- Email: [email protected]
Comparison of Ultra-Early Hematoma Growth and Common Noncontrast Computed Tomography Features in Predicting Hematoma Enlargement in Patients with Spontaneous Intracerebral Hemorrhage
Journal Pre-proof Comparison of ultra-early hematoma growth and common non-contrast computed tomography features in predicting hematoma enlargement in patients with spontaneous intracerebral hemorrhage Yilan Xiang, MD, Tingting Zhang, MD, Yanxuan Li, MD, Jinjin Liu, MD, Haoli Xu, MD, Wenwen He, MD, Qian Chen, MD, Yunjun Yang, PhD PII:
S1878-8750(19)32483-0
DOI:
https://doi.org/10.1016/j.wneu.2019.09.053
Reference:
WNEU 13352
To appear in:
World Neurosurgery
Received Date: 5 June 2019 Revised Date:
9 September 2019
Accepted Date: 10 September 2019
Please cite this article as: Xiang Y, Zhang T, Li Y, Liu J, Xu H, He W, Chen Q, Yang Y, Comparison of ultra-early hematoma growth and common non-contrast computed tomography features in predicting hematoma enlargement in patients with spontaneous intracerebral hemorrhage, World Neurosurgery (2019), doi: https://doi.org/10.1016/j.wneu.2019.09.053. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Elsevier Inc. All rights reserved.
Table 1. Baseline characteristics of patients with hematoma enlargement and patients without
All(n=920)
Patients
Patients with P
without HE
HE
(n =734 )
(n =186)
Age(year)
61.2±12.1
61.4±12.1
60.8±12.0
0.616
Sex female,n(%)
307(33.37%)
263(35.83%)
44(23.66%)
0.002
Alcohol
295(32.07%)
225(30.65%)
70(37.63%)
0.068
Smoking,n(%)
295(32.07%)
230(31.34%)
65(34.95%)
0.346
Hypertension,n(%)
775(84.24%)
629(85.69%)
146(78.49%)
0.016
Diabetes
108(11.74%)
86(11.72%)
22(11.83%)
0.966
27(3.69%)
15(8.06%)
0.010
36(4.90%)
7(3.77%)
0.510
consumption,n(%)
mellitus,n(%) Previous
brain 42(4.57%)
hemorrhage,n(%) Previous
brain 43(4.67%)
infarction,n(%) GCS on admission
12.04±3.21
12.41±3.01
10.55±3.52
<0.001
Baseline
20.09±14.43
18.69±13.11
25.62±17.77
<0.001
3.17±1.42
2.63±1.31
<0.001
hematoma volume (ml) Time from onset to 3.06±1.42 baseline CT(h) IVH
316(34.02%)
249(33.92%)
67(36.02%)
0.590
Midline shift
3.4±2.5
3.3±2.4
4.0±2.9
<0.001
Black hole sign
131(14.24%)
86(11.72%)
45(24.19%)
<0.001
Blend sign
163(17.72%)
95(12.94%)
68(36.56%)
<0.001
uHG (mL/h)
8.11±7.21
7.2±6.4
11.3±8.8
<0.001
uHG >6.46 ml/h
441(47.93%)
310(42.23%)
131(70.43%)
<0.001
IVH: intraventricular hemorrhage; uHG: ultra-early hematoma growth; HE: hematoma enlargement
Table 2. Multivariate analysis of factors associated with hematoma enlargement Baseline characteristics
OR
95%CI
P
Sex female,n(%)
1.80
1.24-2.61
0.002
GCS on admission
0.84
0.80-0.89
<0.001
Black hole sign
2.41
1.61-3.60
<0.001
Blend sign
3.88
2.68-5.60
<0.001
uHG >6.46ml/h
3.26
2.30-4.61
<0.001
GCS: Glasgow Coma Scale; OR: odds ratio; CI: confidence interval; uHG: ultra-early hematoma growth
Table 3. The connection among the black hole sign, blend sign, and uHG ≤ 6.46 mL/h uHG>6.46mL/h
uHG≤6.46mL/h
p
Black hole sign
91
40
<0.001
Blend sign
112
51
<0.001
uHG: ultra-early hematoma growth
Table 4. Prediction power of the black hole sign, blend sign, and uHG > 6.46 mL/h for hematoma enlargement Variable
AUC
Sensitivity
Specificity(%)
PPV(%)
NPV(%)
(%) Black hole sign
0.562
24.19
88.28
34.35
82.13
Blend sign
0.618
36.56
87.06
41.72
84.41
uHG>6.46mL/h
0.641
70.43
57.77
29.71
88.52
AUC: area under the curve; PPV: positive predictive value; NPV: negative predictive value; uHG: ultra-early hematoma growth
Xiang
Abstract Objective: Ultra-early hematoma growth(uHG), the black hole sign, and the blend sign are common predictors of hematoma enlargement(HE). This study aimed to develop a new diagnostic criterion for predicting HE using uHG and to compare the accuracy of uHG, the black hole sign, and the blend sign in predicting HE in patients with spontaneous intracerebral hemorrhage(sICH). Methods: We retrospectively analyzed data of 920 patients with sICH from August 2013 to January 2018. Receiver operating characteristic curves were plotted to determine the optimal threshold values of uHG to predict HE. The effects of the black hole sign, blend sign, and uHG on HE were assessed using univariate and multivariate logistic regression models, and their prediction accuracies were analyzed using receiver operator analyses. Results: The black hole sign was identified in 131 patients, the blend sign in 163 patients, and uHG> 6.46mL/h in 441 patients. Logistic analysis showed that the black hole sign, blend sign, and uHG> 6.46mL/h were independent predictors of HE. The sensitivity values of uHG> 6.46mL/h, the black hole sign, and the blend sign were 70.43%, 24.19%, and 36.56%, respectively, and specificity values were 57.77%, 88.28%, and 87.06%, respectively. uHG had the greatest area under the curve. The black hole and blend signs were more commonly found in patients with uHG> 6.46mL/h(P < 0.001). Conclusion: uHG> 6.46mL/h was the optimal predictor used for identifying patients at high risk of developing HE. A higher uHG value was associated with an increased prevalence of the black hole and blend signs. 1
Xiang
Introduction Spontaneous intracerebral hemorrhage (sICH), a devastating type of stroke, accounts for 10–15% of all stroke cases and approximately 50% of stroke-related mortality and disability worldwide.1, 2 After sICH, patients have an increased risk of negative outcomes, including stroke and cognitive disorders. Hematoma enlargement (HE) appears in 20-40% of cases and has been reported as an independent predictor of mortality and poor clinical outcomes in patients with sICH.3, 4 Therefore, accurate stratification of HE risk is crucial in order to identify patients with the highest risk of clinical deterioration due to active bleeding and would therefore, be most likely to benefit from anti-enlargement therapies. Most studies reported that the volume of the hematoma on baseline non-contrast computed tomography (NCCT) and the onset-to-imaging time (OTT) are related to HE after sICH.5-7 The scholars have proposed using ultra-early hematoma growth (uHG), which is based on the ratio of hematoma volume on the NCCT to OTT (mL/h), to predict HE, and they have proven its crucial value in predicting HE in sICH patients. Furthermore, they determined that a cutoff value of 4.7 mL/h may be used to identify patients at high risk of HE.8 Recently, NCCT has been reported as the standard research tool for identifying signs of acute sICH, such as the blend sign,9 black hole sign,10 and island sign.11 The black hole sign and blend sign are both related to the density of the hematoma and are reported at low-density areas in the density-heterogeneous hematoma, indicating active bleeding.12, 13 These two signs have not only been shown to be neuroimaging predictors of HE, but are also considered reliable 2
Xiang
factors for predicting clinical outcomes and mortality in patients with sICH.14, 15 However, to date, there are few articles using big data samples to further explore the use of uHG in predicting HE, such as proposing a new cutoff value of uHG to predict HE with greater accuracy and convenience. Moreover, the predictive values of uHG and the common density-related imaging signs have not been compared in a single cohort. However, determining the optimal predictor of HE before developing a reliable predicting scale for clinical practice is essential. Thus, this study aimed to propose a new uHG diagnostic criterion for predicting HE and to compare the accuracy of this new uHG criteria and common NCCT imaging signs in predicting HE.
Materials and Methods Patients We retrospectively reviewed the data of patients with sICH from September 2013 to January 2018 in the First Affiliated Hospital of Wenzhou Medical University. A total of 7742 patients, diagnosed with sICH on baseline NCCT, were initially reviewed. The inclusion criteria were: (1) initial diagnostic NCCT performed within 6 h of the sICH onset; (2) follow-up NCCT performed within 72 h of the sICH onset; and (3) the patient’s age was > 18 years old. The exclusion criteria were: (1) intracerebral hemorrhage secondary to intracranial aneurysm, arteriovenous malformation, moyamoya disease, infarction, or tumor; (2) intracerebral hemorrhage caused by trauma; (3) patients who underwent surgical hematoma evacuation 3
Xiang
before the follow-up NCCT scan; (4) hemorrhagic transformation of a brain infarction; and (5) patient's OTT or previous medical history were missing or unclear. Finally, 920 patients were included in this study. Figure 1 depicts the patient selection flow diagram. This study was approved by the ethics committee of the First Affiliated Hospital of Wenzhou Medical University. Data on patients’ age, gender, hypertension history, history of sICH, previous stroke, blood pressure, smoking, alcohol consumption, Glasgow Coma Scale (GCS) score on admission, and OTT were collected.
Imaging data and analysis The admission and follow-up NCCT scans were performed using standard clinical parameters in the axial plane with a 5-mm section thickness. Imaging features included the black hole sign, blend sign, intraventricular hemorrhage, subarachnoid hemorrhage, midline shift, hematoma volume, and uHG. uHG was defined as the baseline sICH volume divided by OTT. Figure 2 shows typical examples of the black hole sign, and blend sign. Hematoma regions in each slice were manually segmented, and the corresponding hematoma volume was measured after three-dimensional reconstruction of the segmented hematoma regions (Figure 3). All imaging assessments and volume measurements of hematomas were completed at the post-processing working station (version 4.6; GE Healthcare). HE was defined as an increase in hematoma volume of > 33% or > 6 mL on follow-up NCCT scan. Two experienced neuroradiologists, who were blinded to the clinical information of the patients, independently evaluated all the NCCT images. 4
Xiang
Statistical analysis Statistical analyses were performed using the SPSS software (Version 22.0; IBM Corp., Armonk, New York, USA). Continuous variables were analyzed using the t-test, and their values are indicated as the mean ± standard deviation. Categorical variables were presented as frequency with percentage and analyzed by the chi-squared test. For the identification of potential risk factors associated with HE, univariate analyses with relevant parameters were performed. Variables with a P value of less than 0.05 in the univariate analysis were included in the multivariate analysis. Hematoma volume and OTT were also excluded in the analysis because they were components used to identify uHG. Receiver operating characteristic (ROC) curves were used to determine the cutoff value of uHG at which the prognostic difference was the most significant. Accuracy, sensitivity, specificity, positive predictive value, and negative predictive value were calculated to evaluate the performance of the black hole sign, blend sign, and uHG > 6.46 mL/h in predicting HE.
Results Baseline clinical characteristics A total of 920 patients (613 men and 307 women) with sICH were enrolled in our study. The median (range) age of the patients was 61 years (52-70 years). HE was observed in 186 5
Xiang
patients (20.22%) with sICH. The demographic and clinical variables including age, smoking, alcohol drinking, subarachnoid hemorrhage, and intraventricular hemorrhage did not differ significantly between the patients in the HE and non-HE groups. Patients who experienced early HE presented with shorter OTTs and lower GCS scores on admission compared to that in patients that had not an experience of early HE (P < 0.001). Detailed clinical data comparing patients with and without HE are displayed in Table 1.
Imaging characteristics There was a statistically significant difference in uHG between patients in the HE and non-HE groups (11.3 ± 8.8 mL/h vs 7.2 ± 6.4 mL/h, P < 0.001). The cutoff value for uHG associated with HE was 6.46 mL/h, obtained by means of optimum stratification. Of the 920 patients with sICH, 131 (14.24%) presented with the black hole sign, 163 (17.72%) with the blend sign on initial imaging, and 441 (47.93%) with an uHG > 6.46 mL/h (Table 1). The analysis showed that the prevalence of the three factors was significantly higher in patients with HE. Multivariate analysis using 5 independent predictors, gender [odds ratio (OR) = 1.80, P = 0.002], GCS score on admission (OR = 0.84, P < 0.001), the black hole sign (OR = 2.41, P < 0.001), the blend sign (OR = 3.88, P < 0.001), and uHG > 6.46 mL/h (OR = 3.26, P < 0.001), showed that these were independently associated with HE (Table 2). In addition, the black hole sign and blend sign were found more frequently in patients with uHG > 6.46 mL/h (P < 0.001), suggesting that these two signs have a positive correlation with uHG > 6.46 mL/h (Table 3). 6
Xiang
Comparison of predictive value in the black hole sign, blend sign, and uHG > 6.46 mL/h The sensitivity, specificity, and positive and negative predictive values of uHG for predicting HE were 70.43%, 57.77%, 29.71%, and 88.52%, respectively. The sensitivity, specificity, and positive and negative predictive values of the black hole sign for predicting HE were 24.19%, 88.28%, 34.35%, and 82.13%, respectively. The sensitivity, specificity, and positive and negative predictive values of the blend sign for predicting HE were 36.56%, 87.06%, 41.72%, and 84.41% (Table 4). In addition, the black hole sign had the smallest area under the curve (AUC) (0.562), and uHG > 6.46 mL/h had the greatest AUC (0.641), while the blend sign had a moderate AUC (0.618). There was a significant difference in the AUC between the black hole sign and the blend sign (P = 0.018), as well as in the black hole sign and uHG > 6.46 mL/h (P = 0.001). However, no significant difference in the AUC was found between the blend sign and uHG > 6.46 mL/h (P = 0.34; Figure 4).
Discussion This study further validates the value of uHG in predicting HE. Although previous studies have demonstrated the value of uHG in predicting HE, this study proposed a new cutoff value of 6.46 mL/h based on a larger sample size for more convenient prediction of HE. Moreover, we compared the accuracy of uHG, the black hole sign, and the blend sign in predicting HE in sICH patients. The results suggest that uHG > 6.46 mL/h is the optimal predictor of HE. In 7
Xiang
addition, we also found that the prevalence of the black hole sign and blend sign had a positive correlation with uHG > 6.46 mL/h. Early NCCT examinations can often improve the rate of HE detection. OTT was statistically different between the patients in the HE and non-HE groups in our study. However, no clear clinical evidence has indicated its predictive value in HE.16 Meanwhile, the effect of the hematoma volume, an independent predictor of the enlargement of the hematoma, on HE was susceptible to changes in OTTs, which in turn leads to limited prediction accuracy.17 Thus, subsequent studies analyzed the predictive value of hematoma volume combined with OTT. Rodriguez-Luna first proposed uHG as the adjustment of the initial hematoma volume by OTT and showed its value in predicting the prognosis in patients with sICH.8 An increased velocity of change in uHG reflects larger culprit plaque rupture points or fragility of the vessel walls, which may lead to a greater possibility of continuous bleeding, rebleeding, and subsequent HE.18-20 In addition, rapid growth of the hematoma may result in secondary damage to small blood vessels surrounding the hematoma, leading to further growth of the hematoma.21, 22 Rodriguez-Luna found that patients with HE had a higher uHG, and uHG was considered an independent risk factor for predicting HE and a poor 90-day prognosis. Moreover, according to the ROC curve, studies defined 4.7 mL/h as the threshold for predicting the prognosis of HE. When uHG ≥ 4.7 mL/h, the sensitivity, specificity, positive predictive value, and negative predictive value of HE prediction were 73.9%, 57.7%, 76.5%, and 57.3%, respectively.23 In the study by Shoichiro et al., uHG > 5 mL/h was considered to be the optimal threshold for predicting HE, 90-day mortality, and poor clinical outcomes.24 8
Xiang
Similarly, in our study, we found that uHG was strongly positively correlated with early HE. When the uHG cutoff value was 6.46 mL/h, it could best be used to identify a subgroup of patients at high risk of HE, which was different from that previously reported. The discrepancies in uHG cutoff values for predicting HE may be explained by many potential factors. Firstly, the initial hematoma volume varies from between institutions. The average initial volume of hematomas in this study (22.89 mL in the HE group and 15.85 mL in the non-HE group) was greater than that reported in the Rodriguez-Luna study (18.2 mL in the HE group and 8.8 mL in the non-HE group). Secondly, our study used three-dimensional reconstruction of continuous regions of interest to measure the volume of the hematomas, which is more accurate than the volume obtained using the ABC/2 method25. Finally, the ethnic characteristics between patients from the East (in our study) and the West (in the Rodriguez-Luna study) may also contribute to these differences. The potential clinical implications of the association of uHG with HE should be further investigated. Among the 920 patients with sICH, the number of patients with an uHG > 6.46 mL/h was found to be greater than the number of patients with HE (441 vs 186), resulting in a higher sensitivity but low specificity of the numerically represented predictors. Therefore, the combination of other predictive signs and the highly sensitive uHG cutoff value may be beneficial in developing a new predictive criteria with improved specificity and sensitivity. Further studies are required to identify a diagnostic criteria with greater predictive value. In addition, an HE screening protocol with improved sensitivity is urgently needed in clinical practice. Thus, uHG > 6.46 mL/h may serve as a prescreening imaging marker for anti-HE treatment. Moreover, our new criteria may provide the rationale for using this in clinical trials for selecting patients for 9
Xiang
anti-enlargement therapies with a favorable safety profile to capture as many patients as possible who are at risk of HE. Hematoma density heterogeneity has been associated with early HE in several studies.26, 27 The black hole sign and blend sign were both considered special types of density heterogeneity and often increase the probability of HE. Some studies have compared the diagnostic performance of the black hole sign and blend sign. In their analysis, the blend sign and the black hole sign on NCCT were both considered good predictors of HE in patients with sICH. However, the blend sign seemed to have a better predictive power. In addition, the two signs showed similar predictive characteristics with good specificity and poor sensitivity.28, 29 In this study, we obtained similar results for the two density-related imaging signs, and when we compared the two signs and the new uHG criterion in this single cohort, the black hole sign had the highest specificity. The rigorous definition of a clear border and a delta of 28 HU between 2 density regions may explain the improved reliability and subjectivity of the predictive characteristics of the black hole sign. The greater specificity of the two imaging sign used for predicting HE may be preferred in studies targeting HE with hemostatic drugs that may carry thromboembolic risks to reduce the risk of potential harm in patients with a low likelihood of sICH expansion30. Another interesting result was that the prevalence of the black hole sign or blend sign increased with increasing uHG. In a study of 133 patients with acute (< 6 hours) sICH, Rodriguez-Luna et al. reported that uHG was significantly faster in patients with the spot sign.31 Li Qi et al. also demonstrated that uHG had a positive correlation with the blend 10
Xiang
sign.15 These results further validate the value of uHG as a predictor of HE. We hypothesized that the faster the uHG increased, the higher the degree of density heterogeneity of the hematoma. However, this hypothesis should be further tested in future related prospective studies in order to draw a statistically significant conclusion. NCCT predictors are relatively simple to identify in the acute phase of sICH and can enhance the accuracy of HE prediction models. Once patients with a high risk of HE can be identified, targeted therapies can be administered early and can improve clinical outcomes in these patients. Thus, the development of a new HE-predicting criteria, based on NCCT predictors, is needed to improve the accuracy of HE prediction in clinical work. Our results showed that uHG > 6.46 mL/h is a convenient factor that could be included in the predictive model. This study had a large sample size, independent and blinded review of imaging, including three-dimensional measurements of the cerebral hematoma masses, and a rigorous retrospective assessment of patients with acute sICH. However, it is undeniable that there are some limitations in our research. Firstly, data of 920 patients with sICH from a single center were retrospectively evaluated. Thus, some systematic errors may be present in the study. Therefore, we aim to carry out further related research in multiple centers in the future. Secondly, we excluded patients with unclear OTTs due to incomplete clinical records or a lack of time needed to determine the OTT, especially in patient who presented during the night. Patients were also excluded if they had not received a follow-up CT due to rapid clinical deterioration, death, no or little symptoms, or the immediate intervention of surgery. This may have resulted in a selection bias. Thirdly, an explanation regarding the mechanism 11
Xiang
of the raised uHG value in patients with HE was lacking. We aim to explore this in future studies. Fourthly, this study excluded other imaging signs, such as the spot sign, island sign, and satellite sign, that have been should to have a good predictive value in HE. In conclusion, our results provide further insights into the use of uHG in the prediction of HE. We found that uHG > 6.46 mL/h was the optimal predictor for HE in patients with sICH and could be included in future predictive criteria for HE. Furthermore, a higher uHG value was associated with an increased prevalence of the black hole and blend signs. Acknowledgements The authors thank all the patients who participated in this study and also to everyone who made a contribution to make this research possible.
Funding Sources Science and Technology Planning Projects of Wenzhou (Grant No.Y20180112),Health Foundation for Creative Talents in Zhejiang Province,and Project Foundation for the College Young and Middle-aged Academic Leader of Zhejiang Province.
References 1.
Qureshi AI, Mendelow AD, Hanley DF. Intracerebral haemorrhage. Lancet. 2009;373:1632-1644
2.
van Asch CJ, Luitse MJ, Rinkel GJ, van der Tweel I, Algra A, Klijn CJ. Incidence, case fatality, and 12
Xiang
functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: A systematic review and meta-analysis. Lancet Neurol. 2010;9:167-176 3.
Ovesen C, Christensen AF, Krieger DW, Rosenbaum S, Havsteen I, Christensen H. Time course of early postadmission
hematoma
expansion
in
spontaneous
intracerebral
hemorrhage.
Stroke.
2014;45:994-999 4.
Yaghi S, Dibu J, Achi E, Patel A, Samant R, Hinduja A. Hematoma expansion in spontaneous intracerebral hemorrhage: Predictors and outcome. Int J Neurosci. 2014;124:890-893
5.
Broderick JP, Brott TG, Duldner JE, Tomsick T, Huster G. Volume of intracerebral hemorrhage. A powerful and easy-to-use predictor of 30-day mortality. Stroke. 1993;24:987-993
6.
A B, HB B, CD A, TW B, AM A, A V, et al. Predictors of hematoma volume in deep and lobar supratentorial intracerebral hemorrhage.%a falcone gj. JAMA neurology. 2013;70:988-994
7.
Falcone GJ, Biffi A, Brouwers HB, Anderson CD, Battey TW, Ayres AM, et al. Predictors of hematoma volume in deep and lobar supratentorial intracerebral hemorrhage. JAMA Neurol. 2013;70:988-994
8.
D R-L, M R, M R, P C, S P, J P, et al. Ultraearly hematoma growth predicts poor outcome after acute intracerebral hemorrhage. Neurology. 2011;77:1599-1604
9.
Li Q, Zhang G, Huang YJ, Dong MX, Lv FJ, Wei X, et al. Blend sign on computed tomography: Novel and reliable predictor for early hematoma growth in patients with intracerebral hemorrhage. Stroke. 2015;46:2119-2123
10.
Qi Li M, PhD; , Gang Zhang M, Xin Xiong M, Xing-Chen Wang M, Wen-Song Yang M, Ke-Wei Li M, et al. Black hole sign: Novel imaging marker that predicts hematoma growth in patients with intracerebral hemorrhage.%a li q. Stroke. 2016;47:1777-1781
11.
Li Q, Liu QJ, Yang WS, Wang XC, Zhao LB, Xiong X, et al. Island sign: An imaging predictor for early hematoma expansion and poor outcome in patients with intracerebral hemorrhage. Stroke. 2017;48:3019-3025
12.
New PF, Aronow S. Attenuation measurements of whole blood and blood fractions in computed tomography. Radiology. 1976;121:635-640
13.
Schlunk F, Greenberg SM. The pathophysiology of intracerebral hemorrhage formation and expansion.
14.
Li Q, Yang WS, Chen SL, Lv FR, Lv FJ, Hu X, et al. Black hole sign predicts poor outcome in patients with
Transl Stroke Res. 2015;6:257-263 intracerebral hemorrhage. Cerebrovasc Dis. 2018;45:48-53 15.
Li Q, Yang WS, Wang XC, Cao D, Zhu D, Lv FJ, et al. Blend sign predicts poor outcome in patients with intracerebral hemorrhage. Plos One. 2017;12:e0183082
16.
Craig S A. Rapid blood-pressure lowering in patients with acute intracerebral hemorrhage. The New England journal of medicine. 2013;25
17.
Boulouis G, Morotti A, Brouwers HB, Charidimou A, Jessel MJ, Auriel E, et al. Noncontrast computed tomography hypodensities predict poor outcome in intracerebral hemorrhage patients. Stroke. 2016;47:2511-2516
18.
Marti-Fabregas J, Delgado-Mederos R, Granell E, Morenas Rodriguez E, Marin Lahoz J, Dinia L, et al. Microbleed burden and hematoma expansion in acute intracerebral hemorrhage. Eur Neurol. 2013;70:175-178
19.
M L, A A-H, RP G, V N, M S. Relationship between white-matter hyperintensities and hematoma volume and growth in patients with intracerebral hemorrhage. Stroke. 2010;41:34-40
20.
Brouwers HB, Biffi A, McNamara KA, Ayres AM, Valant V, Schwab K, et al. Apolipoprotein e genotype is 13
Xiang
associated
with
ct
angiography
spot
sign
in
lobar
intracerebral
hemorrhage.
Stroke.
2012;43:2120-2125 21.
Fisher CM. Pathological observations in hypertensive cerebral hemorrhage. J Neuropathol Exp Neurol. 1971;30:536-550
22.
Brouwers HB, Greenberg SM. Hematoma expansion following acute intracerebral hemorrhage. Cerebrovasc Dis. 2013;35:195-201
23.
Rodriguez-Luna D, Coscojuela P, Rubiera M, Hill MD, Dowlatshahi D, Aviv RI, et al. Ultraearly
24.
Sato S, Arima H, Hirakawa Y, Heeley E, Delcourt C, Beer R, et al. The speed of ultraearly hematoma
hematoma growth in active intracerebral hemorrhage. Neurology. 2016;87:357-364 growth in acute intracerebral hemorrhage. Neurology. 2014;83:2232-2238 25.
RU K, T B, JP B, WG B, LR S, M Z, et al. The abcs of measuring intracerebral hemorrhage volumes. Stroke. 1996;27:1304-1305
26.
Barras CD, Tress BM, Christensen S, MacGregor L, Collins M, Desmond PM, et al. Density and shape as ct predictors of intracerebral hemorrhage growth. Stroke. 2009;40:1325-1331
27.
Blacquiere D, Demchuk AM, Al-Hazzaa M, Deshpande A, Petrcich W, Aviv RI, et al. Intracerebral hematoma morphologic appearance on noncontrast computed tomography predicts significant hematoma expansion. Stroke. 2015;46:3111-3116
28.
Li RL, Yang MF. A comparative study of the blend sign and the black hole sign on ct as a predictor of hematoma expansion in spontaneous intracerebral hemorrhage. Biosci Trends. 2017;11:682-687
29.
Sporns PB, Schwake M, Kemmling A, Minnerup J, Schwindt W, Niederstadt T, et al. Comparison of spot sign, blend sign and black hole sign for outcome prediction in patients with intracerebral hemorrhage. J Stroke. 2017;19:333-339
30.
Morotti A, Boulouis G, Charidimou A, Schwab K, Kourkoulis C, Anderson CD, et al. Integration of computed tomographic angiography spot sign and noncontrast computed tomographic hypodensities to predict hematoma expansion. Stroke. 2018;49:2067-2073
31.
D. Rodriguez-Luna M, M. Rubiera M, PhD , M. Ribo M, PhD, P. Coscojuela M, S. Pin˜eiro MJP, MD, PhD, M. Hernandez-Guillamon P, et al.. Neurology. 2011
14
Table 1. Baseline characteristics of patients with hematoma enlargement and patients without
All(n=920)
Patients
Patients with P
without HE
HE
(n =734 )
(n =186)
Age(year)
61.2±12.1
61.4±12.1
60.8±12.0
0.616
Sex female,n(%)
307(33.37%)
263(35.83%)
44(23.66%)
0.002
Alcohol
295(32.07%)
225(30.65%)
70(37.63%)
0.068
Smoking,n(%)
295(32.07%)
230(31.34%)
65(34.95%)
0.346
Hypertension,n(%)
775(84.24%)
629(85.69%)
146(78.49%)
0.016
Diabetes
108(11.74%)
86(11.72%)
22(11.83%)
0.966
27(3.69%)
15(8.06%)
0.010
36(4.90%)
7(3.77%)
0.510
consumption,n(%)
mellitus,n(%) Previous
brain 42(4.57%)
hemorrhage,n(%) Previous
brain 43(4.67%)
infarction,n(%) GCS on admission
12.04±3.21
12.41±3.01
10.55±3.52
<0.001
Baseline
20.09±14.43
18.69±13.11
25.62±17.77
<0.001
3.17±1.42
2.63±1.31
<0.001
hematoma volume (ml) Time from onset to 3.06±1.42 baseline CT(h) IVH
316(34.02%)
249(33.92%)
67(36.02%)
0.590
Midline shift
3.4±2.5
3.3±2.4
4.0±2.9
<0.001
Black hole sign
131(14.24%)
86(11.72%)
45(24.19%)
<0.001
Blend sign
163(17.72%)
95(12.94%)
68(36.56%)
<0.001
uHG (mL/h)
8.11±7.21
7.2±6.4
11.3±8.8
<0.001
uHG >6.46 ml/h
441(47.93%)
310(42.23%)
131(70.43%)
<0.001
IVH: intraventricular hemorrhage; uHG: ultra-early hematoma growth; HE: hematoma enlargement
Table 2. Multivariate analysis of factors associated with hematoma enlargement Baseline characteristics
OR
95%CI
P
Sex female,n(%)
1.80
1.24-2.61
0.002
GCS on admission
0.84
0.80-0.89
<0.001
Black hole sign
2.41
1.61-3.60
<0.001
Blend sign
3.88
2.68-5.60
<0.001
uHG >6.46ml/h
3.26
2.30-4.61
<0.001
GCS: Glasgow Coma Scale; OR: odds ratio; CI: confidence interval; uHG: ultra-early hematoma growth
Table 3. The connection among the black hole sign, blend sign, and uHG ≤ 6.46 mL/h uHG>6.46mL/h
uHG≤6.46mL/h
p
Black hole sign
91
40
<0.001
Blend sign
112
51
<0.001
uHG: ultra-early hematoma growth
Table 4. Prediction power of the black hole sign, blend sign, and uHG > 6.46 mL/h for hematoma enlargement Variable
AUC
Sensitivity
Specificity(%)
PPV(%)
NPV(%)
(%) Black hole sign
0.562
24.19
88.28
34.35
82.13
Blend sign
0.618
36.56
87.06
41.72
84.41
uHG>6.46mL/h
0.641
70.43
57.77
29.71
88.52
AUC: area under the curve; PPV: positive predictive value; NPV: negative predictive value; uHG: ultra-early hematoma growth
Abbreviations List AUC area under the curve GCS Glasgow Coma Scale HE hematoma enlargement NCCT non-contrast computed tomography OR odds ratio CI confidence interval OTT onset-to-imaging time ROC receiver operating characteristic sICH spontaneous intracerebral hemorrhage uHG ultra-early hematoma growth IVH intraventricular hemorrhage BHS black hole sign BS blend sign
Disclosures of conflict of interest None.
Figure 1 ICH patients Sep. 2013- Jan. 2018 n = 8374 patients
ICH patients n = 2215 patients
ICH patients n = 1531 patients
ICH patients n = 920 patients
Excluded: Time from symptom onset to initial CT scan >6h or with unclear onset-to-imaging time （ 4343 ） or to follow-up CT scan >72h （ 1816 ） n = 6159 patients Excluded: Hemorrhage caused by cerebral aneurysm （ 28 ） , arteriovenous malformation （ 35 ） , infarction(66), tumor （ 10 ） , and trauma brain injury （ 545 ） n = 684 patients Excluded: Surgical interventions before follow-up CT scan(452); patients taking anticoagulant drugs(25);radiation therapy or chemotherapy （ 5 ）； missing previous medical history(129) n = 611patients
Fig. 1.Flowchart of study patients: ICH indicates intracerebral hemorrhage
S1878-8750(19)32483-0
DOI:
https://doi.org/10.1016/j.wneu.2019.09.053
Reference:
WNEU 13352
To appear in:
World Neurosurgery
Received Date: 5 June 2019 Revised Date:
9 September 2019
Accepted Date: 10 September 2019
Please cite this article as: Xiang Y, Zhang T, Li Y, Liu J, Xu H, He W, Chen Q, Yang Y, Comparison of ultra-early hematoma growth and common non-contrast computed tomography features in predicting hematoma enlargement in patients with spontaneous intracerebral hemorrhage, World Neurosurgery (2019), doi: https://doi.org/10.1016/j.wneu.2019.09.053. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Elsevier Inc. All rights reserved.
Table 1. Baseline characteristics of patients with hematoma enlargement and patients without
All(n=920)
Patients
Patients with P
without HE
HE
(n =734 )
(n =186)
Age(year)
61.2±12.1
61.4±12.1
60.8±12.0
0.616
Sex female,n(%)
307(33.37%)
263(35.83%)
44(23.66%)
0.002
Alcohol
295(32.07%)
225(30.65%)
70(37.63%)
0.068
Smoking,n(%)
295(32.07%)
230(31.34%)
65(34.95%)
0.346
Hypertension,n(%)
775(84.24%)
629(85.69%)
146(78.49%)
0.016
Diabetes
108(11.74%)
86(11.72%)
22(11.83%)
0.966
27(3.69%)
15(8.06%)
0.010
36(4.90%)
7(3.77%)
0.510
consumption,n(%)
mellitus,n(%) Previous
brain 42(4.57%)
hemorrhage,n(%) Previous
brain 43(4.67%)
infarction,n(%) GCS on admission
12.04±3.21
12.41±3.01
10.55±3.52
<0.001
Baseline
20.09±14.43
18.69±13.11
25.62±17.77
<0.001
3.17±1.42
2.63±1.31
<0.001
hematoma volume (ml) Time from onset to 3.06±1.42 baseline CT(h) IVH
316(34.02%)
249(33.92%)
67(36.02%)
0.590
Midline shift
3.4±2.5
3.3±2.4
4.0±2.9
<0.001
Black hole sign
131(14.24%)
86(11.72%)
45(24.19%)
<0.001
Blend sign
163(17.72%)
95(12.94%)
68(36.56%)
<0.001
uHG (mL/h)
8.11±7.21
7.2±6.4
11.3±8.8
<0.001
uHG >6.46 ml/h
441(47.93%)
310(42.23%)
131(70.43%)
<0.001
IVH: intraventricular hemorrhage; uHG: ultra-early hematoma growth; HE: hematoma enlargement
Table 2. Multivariate analysis of factors associated with hematoma enlargement Baseline characteristics
OR
95%CI
P
Sex female,n(%)
1.80
1.24-2.61
0.002
GCS on admission
0.84
0.80-0.89
<0.001
Black hole sign
2.41
1.61-3.60
<0.001
Blend sign
3.88
2.68-5.60
<0.001
uHG >6.46ml/h
3.26
2.30-4.61
<0.001
GCS: Glasgow Coma Scale; OR: odds ratio; CI: confidence interval; uHG: ultra-early hematoma growth
Table 3. The connection among the black hole sign, blend sign, and uHG ≤ 6.46 mL/h uHG>6.46mL/h
uHG≤6.46mL/h
p
Black hole sign
91
40
<0.001
Blend sign
112
51
<0.001
uHG: ultra-early hematoma growth
Table 4. Prediction power of the black hole sign, blend sign, and uHG > 6.46 mL/h for hematoma enlargement Variable
AUC
Sensitivity
Specificity(%)
PPV(%)
NPV(%)
(%) Black hole sign
0.562
24.19
88.28
34.35
82.13
Blend sign
0.618
36.56
87.06
41.72
84.41
uHG>6.46mL/h
0.641
70.43
57.77
29.71
88.52
AUC: area under the curve; PPV: positive predictive value; NPV: negative predictive value; uHG: ultra-early hematoma growth
Xiang
Abstract Objective: Ultra-early hematoma growth(uHG), the black hole sign, and the blend sign are common predictors of hematoma enlargement(HE). This study aimed to develop a new diagnostic criterion for predicting HE using uHG and to compare the accuracy of uHG, the black hole sign, and the blend sign in predicting HE in patients with spontaneous intracerebral hemorrhage(sICH). Methods: We retrospectively analyzed data of 920 patients with sICH from August 2013 to January 2018. Receiver operating characteristic curves were plotted to determine the optimal threshold values of uHG to predict HE. The effects of the black hole sign, blend sign, and uHG on HE were assessed using univariate and multivariate logistic regression models, and their prediction accuracies were analyzed using receiver operator analyses. Results: The black hole sign was identified in 131 patients, the blend sign in 163 patients, and uHG> 6.46mL/h in 441 patients. Logistic analysis showed that the black hole sign, blend sign, and uHG> 6.46mL/h were independent predictors of HE. The sensitivity values of uHG> 6.46mL/h, the black hole sign, and the blend sign were 70.43%, 24.19%, and 36.56%, respectively, and specificity values were 57.77%, 88.28%, and 87.06%, respectively. uHG had the greatest area under the curve. The black hole and blend signs were more commonly found in patients with uHG> 6.46mL/h(P < 0.001). Conclusion: uHG> 6.46mL/h was the optimal predictor used for identifying patients at high risk of developing HE. A higher uHG value was associated with an increased prevalence of the black hole and blend signs. 1
Xiang
Introduction Spontaneous intracerebral hemorrhage (sICH), a devastating type of stroke, accounts for 10–15% of all stroke cases and approximately 50% of stroke-related mortality and disability worldwide.1, 2 After sICH, patients have an increased risk of negative outcomes, including stroke and cognitive disorders. Hematoma enlargement (HE) appears in 20-40% of cases and has been reported as an independent predictor of mortality and poor clinical outcomes in patients with sICH.3, 4 Therefore, accurate stratification of HE risk is crucial in order to identify patients with the highest risk of clinical deterioration due to active bleeding and would therefore, be most likely to benefit from anti-enlargement therapies. Most studies reported that the volume of the hematoma on baseline non-contrast computed tomography (NCCT) and the onset-to-imaging time (OTT) are related to HE after sICH.5-7 The scholars have proposed using ultra-early hematoma growth (uHG), which is based on the ratio of hematoma volume on the NCCT to OTT (mL/h), to predict HE, and they have proven its crucial value in predicting HE in sICH patients. Furthermore, they determined that a cutoff value of 4.7 mL/h may be used to identify patients at high risk of HE.8 Recently, NCCT has been reported as the standard research tool for identifying signs of acute sICH, such as the blend sign,9 black hole sign,10 and island sign.11 The black hole sign and blend sign are both related to the density of the hematoma and are reported at low-density areas in the density-heterogeneous hematoma, indicating active bleeding.12, 13 These two signs have not only been shown to be neuroimaging predictors of HE, but are also considered reliable 2
Xiang
factors for predicting clinical outcomes and mortality in patients with sICH.14, 15 However, to date, there are few articles using big data samples to further explore the use of uHG in predicting HE, such as proposing a new cutoff value of uHG to predict HE with greater accuracy and convenience. Moreover, the predictive values of uHG and the common density-related imaging signs have not been compared in a single cohort. However, determining the optimal predictor of HE before developing a reliable predicting scale for clinical practice is essential. Thus, this study aimed to propose a new uHG diagnostic criterion for predicting HE and to compare the accuracy of this new uHG criteria and common NCCT imaging signs in predicting HE.
Materials and Methods Patients We retrospectively reviewed the data of patients with sICH from September 2013 to January 2018 in the First Affiliated Hospital of Wenzhou Medical University. A total of 7742 patients, diagnosed with sICH on baseline NCCT, were initially reviewed. The inclusion criteria were: (1) initial diagnostic NCCT performed within 6 h of the sICH onset; (2) follow-up NCCT performed within 72 h of the sICH onset; and (3) the patient’s age was > 18 years old. The exclusion criteria were: (1) intracerebral hemorrhage secondary to intracranial aneurysm, arteriovenous malformation, moyamoya disease, infarction, or tumor; (2) intracerebral hemorrhage caused by trauma; (3) patients who underwent surgical hematoma evacuation 3
Xiang
before the follow-up NCCT scan; (4) hemorrhagic transformation of a brain infarction; and (5) patient's OTT or previous medical history were missing or unclear. Finally, 920 patients were included in this study. Figure 1 depicts the patient selection flow diagram. This study was approved by the ethics committee of the First Affiliated Hospital of Wenzhou Medical University. Data on patients’ age, gender, hypertension history, history of sICH, previous stroke, blood pressure, smoking, alcohol consumption, Glasgow Coma Scale (GCS) score on admission, and OTT were collected.
Imaging data and analysis The admission and follow-up NCCT scans were performed using standard clinical parameters in the axial plane with a 5-mm section thickness. Imaging features included the black hole sign, blend sign, intraventricular hemorrhage, subarachnoid hemorrhage, midline shift, hematoma volume, and uHG. uHG was defined as the baseline sICH volume divided by OTT. Figure 2 shows typical examples of the black hole sign, and blend sign. Hematoma regions in each slice were manually segmented, and the corresponding hematoma volume was measured after three-dimensional reconstruction of the segmented hematoma regions (Figure 3). All imaging assessments and volume measurements of hematomas were completed at the post-processing working station (version 4.6; GE Healthcare). HE was defined as an increase in hematoma volume of > 33% or > 6 mL on follow-up NCCT scan. Two experienced neuroradiologists, who were blinded to the clinical information of the patients, independently evaluated all the NCCT images. 4
Xiang
Statistical analysis Statistical analyses were performed using the SPSS software (Version 22.0; IBM Corp., Armonk, New York, USA). Continuous variables were analyzed using the t-test, and their values are indicated as the mean ± standard deviation. Categorical variables were presented as frequency with percentage and analyzed by the chi-squared test. For the identification of potential risk factors associated with HE, univariate analyses with relevant parameters were performed. Variables with a P value of less than 0.05 in the univariate analysis were included in the multivariate analysis. Hematoma volume and OTT were also excluded in the analysis because they were components used to identify uHG. Receiver operating characteristic (ROC) curves were used to determine the cutoff value of uHG at which the prognostic difference was the most significant. Accuracy, sensitivity, specificity, positive predictive value, and negative predictive value were calculated to evaluate the performance of the black hole sign, blend sign, and uHG > 6.46 mL/h in predicting HE.
Results Baseline clinical characteristics A total of 920 patients (613 men and 307 women) with sICH were enrolled in our study. The median (range) age of the patients was 61 years (52-70 years). HE was observed in 186 5
Xiang
patients (20.22%) with sICH. The demographic and clinical variables including age, smoking, alcohol drinking, subarachnoid hemorrhage, and intraventricular hemorrhage did not differ significantly between the patients in the HE and non-HE groups. Patients who experienced early HE presented with shorter OTTs and lower GCS scores on admission compared to that in patients that had not an experience of early HE (P < 0.001). Detailed clinical data comparing patients with and without HE are displayed in Table 1.
Imaging characteristics There was a statistically significant difference in uHG between patients in the HE and non-HE groups (11.3 ± 8.8 mL/h vs 7.2 ± 6.4 mL/h, P < 0.001). The cutoff value for uHG associated with HE was 6.46 mL/h, obtained by means of optimum stratification. Of the 920 patients with sICH, 131 (14.24%) presented with the black hole sign, 163 (17.72%) with the blend sign on initial imaging, and 441 (47.93%) with an uHG > 6.46 mL/h (Table 1). The analysis showed that the prevalence of the three factors was significantly higher in patients with HE. Multivariate analysis using 5 independent predictors, gender [odds ratio (OR) = 1.80, P = 0.002], GCS score on admission (OR = 0.84, P < 0.001), the black hole sign (OR = 2.41, P < 0.001), the blend sign (OR = 3.88, P < 0.001), and uHG > 6.46 mL/h (OR = 3.26, P < 0.001), showed that these were independently associated with HE (Table 2). In addition, the black hole sign and blend sign were found more frequently in patients with uHG > 6.46 mL/h (P < 0.001), suggesting that these two signs have a positive correlation with uHG > 6.46 mL/h (Table 3). 6
Xiang
Comparison of predictive value in the black hole sign, blend sign, and uHG > 6.46 mL/h The sensitivity, specificity, and positive and negative predictive values of uHG for predicting HE were 70.43%, 57.77%, 29.71%, and 88.52%, respectively. The sensitivity, specificity, and positive and negative predictive values of the black hole sign for predicting HE were 24.19%, 88.28%, 34.35%, and 82.13%, respectively. The sensitivity, specificity, and positive and negative predictive values of the blend sign for predicting HE were 36.56%, 87.06%, 41.72%, and 84.41% (Table 4). In addition, the black hole sign had the smallest area under the curve (AUC) (0.562), and uHG > 6.46 mL/h had the greatest AUC (0.641), while the blend sign had a moderate AUC (0.618). There was a significant difference in the AUC between the black hole sign and the blend sign (P = 0.018), as well as in the black hole sign and uHG > 6.46 mL/h (P = 0.001). However, no significant difference in the AUC was found between the blend sign and uHG > 6.46 mL/h (P = 0.34; Figure 4).
Discussion This study further validates the value of uHG in predicting HE. Although previous studies have demonstrated the value of uHG in predicting HE, this study proposed a new cutoff value of 6.46 mL/h based on a larger sample size for more convenient prediction of HE. Moreover, we compared the accuracy of uHG, the black hole sign, and the blend sign in predicting HE in sICH patients. The results suggest that uHG > 6.46 mL/h is the optimal predictor of HE. In 7
Xiang
addition, we also found that the prevalence of the black hole sign and blend sign had a positive correlation with uHG > 6.46 mL/h. Early NCCT examinations can often improve the rate of HE detection. OTT was statistically different between the patients in the HE and non-HE groups in our study. However, no clear clinical evidence has indicated its predictive value in HE.16 Meanwhile, the effect of the hematoma volume, an independent predictor of the enlargement of the hematoma, on HE was susceptible to changes in OTTs, which in turn leads to limited prediction accuracy.17 Thus, subsequent studies analyzed the predictive value of hematoma volume combined with OTT. Rodriguez-Luna first proposed uHG as the adjustment of the initial hematoma volume by OTT and showed its value in predicting the prognosis in patients with sICH.8 An increased velocity of change in uHG reflects larger culprit plaque rupture points or fragility of the vessel walls, which may lead to a greater possibility of continuous bleeding, rebleeding, and subsequent HE.18-20 In addition, rapid growth of the hematoma may result in secondary damage to small blood vessels surrounding the hematoma, leading to further growth of the hematoma.21, 22 Rodriguez-Luna found that patients with HE had a higher uHG, and uHG was considered an independent risk factor for predicting HE and a poor 90-day prognosis. Moreover, according to the ROC curve, studies defined 4.7 mL/h as the threshold for predicting the prognosis of HE. When uHG ≥ 4.7 mL/h, the sensitivity, specificity, positive predictive value, and negative predictive value of HE prediction were 73.9%, 57.7%, 76.5%, and 57.3%, respectively.23 In the study by Shoichiro et al., uHG > 5 mL/h was considered to be the optimal threshold for predicting HE, 90-day mortality, and poor clinical outcomes.24 8
Xiang
Similarly, in our study, we found that uHG was strongly positively correlated with early HE. When the uHG cutoff value was 6.46 mL/h, it could best be used to identify a subgroup of patients at high risk of HE, which was different from that previously reported. The discrepancies in uHG cutoff values for predicting HE may be explained by many potential factors. Firstly, the initial hematoma volume varies from between institutions. The average initial volume of hematomas in this study (22.89 mL in the HE group and 15.85 mL in the non-HE group) was greater than that reported in the Rodriguez-Luna study (18.2 mL in the HE group and 8.8 mL in the non-HE group). Secondly, our study used three-dimensional reconstruction of continuous regions of interest to measure the volume of the hematomas, which is more accurate than the volume obtained using the ABC/2 method25. Finally, the ethnic characteristics between patients from the East (in our study) and the West (in the Rodriguez-Luna study) may also contribute to these differences. The potential clinical implications of the association of uHG with HE should be further investigated. Among the 920 patients with sICH, the number of patients with an uHG > 6.46 mL/h was found to be greater than the number of patients with HE (441 vs 186), resulting in a higher sensitivity but low specificity of the numerically represented predictors. Therefore, the combination of other predictive signs and the highly sensitive uHG cutoff value may be beneficial in developing a new predictive criteria with improved specificity and sensitivity. Further studies are required to identify a diagnostic criteria with greater predictive value. In addition, an HE screening protocol with improved sensitivity is urgently needed in clinical practice. Thus, uHG > 6.46 mL/h may serve as a prescreening imaging marker for anti-HE treatment. Moreover, our new criteria may provide the rationale for using this in clinical trials for selecting patients for 9
Xiang
anti-enlargement therapies with a favorable safety profile to capture as many patients as possible who are at risk of HE. Hematoma density heterogeneity has been associated with early HE in several studies.26, 27 The black hole sign and blend sign were both considered special types of density heterogeneity and often increase the probability of HE. Some studies have compared the diagnostic performance of the black hole sign and blend sign. In their analysis, the blend sign and the black hole sign on NCCT were both considered good predictors of HE in patients with sICH. However, the blend sign seemed to have a better predictive power. In addition, the two signs showed similar predictive characteristics with good specificity and poor sensitivity.28, 29 In this study, we obtained similar results for the two density-related imaging signs, and when we compared the two signs and the new uHG criterion in this single cohort, the black hole sign had the highest specificity. The rigorous definition of a clear border and a delta of 28 HU between 2 density regions may explain the improved reliability and subjectivity of the predictive characteristics of the black hole sign. The greater specificity of the two imaging sign used for predicting HE may be preferred in studies targeting HE with hemostatic drugs that may carry thromboembolic risks to reduce the risk of potential harm in patients with a low likelihood of sICH expansion30. Another interesting result was that the prevalence of the black hole sign or blend sign increased with increasing uHG. In a study of 133 patients with acute (< 6 hours) sICH, Rodriguez-Luna et al. reported that uHG was significantly faster in patients with the spot sign.31 Li Qi et al. also demonstrated that uHG had a positive correlation with the blend 10
Xiang
sign.15 These results further validate the value of uHG as a predictor of HE. We hypothesized that the faster the uHG increased, the higher the degree of density heterogeneity of the hematoma. However, this hypothesis should be further tested in future related prospective studies in order to draw a statistically significant conclusion. NCCT predictors are relatively simple to identify in the acute phase of sICH and can enhance the accuracy of HE prediction models. Once patients with a high risk of HE can be identified, targeted therapies can be administered early and can improve clinical outcomes in these patients. Thus, the development of a new HE-predicting criteria, based on NCCT predictors, is needed to improve the accuracy of HE prediction in clinical work. Our results showed that uHG > 6.46 mL/h is a convenient factor that could be included in the predictive model. This study had a large sample size, independent and blinded review of imaging, including three-dimensional measurements of the cerebral hematoma masses, and a rigorous retrospective assessment of patients with acute sICH. However, it is undeniable that there are some limitations in our research. Firstly, data of 920 patients with sICH from a single center were retrospectively evaluated. Thus, some systematic errors may be present in the study. Therefore, we aim to carry out further related research in multiple centers in the future. Secondly, we excluded patients with unclear OTTs due to incomplete clinical records or a lack of time needed to determine the OTT, especially in patient who presented during the night. Patients were also excluded if they had not received a follow-up CT due to rapid clinical deterioration, death, no or little symptoms, or the immediate intervention of surgery. This may have resulted in a selection bias. Thirdly, an explanation regarding the mechanism 11
Xiang
of the raised uHG value in patients with HE was lacking. We aim to explore this in future studies. Fourthly, this study excluded other imaging signs, such as the spot sign, island sign, and satellite sign, that have been should to have a good predictive value in HE. In conclusion, our results provide further insights into the use of uHG in the prediction of HE. We found that uHG > 6.46 mL/h was the optimal predictor for HE in patients with sICH and could be included in future predictive criteria for HE. Furthermore, a higher uHG value was associated with an increased prevalence of the black hole and blend signs. Acknowledgements The authors thank all the patients who participated in this study and also to everyone who made a contribution to make this research possible.
Funding Sources Science and Technology Planning Projects of Wenzhou (Grant No.Y20180112),Health Foundation for Creative Talents in Zhejiang Province,and Project Foundation for the College Young and Middle-aged Academic Leader of Zhejiang Province.
References 1.
Qureshi AI, Mendelow AD, Hanley DF. Intracerebral haemorrhage. Lancet. 2009;373:1632-1644
2.
van Asch CJ, Luitse MJ, Rinkel GJ, van der Tweel I, Algra A, Klijn CJ. Incidence, case fatality, and 12
Xiang
functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: A systematic review and meta-analysis. Lancet Neurol. 2010;9:167-176 3.
Ovesen C, Christensen AF, Krieger DW, Rosenbaum S, Havsteen I, Christensen H. Time course of early postadmission
hematoma
expansion
in
spontaneous
intracerebral
hemorrhage.
Stroke.
2014;45:994-999 4.
Yaghi S, Dibu J, Achi E, Patel A, Samant R, Hinduja A. Hematoma expansion in spontaneous intracerebral hemorrhage: Predictors and outcome. Int J Neurosci. 2014;124:890-893
5.
Broderick JP, Brott TG, Duldner JE, Tomsick T, Huster G. Volume of intracerebral hemorrhage. A powerful and easy-to-use predictor of 30-day mortality. Stroke. 1993;24:987-993
6.
A B, HB B, CD A, TW B, AM A, A V, et al. Predictors of hematoma volume in deep and lobar supratentorial intracerebral hemorrhage.%a falcone gj. JAMA neurology. 2013;70:988-994
7.
Falcone GJ, Biffi A, Brouwers HB, Anderson CD, Battey TW, Ayres AM, et al. Predictors of hematoma volume in deep and lobar supratentorial intracerebral hemorrhage. JAMA Neurol. 2013;70:988-994
8.
D R-L, M R, M R, P C, S P, J P, et al. Ultraearly hematoma growth predicts poor outcome after acute intracerebral hemorrhage. Neurology. 2011;77:1599-1604
9.
Li Q, Zhang G, Huang YJ, Dong MX, Lv FJ, Wei X, et al. Blend sign on computed tomography: Novel and reliable predictor for early hematoma growth in patients with intracerebral hemorrhage. Stroke. 2015;46:2119-2123
10.
Qi Li M, PhD; , Gang Zhang M, Xin Xiong M, Xing-Chen Wang M, Wen-Song Yang M, Ke-Wei Li M, et al. Black hole sign: Novel imaging marker that predicts hematoma growth in patients with intracerebral hemorrhage.%a li q. Stroke. 2016;47:1777-1781
11.
Li Q, Liu QJ, Yang WS, Wang XC, Zhao LB, Xiong X, et al. Island sign: An imaging predictor for early hematoma expansion and poor outcome in patients with intracerebral hemorrhage. Stroke. 2017;48:3019-3025
12.
New PF, Aronow S. Attenuation measurements of whole blood and blood fractions in computed tomography. Radiology. 1976;121:635-640
13.
Schlunk F, Greenberg SM. The pathophysiology of intracerebral hemorrhage formation and expansion.
14.
Li Q, Yang WS, Chen SL, Lv FR, Lv FJ, Hu X, et al. Black hole sign predicts poor outcome in patients with
Transl Stroke Res. 2015;6:257-263 intracerebral hemorrhage. Cerebrovasc Dis. 2018;45:48-53 15.
Li Q, Yang WS, Wang XC, Cao D, Zhu D, Lv FJ, et al. Blend sign predicts poor outcome in patients with intracerebral hemorrhage. Plos One. 2017;12:e0183082
16.
Craig S A. Rapid blood-pressure lowering in patients with acute intracerebral hemorrhage. The New England journal of medicine. 2013;25
17.
Boulouis G, Morotti A, Brouwers HB, Charidimou A, Jessel MJ, Auriel E, et al. Noncontrast computed tomography hypodensities predict poor outcome in intracerebral hemorrhage patients. Stroke. 2016;47:2511-2516
18.
Marti-Fabregas J, Delgado-Mederos R, Granell E, Morenas Rodriguez E, Marin Lahoz J, Dinia L, et al. Microbleed burden and hematoma expansion in acute intracerebral hemorrhage. Eur Neurol. 2013;70:175-178
19.
M L, A A-H, RP G, V N, M S. Relationship between white-matter hyperintensities and hematoma volume and growth in patients with intracerebral hemorrhage. Stroke. 2010;41:34-40
20.
Brouwers HB, Biffi A, McNamara KA, Ayres AM, Valant V, Schwab K, et al. Apolipoprotein e genotype is 13
Xiang
associated
with
ct
angiography
spot
sign
in
lobar
intracerebral
hemorrhage.
Stroke.
2012;43:2120-2125 21.
Fisher CM. Pathological observations in hypertensive cerebral hemorrhage. J Neuropathol Exp Neurol. 1971;30:536-550
22.
Brouwers HB, Greenberg SM. Hematoma expansion following acute intracerebral hemorrhage. Cerebrovasc Dis. 2013;35:195-201
23.
Rodriguez-Luna D, Coscojuela P, Rubiera M, Hill MD, Dowlatshahi D, Aviv RI, et al. Ultraearly
24.
Sato S, Arima H, Hirakawa Y, Heeley E, Delcourt C, Beer R, et al. The speed of ultraearly hematoma
hematoma growth in active intracerebral hemorrhage. Neurology. 2016;87:357-364 growth in acute intracerebral hemorrhage. Neurology. 2014;83:2232-2238 25.
RU K, T B, JP B, WG B, LR S, M Z, et al. The abcs of measuring intracerebral hemorrhage volumes. Stroke. 1996;27:1304-1305
26.
Barras CD, Tress BM, Christensen S, MacGregor L, Collins M, Desmond PM, et al. Density and shape as ct predictors of intracerebral hemorrhage growth. Stroke. 2009;40:1325-1331
27.
Blacquiere D, Demchuk AM, Al-Hazzaa M, Deshpande A, Petrcich W, Aviv RI, et al. Intracerebral hematoma morphologic appearance on noncontrast computed tomography predicts significant hematoma expansion. Stroke. 2015;46:3111-3116
28.
Li RL, Yang MF. A comparative study of the blend sign and the black hole sign on ct as a predictor of hematoma expansion in spontaneous intracerebral hemorrhage. Biosci Trends. 2017;11:682-687
29.
Sporns PB, Schwake M, Kemmling A, Minnerup J, Schwindt W, Niederstadt T, et al. Comparison of spot sign, blend sign and black hole sign for outcome prediction in patients with intracerebral hemorrhage. J Stroke. 2017;19:333-339
30.
Morotti A, Boulouis G, Charidimou A, Schwab K, Kourkoulis C, Anderson CD, et al. Integration of computed tomographic angiography spot sign and noncontrast computed tomographic hypodensities to predict hematoma expansion. Stroke. 2018;49:2067-2073
31.
D. Rodriguez-Luna M, M. Rubiera M, PhD , M. Ribo M, PhD, P. Coscojuela M, S. Pin˜eiro MJP, MD, PhD, M. Hernandez-Guillamon P, et al.
14
Table 1. Baseline characteristics of patients with hematoma enlargement and patients without
All(n=920)
Patients
Patients with P
without HE
HE
(n =734 )
(n =186)
Age(year)
61.2±12.1
61.4±12.1
60.8±12.0
0.616
Sex female,n(%)
307(33.37%)
263(35.83%)
44(23.66%)
0.002
Alcohol
295(32.07%)
225(30.65%)
70(37.63%)
0.068
Smoking,n(%)
295(32.07%)
230(31.34%)
65(34.95%)
0.346
Hypertension,n(%)
775(84.24%)
629(85.69%)
146(78.49%)
0.016
Diabetes
108(11.74%)
86(11.72%)
22(11.83%)
0.966
27(3.69%)
15(8.06%)
0.010
36(4.90%)
7(3.77%)
0.510
consumption,n(%)
mellitus,n(%) Previous
brain 42(4.57%)
hemorrhage,n(%) Previous
brain 43(4.67%)
infarction,n(%) GCS on admission
12.04±3.21
12.41±3.01
10.55±3.52
<0.001
Baseline
20.09±14.43
18.69±13.11
25.62±17.77
<0.001
3.17±1.42
2.63±1.31
<0.001
hematoma volume (ml) Time from onset to 3.06±1.42 baseline CT(h) IVH
316(34.02%)
249(33.92%)
67(36.02%)
0.590
Midline shift
3.4±2.5
3.3±2.4
4.0±2.9
<0.001
Black hole sign
131(14.24%)
86(11.72%)
45(24.19%)
<0.001
Blend sign
163(17.72%)
95(12.94%)
68(36.56%)
<0.001
uHG (mL/h)
8.11±7.21
7.2±6.4
11.3±8.8
<0.001
uHG >6.46 ml/h
441(47.93%)
310(42.23%)
131(70.43%)
<0.001
IVH: intraventricular hemorrhage; uHG: ultra-early hematoma growth; HE: hematoma enlargement
Table 2. Multivariate analysis of factors associated with hematoma enlargement Baseline characteristics
OR
95%CI
P
Sex female,n(%)
1.80
1.24-2.61
0.002
GCS on admission
0.84
0.80-0.89
<0.001
Black hole sign
2.41
1.61-3.60
<0.001
Blend sign
3.88
2.68-5.60
<0.001
uHG >6.46ml/h
3.26
2.30-4.61
<0.001
GCS: Glasgow Coma Scale; OR: odds ratio; CI: confidence interval; uHG: ultra-early hematoma growth
Table 3. The connection among the black hole sign, blend sign, and uHG ≤ 6.46 mL/h uHG>6.46mL/h
uHG≤6.46mL/h
p
Black hole sign
91
40
<0.001
Blend sign
112
51
<0.001
uHG: ultra-early hematoma growth
Table 4. Prediction power of the black hole sign, blend sign, and uHG > 6.46 mL/h for hematoma enlargement Variable
AUC
Sensitivity
Specificity(%)
PPV(%)
NPV(%)
(%) Black hole sign
0.562
24.19
88.28
34.35
82.13
Blend sign
0.618
36.56
87.06
41.72
84.41
uHG>6.46mL/h
0.641
70.43
57.77
29.71
88.52
AUC: area under the curve; PPV: positive predictive value; NPV: negative predictive value; uHG: ultra-early hematoma growth
Abbreviations List AUC area under the curve GCS Glasgow Coma Scale HE hematoma enlargement NCCT non-contrast computed tomography OR odds ratio CI confidence interval OTT onset-to-imaging time ROC receiver operating characteristic sICH spontaneous intracerebral hemorrhage uHG ultra-early hematoma growth IVH intraventricular hemorrhage BHS black hole sign BS blend sign
Disclosures of conflict of interest None.
Figure 1 ICH patients Sep. 2013- Jan. 2018 n = 8374 patients
ICH patients n = 2215 patients
ICH patients n = 1531 patients
ICH patients n = 920 patients
Excluded: Time from symptom onset to initial CT scan >6h or with unclear onset-to-imaging time （ 4343 ） or to follow-up CT scan >72h （ 1816 ） n = 6159 patients Excluded: Hemorrhage caused by cerebral aneurysm （ 28 ） , arteriovenous malformation （ 35 ） , infarction(66), tumor （ 10 ） , and trauma brain injury （ 545 ） n = 684 patients Excluded: Surgical interventions before follow-up CT scan(452); patients taking anticoagulant drugs(25);radiation therapy or chemotherapy （ 5 ）； missing previous medical history(129) n = 611patients
Fig. 1.Flowchart of study patients: ICH indicates intracerebral hemorrhage