuous variables had missing data [101 (13.8%) for CRP; 10 (1.4%) for glomerular filtration

rate (GFR); 25 (3.4%) for glucose; 65 (8.9%) for magnesium; 49 (6.7%)

for prothrombin time , fibrinogen, activated partial thromboplastin time , thrombin time (TT); 21 (2.9%) for international normal- ized ratio (INR); 48 (6.5%) for triglyceride, high-density lipoprotein (HDL), Low-density lipoprotein ], but the rates of missing data were <15% and were imputed as the mean value of the remaining avail- able data [18]. The variables with P values<0.1 from the univariate anal- ysis and variables previously shown to be predictive of HE were included in the multivariate analysis using Wald forward logistic regres- sion model to evaluate the association between the laboratory indica- tors and HE and then to screen the candidate biomarkers of HE. Then the restricted cubic spline (RCS) and receiver operating characteristic (ROC) analyses were performed for the candidate biomarkers which were significant in both the univariate and multivariate analysis. The RCS model with three knots was used to test the linear association be- tween candidate biomarker levels and HE. ROC analysis was performed to determine the cutoff values of candidate biomarkers for predicting hematoma expansion. The diagnostic value of candidate biomarkers for HE was assessed by calculating the area under the curve (AUC) of the ROC curve in the validation cohort. An additional logistic regression model assessed the association of candidate biomarker levels with poor 3-month clinical outcomes after adjusting for variables with P values<0.1 from the univariate analysis. A mediation analysis was per- formed to estimate whether HE (as the mediator) was the driving factor for any relationship between candidate biomarkers (independent vari- able) and poor outcomes (dependent variable) by regression analysis of all three variables together [19]. Two-sided P values <0.05 were con- sidered statistically significant. All analyses were performed with IBM SPSS software, version 26 (SPSS Inc., Chicago, IL, USA) and R software version 4.1.1 (The R Foundation for Statistical Computing, Vienna, Austria. http://www.r-project.org). We prepared this article using STROBE, which is the guideline for reports of cohort studies.

  1. Results

A total of 1705 patients with primary ICH were screened, and 734 patients were included according to the inclusion and exclusion criteria (male: 66.2%; mean age: 56.3 +- 12.2 years, Fig. 1). There were 163 (22.2%) patients presenting HE. In addition to infratentorial ICH and white blood cells, the other baseline characteristics were similar be- tween the study participants and the excluded patients (Supplementary Table S1).

Intergroup differences between patients with HE and those without are shown in Table 1. Patients with HE had shorter time intervals from symptom onset to the first CT scan, more frequent presence of IVH, lower platelet counts, higher glucose levels and worse NIHSS and GCS scores (P < 0.05). In the multivariate analysis, as shown in Table 2, time from symptom onset to first CT scan, presence of IVH, NIHSS on ad- mission and direct DBil were associated with HE. Therefore, DBil was regarded as a candidate biomarker of HE. The dose-response relation- ship between the DBil level and the risk of HE was further demonstrated with RCS (P for nonlinearity = 0.465, Fig. 2A), and DBil predicted the probability of HE in a linear, dose-dependent relationship (r = 0.243,

Image of Fig. 1

Fig. 1. Flowchart of study patients. ICH, intracerebral hemorrhage.

P < 0.001, Fig. 2B). According to ROC analysis, we used 5.65 umol/L as the cutoff value.

A total of 96 patients were included in the validation cohort (male: 74.0%; mean age: 55.2 +- 12.9 years, Fig. 1). There were 35 (36.5%) pa- tients presenting HE. Intergroup comparisons between patients with HE and those without are shown in Supplementary Table S2. The multi- variate logistic regression model revealed that DBil >5.65 umol/L was an independent predictor of HE (adjusted OR 4.476; 95% CI, 1.404-14.270; P = 0.011 in model 1 and 3.871; 95% CI, 1.283-11.675; P = 0.016 in

model 2, Table 3). In the ROC curve, the AUC was 0.656, with a 95% CI of 0.535-0.776 (P = 0.013, Fig. 3).

A total of 87 patients were lost to 3-month follow-up due to provid- ing the wrong phone number or failure to answer the phone call. We compared the baseline characteristics between the group with com- plete follow-up and the group lost to follow-up. Lobar ICH, deep ICH and the presence of IVH were significantly different between the two groups, while the other characteristics showed no differences (Supple- mentary Table S3). Intergroup differences between patients with favor- able and unfavorable outcomes are shown in Supplementary Table S4. In the evaluation of the relationships between DBil levels and 3- month outcomes, we identified an association of higher DBil levels with increased odds of poor 3-month outcomes (adjusted OR per

1.0 umol/L change 1.145; 95% CI 1.058-1.245; P = 0.001, Supplemen- tary Table S4). When regressing DBil, HE, and outcome together, the mediation analysis revealed that HE significantly mediated 21.69% of the association between admission DBil level and poor outcome (Fig. 4).

  1. Discussion

In this study, we investigated the association between biomarkers from conventional laboratory tests and HE after ICH and found that

DBil was an independent predictor of HE with a dose-response relation- ship. In addition, we verified that the DBil level was associated with a 3- month unfavorable outcome after ICH, and HE appeared to partially mediate this relationship.

It is of great clinical significance to predict and control HE after ICH considering the disastrous effect of HE on the post-ICH progno- sis. Exploring optimal biomarkers is clearly a priority. Previously reported biomarkers indicated that dysregulation of pathways, in- cluding coagulation/hemostasis, cerebrovascular integrity and in- flammation, is critical for HE. Regarding proteins, a low level of fibrinogen and low hemoglobin may induce HE due to an impair- ment of coagulation and hemostasis functions [9,11]. Plasma cellular fibronectin (c-Fn) and MMP-9 were demonstrated to have favorable predictive value for HE resulting in Vascular injury [10,13]. Interleukin-6 (IL-6), CRP and Lactate dehydrogenase related to the inflammatory response were revealed by HE [7,8,10]. Regard- ing electrolytes, low serum levels of magnesium [15] and calcium

[14] have been reported to be associated with HE, with evidence of coagulopathy. Glucose that are involved in carbohydrate and amino acid related metabolic pathways has been found to be associ- ated with HE [20] and bradykinin-mediated hemostasis inhibition was suggested to contribute to the relationship [21]. In our study, al- though not significant in multivariate analysis, patients with HE showed lower magnesium and higher glucose levels than did non- HE patients in univariate analysis (P < 0.05). When we regard hema- toma growth as a continuous variable similar to Liotta EM et al. and Liotta EM et al. and use linear regression to investigate the associa- tion between glucose as well as magnesium and hematoma growth, we found higher glucose was significant associated with greater he- matoma growth in the univariate and multivariate analysis, while lower magnesium was significant in the univariate analysis and

Table 1

Comparison between ICH patients with and without hematoma expansion.

Total

(n = 734)

Non-HE (n = 571)

HE

(n = 163)

P Value

Age, y

56.3 +- 12.2

56.0 +- 12.1

57.2 +- 12.5

0.281

Male sex

486 (66.2%)

372 (65.1%)

114 (69.9%)

0.254

Hypertension

526 (71.7%)

414 (72.5%)

112 (68.7%)

0.343

Diabetes mellitus

73 (9.9%)

54 (9.5%)

19 (11.7%)

0.408

Coronary heart disease

43 (5.9%)

32 (5.6%)

11 (6.7%)

0.583

Atrial fibrillation

6 (0.8%)

4 (0.7%)

2 (1.2%)

0.619

Oral anticoagulation

19 (2.6%)

13 (2.3%)

6 (3.7%)

0.399

Previous stroke

112 (15.3%)

92 (16.1%)

20 (12.3%)

0.229

Smoking habit

270 (36.8%)

206 (36.1%)

64 (39.3%)

0.457

Drinking habit

242 (33.0%)

191 (33.5%)

51 (31.3%)

0.605

Initial ICH volume, mL

18.1 (8-36.1)

17.6 (8.7-33.6)

18.5 (7.9-36.8)

0.734

Lobar ICH

200 (27.2%)

157 (27.5%)

43 (26.4%)

0.778

Deep ICH

551 (75.1%)

424 (74.3%)

127 (77.9%)

0.341

Infratentorial ICH

41 (5.6%)

29 (5.1%)

12 (7.4%)

0.263

Subarachnoid space extension

70 (9.5%)

49 (8.6%)

21 (12.9%)

0.099

Presence of IVH

244 (33.2%)

172 (30.1%)

72 (44.2%)

<0.001

Midline shift

144 (19.6%)

105 (18.4%)

39 (23.9%)

0.116

Time from symptom onset to first CT scan, h

4 (2-6)

4 (2-6)

3 (2-5)

<0.001

Admission SBP, mmHg

156.84 +- 24.65

156.39 +- 24.66

158.39 +- 24.61

0.363

Admission DBP, mmHg

91.39 +- 16.90

91.22 +- 17.03

91.97 +- 16.48

0.619

WBC, 10^9/L

10 (7.6-12.8)

9.9 (7.6-12.5)

10.4 (7.6-13.2)

0.448

Neutrophil, 10^9/L

8.2 (5.7-10.9)

8 (5.7-10.8)

8.7 (5.7-11.6)

0.235

Neutrophil percentage, %

82.3 (74.0-88.3)

82.3 (73.6-88.0)

82.7 (75.0-89.3)

0.074

RBC, 10^12/L

4.7 (4.3-5.0)

4.7 (4.3-5.0)

4.6 (4.3-5.1)

0.844

Hemoglobin, g/L

141 (129-152)

141 (129-151)

141 (130.5-152)

0.464

Platelet count, 10^9/L

199.8 +- 60.9

203 +- 60.6

188.5 +- 61.1

0.007

ALT, U/L

16 (12-23)

16 (12-23)

17 (13-25)

0.301

AST, U/L

20 (16-25)

20 (16-25)

21 (17-25)

0.126

Total Cholesterol, mmol/L

4.4 (3.8-5)

4.4 (3.8-5.1)

4.4 (3.8-4.9)

0.274

Triglyceride, mmol/L

1.1 (0.8-1.5)

1.1 (0.8-1.5)

1.1 (0.8-1.4)

0.851

HDL, mmol/L

1.2 (1-1.4)

1.2 (1-1.4)

1.2 (1-1.4)

0.572

LDL, mmol/L

2.8 (2.2-3.2)

2.8 (2.2-3.3)

2.7 (2.2-3.2)

0.721

Total bilirubin, umol/L

12.9 (8.8-17.0)

12.8 (8.8-16.7)

13.2 (8.9-17.8)

0.528

Direct bilirubin, umol/L

3.7 (2.5-5.1)

3.6 (2.5-4.9)

3.8 (2.7-5.8)

0.087

Albumin, g/L

42.7 (39.9-45.3)

42.7 (39.8-45.4)

42.7 (40.2-45)

0.972

LDH, U/L

211 (180.0-234.0)

212 (180.0-233.5)

208 (176.0-236.5)

0.595

Creatinine, umol/L

72.5 (59.0-89.0)

72 (58.5-89.0)

75 (59.0-91.0)

0.280

GFR, ml/min/1.73 m^2

90.2 (74.9-105.3)

91 (75.5-105.4)

90.1 (74.7-104.8)

0.403

Uric acid, mmol/L

297.1 (225.9-375.8)

296 (228.2-379.0)

300 (221.1-365.0)

0.426

Magnesium, mmol/L

0.83 +- 0.09

0.83 +- 0.09

0.81 +- 0.09

0.069

Calcium, mmol/L

2.2 (2.2-2.3)

2.2 (2.2-2.3)

2.2 (2.2-2.3)

0.568

Glucose, mmol/L

6.7 (5.7-7.8)

6.6 (5.6-7.7)

7.1 (6.0-8.6)

0.003

PT, s

13.5 (12.9-13.9)

13.5 (13.0-13.9)

13.4 (12.9-14.0)

0.435

Fibrinogen, g/L

3.40 (2.80-3.90)

3.40 (2.81-3.80)

3.30 (2.80-3.90)

0.571

APTT, s

34.9 (32.4-37.1)

34.9 (32.5-37.0)

34.9 (31.8-37.1)

0.917

TT, s

16.5 (15.9-17)

16.5 (15.9-17)

16.5 (15.9-16.9)

0.950

INR

1 (1.0-1.1)

1 (1.0-1.1)

1 (1.0-1.1)

0.259

CRP, mg/L

5.3 (1.7-11.7)

5.2 (1.7-11.7)

5.9 (1.6-11.7)

0.609

Pre-ICH mRS

0 (0-0)

0 (0-0)

0 (0-0)

0.209

NIHSS on admission

11 (7-18)

11 (6-16)

14 (10-22)

<0.001

GCS on admission

13 (10-15)

13 (11-15)

13 (9-15)

0.005

ICH indicates intracerebral hemorrhage; IVH, Intraventricular hemorrhage; SBP, systolic blood pressure; DBP, diastolic blood pressure; WBC, White blood cell; RBC, Red blood cell; ALT, Alanine transaminase; AST, Aspartate Aminotransferase; HDL, High density lipoprotein; LDL, Low density lipoprotein; LDH, lactate dehydrogenase; GFR, Glomerular filtration rate; PT, pro- thrombin time; APTT, activated partial thromboplastin time; TT, thrombin time; INR, international normalized ratio; CRP, C-reactive protein; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; GCS, Glasgow Coma Scale.

marginally significant in the multivariate analysis (Supplementary Table S5), which is consistent with previous reports to some extent. A series of prior clinical studies gave strong clues regarding the role of DBil in HE. The association between DBil and Stroke severity as well as poststroke outcomes has been investigated in several studies [22-26]. In particular, in a few studies that examined the relationship between DBil and ICH, Fu K et al. demonstrated the potential Predictive ability of DBil for initial stroke severity and prognosis; namely, higher DBil levels were associated with greater stroke severity at presentation and worse out- comes at discharge [27]. In our study, we found that DBil was associated with HE and could predict the risk of HE in a linear relationship. Our re- sults indicated that even if the DBil level is within the normal range, a level of 1.0 umol/L may increase the risk of HE by approximately 8%, and a DBil level higher than 5.65 umol/L increased the risk by four- to

five fold. Direct bilirubin, as the end metabolic product that comprehen- sively reflects the multiple pathological pathways, metabolizes rapidly and can timely reflect the state of the body, which is a sensitive warning indicator. Our findings expand our understanding of the role of DBil in ICH, especially in the Early prediction of HE.

The metabolic process and pathological effect of DBil per se may also support its potential Predictive role in HE. There is a redox reaction cycle between biliverdin and bilirubin in the metabolism of bilirubin. In this case, bilirubin plays an antioxidant role, that is, bilirubin is traditionally regarded as an antioxidant factor [28]. Acute ICH will stimulate the body to produce an overall inflammatory response, and the production of a large number of oxygen radicals will induce the increase of bilirubin to play the antioxidant role. However, the antioxidant effect of bilirubin is limited to the state of low content [28], and more bilirubin will

Table 2

Multivariate analysis assessing the association between laboratory indicators and hematoma expansion.

Model 1a

Model 2b

P Value OR (95%CI)

P Value OR (95%CI)

Presence of IVH

Time from symptom onset to first CT scan Direct bilirubin

NIHSS on admission INR

0.022 1.553 (1.066-2.261)

0.001 0.867 (0.797-0.944)

0.023 1.082 (1.011-1.158)

0.003 1.028 (1.010-1.046)

0.026 1.535 (1.052-2.240)

0.001 0.866 (0.795-0.943)

0.048 1.072 (1.001-1.149)

0.004 1.027 (1.008-1.045)

0.053 3.934 (0.973-15.908)

IVH indicates intraventricular hemorrhage; NIHSS, National Institutes of Health Stroke Scale; ICH, intracerebral hemorrhage; SBP, systolic blood pressure; LDH, lactate dehydrogenase; INR, international normalized ratio; CRP, C-reactive protein; GCS, Glasgow Coma Scale; OR, odds ratio; CI, confidence interval. a Model 1 including age, sex and variables with P values<0.1 from the univariate analysis (Subarachnoid space extension, Presence of IVH, Time from

symptom onset to first CT scan, Neutrophil percentage, Platelet count, Direct bilirubin, Magnesium, Glucose, NIHSS on admission).

b Model 2 including variables in Model 1 and variables previously shown to be predictive of HE (Oral anticoagulation, Admission SBP, Albumin, LDH, Calcium, Fibrinogen, INR, CRP, GCS on admission).

Image of Fig. 2

Fig. 2. A. Relationship of direct bilirubin level with the risk of hematoma expansion. Adjusted for age, sex, time from symptom onset to first CT scan, subarachnoid space extension, presence of IVH, PLT, glucose, magnesium, neutrophil percentage, NIHSS on admission, GCS on admission. B. Relationship between direct bilirubin level and predicted probability of hematoma expansion.

produce oxidative stress and other harmful effects. Bilirubin in the blood can reach the brain through the blood-brain barrier(BBB) and cause irreversible brain damage through pathological processes such as oxidative stress and activation of cytokines (including MMPs) [29-32]. Oxidative stress can destroy tight-junction proteins, degrade col- lagen and laminin in the basal membrane [33], and the activation of MMP-2 and MMP-9 can damage the integrity of cerebral microvascular

Endothelial cells, thus destroying the integrity of the BBB [34]. In addition, in the early stage of ICH, red blood cells in hematoma can be lysed [5], thereby releasing hemoglobin and further degrading to heme, which is transported to microglia, macrophages, vascular smooth muscle cells, neurons and endothelial cells and metabolized into bilirubin. Oxidative products are produced in these process and cause oxidative stress [35] which can promote inflammation, endoplasmic reticulum

Table 3

Multivariate analysis assessing association of direct bilirubin levels with hematoma expansion in validation cohort.

Model 1a

Model 2b

P Value OR (95%CI)

P Value OR (95%CI)

Age, y Male sex

Direct bilirubin >5.65 umol/L

0.233 1.030 (0.987-1.077)

0.189 2.505 (0.739-9.774)

0.011 4.476 (1.404-14.270)

0.185 1.037 (0.998-1.082)

0.163 1.959 (0.600-6.733)

0.010 5.088 (1.523-18.964)

OR indicates odds ratio; CI, confidence interval.

a Adjusted for age, sex, time from symptom onset to first CT scan, diabetes mellitus, previous stroke, initial ICH volume, deep ICH, NIHSS on admission.

b Adjusted for covariates from model 1 and further adjusted for admission SBP, GCS on admission.

Image of Fig. 3

Fig. 3. receiver operating characteristic curves for prediction of direct bilirubin level for hematoma expansion in the validation cohort. The ROC curve yielded an area under the curve (AUC) value of 0.656 (95% CI 0.535-0.776, P = 0.013).

Image of Fig. 4

Fig. 4. The mediation analysis among direct bilirubin, hematoma expansion and 3-month functional outcome. Direct bilirubin was the independent variable, hematoma expansion was the intermediate variable, and 3-month functional outcome was the dependent var- iable. The total effect c of the independent variable on the dependent variable was 0.0853, the direct effect c’ was 0.0668, and the indirect effect a*b was 0.0185. The 95% con- fidence interval of the indirect effect and direct effect did not include 0, so hematoma expansion played a partial mediating role, accounting for 21.69% of the whole effect.

stress, autophagy, apoptosis and necrosis, leading to the endothelial cell damage and destruction of BBB [36]. In addition, previous studies have found that elevated bilirubin concentration can inhibit Platelet function, coagulation pathways and hemostatic response [37]. The ef- fects of bilirubin on promoting inflammation, damaging vascular in- tegrity, inhibiting coagulation and hemostasis provided evidence for motivating HE. During the above pathological reactions, bilirubin is

supposed to act and cooperate with the reported biomarkers of HE, such as c-Fn [10], MMP-9 [13,38],IL-6 [7,10], CRP [7] and magnesium [15,39], and these associations are not limited to ICH [40-44]. Indeed,

the pathological mechanism of HE is complex, and these pathological processes are likely to be related to each other [45-47].

This study had several limitations. First, it was a single-center study, which can lead to selection bias. However, Tongji Hospital is a national tertiary hospital and has wide coverage and a large number of patients. In addition, internal validation was performed to increase the consis- tency of the results. It is expected that our observations will be validated

at other institutions. Second, the bilirubin data used in our study were collected within 24 h after system onset, and variations may have oc- curred compared with the initial condition. To guarantee reliability, we limited this time to 6 h in the validation cohorts. Third, data on the history of digestive diseases, which may affect blood bilirubin levels, were incomplete in our retrospective analysis. Nevertheless, the inter- group comparisons of liver function parameters on admission showed no significant differences, as did the history of chronic liver disease in the validation cohorts.

In conclusion, DBil is associated with HE after ICH and has predictive value for HE. DBil also acts as a predictor of poor post-ICH outcomes, and the relationship between DBil and outcomes was partially mediated by HE. Weather the decrement of DBil level takes effect in the development of HE warrants further prospective studies. Additional research is needed to explore the potential Treatment interventions targeting DBil to improve prognosis after ICH.

Funding

This work was supported by the Hubei Technological Innovation Spe- cial Fund (CN) [grant number 2019ACA132]; The Fundamental Research Funds for the Central Universities [grant number 2019kfyXKJC075] and The National Key research and development Program of China [grant number 2017YFC1310000].

CRediT authorship contribution statement

Yuchao Jia: Writing – original draft, Methodology, Investigation, Conceptualization. Xiaodong Ye: Investigation. Guini Song: Investiga- tion. Xianxian Li: Investigation. Jiahe Ye: Investigation. Yuyan Yang: Investigation. Kai Lu: Investigation. Shanshan Huang: Writing – review & editing, Project administration, Funding acquisition, Conceptualization. Suiqiang Zhu: Supervision, Project administration, Funding acquisition.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declaration of Competing Interest

None.

Acknowledgments

We thank all the patients and caregivers who participated in the pro- ject and the clinical staff for their support and contribution.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2023.06.042.

References

  1. Sacco S, Ornello R, Degan D, Tiseo C, Pistoia F, Carolei A. Declining incidence of intra- cerebral hemorrhage over two decades in a population-based study. Eur J Neurol. 2016;23(11):1627-34. https://doi.org/10.1111/ene.13099.
  2. Qureshi AI, Mendelow AD, Hanley DF. Intracerebral haemorrhage. Lancet. 2009;373 (9675):1632-44. https://doi.org/10.1016/S0140-6736(09)60371-8.
  3. van Asch CJ, Luitse MJ, Rinkel GJ, van der Tweel I, Algra A, Klijn CJ. Incidence, case fa- tality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: a systematic review and meta-analysis. Lancet Neurol. 2010;9(2):167-76. https://doi.org/10.1016/S1474-4422(09)70340-0.
  4. Davis SM, Broderick J, Hennerici M, Brun NC, Diringer MN, Mayer SA, et al. Hema- toma growth is a determinant of mortality and poor outcome after intracerebral hemorrhage. Neurology. 2006;66(8):1175-81. https://doi.org/10.1212/01.wnl. 0000208408.98482.99.
  5. Xi G, Keep RF, Hoff JT. Mechanisms of brain injury after intracerebral haemorrhage. Lancet Neurol. 2006;5(1):53-63. https://doi.org/10.1016/S1474-4422(05)70283-0.
  6. Wada R, Aviv RI, Fox AJ, Sahlas DJ, Gladstone DJ, Tomlinson G, et al. CT angiography “spot sign” predicts hematoma expansion in acute intracerebral hemorrhage. Stroke. 2007;38(4):1257-62. https://doi.org/10.1161/01.STR.0000259633.59404.f3.
  7. Di Napoli M, Parry-Jones AR, Smith CJ, Hopkins SJ, Slevin M, Masotti L, et al. C- reactive protein predicts hematoma growth in intracerebral hemorrhage. Stroke. 2014;45(1):59-65. https://doi.org/10.1161/STROKEAHA.113.001721.
  8. Chu H, Huang C, Dong J, Yang X, Xiang J, Dong Q, et al. Lactate dehydrogenase pre- dicts early hematoma expansion and poor outcomes in intracerebral hemorrhage patients. Transl Stroke Res. 2019;10(6):620-9. https://doi.org/10.1007/s12975- 019-0686-7.
  9. Roh DJ, Albers DJ, Magid-Bernstein J, Doyle K, Hod E, Eisenberger A, et al. Low hemo- globin and hematoma expansion after intracerebral hemorrhage. Neurology. 2019; 93(4):e372-80. https://doi.org/10.1212/WNL.0000000000007820.
  10. Silva Y, Leira R, Tejada J, Lainez JM, Castillo J, Davalos A, et al. Molecular signatures of vascular injury are associated with early growth of intracerebral hemorrhage. Stroke. 2005;36(1):86-91. https://doi.org/10.1161/01.STR.0000149615.51204.0b.
  11. Fujii Y, Takeuchi S, Sasaki O, Minakawa T, Tanaka R. Multivariate analysis of predic- tors of hematoma enlargement in spontaneous intracerebral hemorrhage. Stroke. 1998;29(6):1160-6. https://doi.org/10.1161/01.str.29.6.1160.
  12. Takahashi H, Urano T, Nagai N, Takada Y, Takada A. Progressive expansion of hyper- tensive intracerebral hemorrhage by coagulopathy. Am J Hematol. 1998;59(2): 110-4. https://doi.org/10.1002/(sici)1096-8652(199810)59:2<110::aid-ajh2>3.0. co;2-0.
  13. Yang Q, Zhuang X, Peng F, Zheng W. Relationship of plasma matrix metalloproteinase-9 and hematoma expansion in acute hypertensive cerebral hem- orrhage. Int J Neurosci. 2016;126(3):213-8. https://doi.org/10.3109/00207454. 2015.1007372.
  14. Morotti A, Charidimou A, Phuah CL, Jessel MJ, Schwab K, Ayres AM, et al. Association between serum calcium level and extent of bleeding in patients with intracerebral hemorrhage. JAMA Neurol. 2016;73(11):1285-90. https://doi.org/10.1001/ jamaneurol.2016.2252.
  15. Liotta EM, Prabhakaran S, Sangha RS, Bush RA, Long AE, Trevick SA, et al. Magnesium, hemostasis, and outcomes in patients with intracerebral hemorrhage. Neurology. 2017;89(8):813-9. https://doi.org/10.1212/WNL.0000000000004249.
  16. Aho K, Harmsen P, Hatano S, Marquardsen J, Smirnov VE, Strasser T. Cerebrovascular disease in the community: results of a WHO collaborative study. Bull World Health Organ. 1980;58(1):113-30. https://www.ncbi.nlm.nih.gov/pubmed/6966542.
  17. Kothari RU, Brott T, Broderick JP, Barsan WG, Sauerbeck LR, Zuccarello M, et al. The ABCs of measuring intracerebral hemorrhage volumes. Stroke. 1996;27(8):1304-5. https://doi.org/10.1161/01.str.27.8.1304.
  18. Barzi F, Woodward M. Imputations of missing values in practice: results from impu- tations of serum cholesterol in 28 cohort studies. Am J Epidemiol. 2004;160(1): 34-45. https://doi.org/10.1093/aje/kwh175.
  19. Preacher KJ, Hayes AF. SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behav Res Methods Instrum Comput. 2004;36(4): 717-31. https://doi.org/10.3758/bf03206553.
  20. Broderick JP, Diringer MN, Hill MD, Brun NC, Mayer SA, Steiner T, et al. Determinants of intracerebral hemorrhage growth: an exploratory analysis. Stroke. 2007;38(3): 1072-5. https://doi.org/10.1161/01.STR.0000258078.35316.30.
  21. Liu J, Gao BB, Clermont AC, Blair P, Chilcote TJ, Sinha S, et al. Hyperglycemia-induced cerebral hematoma expansion is mediated by plasma kallikrein. Nat Med. 2011;17 (2):206-10. https://doi.org/10.1038/nm.2295.
  22. Luo Y, Li J, Zhang J, Xu Y. Elevated bilirubin after acute ischemic stroke linked to the stroke severity. Int J Dev Neurosci. 2013;31(7):634-8. https://doi.org/10.1016/j. ijdevneu.2013.08.002.
  23. Sagheb Asl E, Taheraghdam A, Rahmani F, Javadrashid R, Golzari SEJ, Ghaemian N, et al. Determination of the predictive value of Serum bilirubin in patients with ische- mic stroke: a prospective descriptive analytical study. Adv Pharm Bull. 2018;8(4): 715-9. https://doi.org/10.15171/apb.2018.080.
  24. Ouyang Q, Wang A, Tian X, Zuo Y, Liu Z, Xu Q, et al. Serum bilirubin levels are asso- ciated with poor functional outcomes in patients with acute ischemic stroke or tran- sient ischemic attack. BMC Neurol. 2021;21(1):373. https://doi.org/10.1186/ s12883-021-02398-z.
  25. Peng Q, Bi R, Chen S, Chen J, Li Z, Li J, et al. Predictive value of different bilirubin sub- types for clinical outcomes in patients with acute ischemic stroke receiving throm- bolysis therapy. CNS Neurosci Ther. 2022;28(2):226-36. https://doi.org/10.1111/ cns.13759.
  26. Jian Y, Zhao L, Wang H, Li T, Zhang L, Sun M, et al. Bilirubin: a novel predictor of hem- orrhagic transformation and symptomatic intracranial hemorrhage after Mechanical thrombectomy. Neurol Sci. 2020;41(4):903-9. https://doi.org/10.1007/s10072-019- 04182-x.
  27. Fu K, Garvan CS, Heaton SC, Nagaraja N, Dore S. Association of serum bilirubin with the severity and outcomes of intracerebral hemorrhages. Antioxidants (Basel). 2021;

10(9). https://doi.org/10.3390/antiox10091346.

  1. Jansen T, Hortmann M, Oelze M, Opitz B, Steven S, Schell R, et al. Conversion of bil- iverdin to bilirubin by biliverdin reductase contributes to endothelial cell protection by heme oxygenase-1-evidence for direct and indirect antioxidant actions of biliru- bin. J Mol Cell Cardiol. 2010;49(2):186-95. https://doi.org/10.1016/j.yjmcc.2010.04. 011.
  2. Clark JF, Loftspring M, Wurster WL, Beiler S, Beiler C, Wagner KR, et al. Bilirubin ox- idation products, oxidative stress, and intracerebral hemorrhage. Acta Neurochir Suppl. 2008;105:7-12. https://doi.org/10.1007/978-3-211-09469-3_2.
  3. Palmela I, Cardoso FL, Bernas M, Correia L, Vaz AR, Silva RF, et al. Elevated levels of bilirubin and long-term exposure impair human brain microvascular endothelial cell integrity. Curr Neurovasc Res. 2011;8(2):153-69. https://doi.org/10.2174/ 156720211795495358.
  4. Cardoso FL, Kittel A, Veszelka S, Palmela I, Toth A, Brites D, et al. Exposure to lipo- polysaccharide and/or unconjugated bilirubin impair the integrity and function of brain microvascular endothelial cells. PloS One. 2012;7(5):e35919. https://doi.org/ 10.1371/journal.pone.0035919.
  5. Fujiwara R, Haag M, Schaeffeler E, Nies AT, Zanger UM, Schwab M. Systemic regula- tion of bilirubin homeostasis: potential benefits of hyperbilirubinemia. Hepatology. 2018;67(4):1609-19. https://doi.org/10.1002/hep.29599.
  6. Sun MS, Jin H, Sun X, Huang S, Zhang FL, Guo ZN, et al. Free radical damage in Ischemia-reperfusion injury: an obstacle in acute ischemic stroke after revasculari- zation therapy. Oxid Med Cell Longev. 2018;2018:3804979. https://doi.org/10. 1155/2018/3804979.
  7. Lattanzi S, Di Napoli M, Ricci S, Divani AA. Matrix metalloproteinases in acute intra- cerebral hemorrhage. Neurotherapeutics. 2020;17(2):484-96. https://doi.org/10. 1007/s13311-020-00839-0.
  8. Clark JF, Sharp FR. Bilirubin oxidation products (BOXes) and their role in cerebral va- sospasm after subarachnoid hemorrhage. J Cereb Blood Flow Metab. 2006;26(10): 1223-33. https://doi.org/10.1038/sj.jcbfm.9600280.
  9. Duan X, Wen Z, Shen H, Shen M, Chen G. Intracerebral hemorrhage, oxidative stress, and antioxidant therapy. Oxid Med Cell Longev. 2016;2016:1203285. https://doi. org/10.1155/2016/1203285.
  10. Cho HS, Lee SW, Kim ES, Shin J, Moon SD, Han JH, et al. Serum bilirubin levels are in- versely associated with PAI-1 and fibrinogen in Korean subjects. Atherosclerosis. 2016;244:204-10. https://doi.org/10.1016/j.atherosclerosis.2015.11.008.
  11. Silva SL, Osorio C, Vaz AR, Barateiro A, Falcao AS, Silva RF, et al. Dynamics of neuron- glia interplay upon exposure to unconjugated bilirubin. J Neurochem. 2011;117(3): 412-24. https://doi.org/10.1111/j.1471-4159.2011.07200.x.
  12. Jafari M, Di Napoli M, Lattanzi S, Mayer SA, Bachour S, Bershad EM, et al. Serum mag- nesium level and hematoma expansion in patients with intracerebral hemorrhage. J Neurol Sci. 2019;398:39-44. https://doi.org/10.1016/j.jns.2019.01.027.
  13. Ding Y, Zhao L, Mei H, Huang ZH, Zhang SL. Alterations of biliary biochemical constit- uents and cytokines in infantile hepatitis syndrome. World J Gastroenterol. 2006;12 (43):7038-41. https://doi.org/10.3748/wjg.v12.i43.7038.
  14. Akboga MK, Canpolat U, Sahinarslan A, Alsancak Y, Nurkoc S, Aras D, et al. Associa- tion of serum total bilirubin level with severity of coronary atherosclerosis is linked to systemic inflammation. Atherosclerosis. 2015;240(1):110-4. https://doi.org/10. 1016/j.atherosclerosis.2015.02.051.
  15. Rosolowski M, Oberle V, Ahnert P, Creutz P, Witzenrath M, Kiehntopf M, et al. Dy- namics of cytokines, immune cell counts and disease severity in patients with community-acquired pneumonia – unravelling potential causal relationships. Cyto- kine. 2020;136:155263. https://doi.org/10.1016/j.cyto.2020.155263.
  16. DeMauro SB, Kilpatrick LE, Gerdes JS, Abbasi S. Early inflammatory markers for pre- diction of cholestasis in very-low-birth-weight infants. Neonatology. 2012;102(3): 229-34. https://doi.org/10.1159/000339960.
  17. Hwang HJ, Lee SW, Kim SH. Relationship between bilirubin and C-reactive protein. Clin Chem Lab Med. 2011;49(11):1823-8. https://doi.org/10.1515/CCLM.2011.662.
  18. Yang GY, Betz AL, Chenevert TL, Brunberg JA, Hoff JT. Experimental intracerebral hemorrhage: relationship between Brain edema, blood flow, and blood-brain barrier permeability in rats. J Neurosurg. 1994;81(1):93-102. https://doi.org/10.3171/jns. 1994.81.1.0093.
  19. Ziai WC. Hematology and inflammatory signaling of intracerebral hemorrhage. Stroke. 2013;44(6 Suppl 1):S74-8. https://doi.org/10.1161/STROKEAHA.111. 000662.
  20. Rosenberg GA, Navratil M. Metalloproteinase inhibition blocks edema in intracere- bral hemorrhage in the rat. Neurology. 1997;48(4):921-6. https://doi.org/10.1212/ wnl.48.4.921.