The Use of Serum Matrix Metalloproteinases in Cerebral Amyloid Angiopathy-Related Intracerebral Hemorrhage and Cognitive Impairment

Abstract

Background:

Neuroimaging has played a primary role in predicting intracerebral hemorrhage (ICH) recurrence of cerebral amyloid angiopathy (CAA); however, the utilities of biomarkers in CAA-related ICH and cognitive impairment remain unexplored.

Objective:

To investigate the correlations of serum levels of matrix metalloproteinase-2 (MMP-2), MMP-3, and MMP-9 with CAA-related MRI markers, ICH recurrence, and cognitive status.

Methods:

68 cases with first probable CAA-ICH and 69 controls were recruited. Clinical and imaging data were obtained at baseline and serum MMPs in the acute phase were measured by Luminex multiplex assays. Cognitive status was assessed with the Chinese version of Mini-Mental State Examination within 10–14 days after ICH onset.

Results:

Serum MMP-2 level was significantly lower in CAA-ICH patients than controls while MMP-9 was significantly higher. In CAA-ICH patients, MMP-3 level was significantly associated with lobar cerebral microbleeds count after adjusting age, sex, and hypertension (adjusted coefficient 0.368, 95%CI 0.099–0.637, p = 0.008). During a median follow-up of 2.4 years, higher level of MMP-2 predicted lower CAA-ICH recurrence after adjusting age (adjusted HR 0.326, 95%CI 0.122–0.871, p = 0.025), ICH volume (adjusted HR 0.259, 95%CI 0.094–0.715, p = 0.009), total MRI burden of SVD score (adjusted HR 0.350, 95%CI 0.131–0.936, p = 0.037) respectively. Besides, higher level of MMP-2 was significantly associated with decreased risk of cognitive impairment independent of age and ICH volume (adjusted OR 0.054, 95%CI 0.005–0.570, p = 0.015).

Conclusion:

Serum MMP-2 in acute phase might be a promising biomarker to predict CAA-ICH recurrence and to evaluate the risk of cognitive impairment.

Keywords

Cerebral amyloid angiopathy cognitive impairment cerebral microbleeds intracerebral hemorrhage matrix metalloproteinases

INTRODUCTION

Cerebral amyloid angiopathy (CAA), defined by the accumulation of amyloid in the walls of leptomeningeal and cortical vessels, is one of the main causes of both intracerebral hemorrhage (ICH) and cognitive impairment in the elderly. In addition to its high incidence in Alzheimer’s disease (AD), CAA is one of the main causes of primary nontraumatic lobar hemorrhage in the elderly, accounting for 20%of all spontaneous ICH in patients over the age of 70 [1]. The rate of ICH recurrence in CAA patients is much higher, about 6.9–10%per year [2, 3], compared with deep ICH, about 3%per year [4], which often leads to disability and mortality.

Deposition of amyloid-β results in the degeneration of the tunica media and loss of cerebrovascular smooth muscle and endothelial cells. However, this is not sufficient for cerebrovascular basement membrane breakdown, which suggests that additional molecular mechanisms are likely involved in the vessel rupture. In this regard, proteolytic systems may participate in the regulation of cerebrovascular integrity.

Matrix metalloproteinases (MMPs) are a large family of zinc-dependent endopeptidase enzymes that degrade almost all of the components composing connective tissue and basement membrane. Therefore, MMPs are considered to play an important role in the homeostasis of vascular integrity and extracellular matrix [5]. In the central nervous system (CNS), MMPs are involved in the breakdown of blood-brain barrier, cerebral hemorrhage, demyelination, axonal injury, and activation of neuroinflammation [6]. They have been implicated in various CNS disorders, including stroke, multiple sclerosis, amyotrophic lateral sclerosis, and AD [7].

Previous studies on MMPs in CAA patients mainly focused on their expression patterns in brain tissues and associations with intracerebral hemorrhage [5 , 9]. However, their roles in CAA-related ICH and cognitive function remain unclear. In this study, we investigated serum levels of MMP-2, MMP-3, and MMP-9 in CAA patients and analyzed correlations of MMPs with CAA-related magnetic resonance imaging (MRI) markers, ICH recurrence, and cognitive status. By this means, we tried to shed light on the role of MMPs in the prognostic prediction for CAA-ICH patients.

MATERIALS AND METHODS

Study population

68 cases with first probable CAA-ICH admitted at Huashan Hospital Fudan University (Shanghai, China) from January 2014 to September 2018 as well as First Affiliated Hospital of University of Science and Technology of China (Hefei, China) from January 2018 to September 2018 were recruited. During the same period, sex- and age-matched healthy controls from participants in Shanghai Aging Study [10] were collected. The inclusion criteria and study flow chart were reported previously [11]. The detailed clinical and demographic characteristics were collected at baseline. All patients were regularly followed up every six months until the ICH recurrence, death, or October 2019. ICH recurrence was defined as a new symptomatic lobar ICH event confirmed by corresponding lesion on computed tomography (CT) or susceptibility weighted imaging (SWI) scan during follow-up period. The median (interquartile range, IQR) follow-up time was 2.4 (1.3, 4.0) years and the median (IQR) time of recurrence was 1.0 (0.6, 3.3) year. Two patients died when ICH recurred, and no patient was lost during follow-up. The study was approved by local ethics committees and all the participants (or their surrogate) were consented.

Blood samples and biomarkers measurement

Blood collection was done within 24 h since admission. Serum samples were obtained by centrifuging at 3,000 rpm for 15 min and kept at –80°C until analysis. The measurements of biomarkers were performed by a team blinded to the clinical data. Quantification of serum biomarkers was using Luminex assay kits (R&D Systems Inc., Minneapolis, MN, USA). Briefly, after centrifuging and vortexing, premixed beads (50μL/well) were added to 96-well plates. Then, standards, blanks, and diluted and duplicated serum samples (1:1 for MMP-3 and 1:19 for MMP-2 and MMP-9) (50μL/well) were added and incubated for 2 h on a plate shaker (800 rpm). After washing for 3 times, the biotin-antibody cocktail (50μL/well) was added and incubated for 1 h (800 rpm). Repeated washing for 3 times, the diluted streptavidin-PE (50μL/well) was added and incubated for 30 min (800 rpm). Beads were washed for 3 times and resuspended with washing buffer. The plate was shaken for 2 min and analyzed with Luminex 200 system (Luminex Corporation, TX, USA).

MRI protocols and imaging analysis

Siemens 3.0T MRI was used to scan CAA patients. The parameters of T1-weighted, T2-weighted, fluid attenuated inversion recovery (FLAIR) and SWI were shown in our previous study [11]. Lobar cerebral microbleeds (CMBs) count, white matter hyperintensities (WMH) Fazekas score, cerebral superficial siderosis (cSS) score, and centrum semiovale perivascular space (CSO-PVS) score were visually evaluated by two experienced raters (J.F. and J.X.) independently [11]. The total MRI burden of SVD was calculated by accumulating CMBs number, WMH severity, cSS presence, and extent and CSO-PVS severity, ranging from 0 to 6 points [12]. ICH volume was quantified by 1/2^*(length^*width^*height) on CT scan. WMH volume was measured by using ITK-SNAP software with semi-automatic segmentation approach [13]. WMH threshold was defined as: average signal intensity of brain parenchymal +3 standard deviations. All images were manually inspected and WMH volume was calculated after adjusting the whole brain volume.

Assessment of cognitive impairment

Global cognitive status was assessed using a modified Chinese version of the Mini-Mental State Examination (MMSE) [14] within 10–14 days after ICH onset. Because MMSE scores are strongly correlated with education level [15], we used education-based cut-off points of MMSE, which are widely accepted and used in China, to define cognitive impairment: < 20 for participants without formal education (illiteracy),<23 for those with 1–6 years of education (primary school), and < 27 for those with more than 6 years of education (middle school and higher) [16 –18].

Statistical analysis

Descriptive data were expressed as mean± standard deviation (SD) for normally distributed continuous variables or median with IQR for non-normally distributed continuous variables. Comparisons were used one-way analysis of variance or Mann-Whitney test for continuous variables and χ² test or Fisher exact test for categorical variables.

We used univariable and multivariable linear regression models to estimate the associations between blood biomarkers and WMH volume and CMBs counts (with age, sex, and hypertension as covariables). Coefficient with 95%confidence intervals (CI) were reported. As for cSS score, CSO-PVS score as well as total MRI burden of SVD score, ordinal regression models were used. Odds ratio (OR) with 95%CI were reported.

Then, we explored the ability of MMPs in predicting ICH recurrence and used univariable and multivariable Cox regression analyses with forward selection to calculate hazard ratio (HR) with 95%CI. X-tile software was used to determine the optimal cut-off values for creating dichotomous variables [19]. Proportional hazard assumption was checked using Schoenfeld residual test and the cumulative hazard plot.

Finally, we used binary logistic regression models to investigate the effect of neuroimaging factors and blood biomarkers on cognitive impairment. MMPs were changed into dichotomous variable by the optimal cut-off values generated from receiver operating characteristic curve. Multivariable regression analyses with forward selection were used to calculate OR with 95%CI.

Statistical analyses were performed using SPSS 22.0 software (SPSS Inc., Chicago, IL, USA) and differences were considered significant if two-tailed p < 0.05.

RESULTS

Serum MMPs levels in healthy controls and CAA-ICH patients

68 cases with first probable CAA-ICH from a Chinese prospective cohort and 69 healthy controls were recruited. Demographic characteristics, clinical and neuroimaging features and serum MMPs levels were shown in Table 1. Compared to healthy controls, cases with CAA-ICH showed significantly decreased MMP-2 (median [IQR], 182796.6 [172364.1, 192363.4] versus 167868.0 [158537.5, 180195.3] pg/mL, p < 0.001), and significantly elevated MMP-9 (median [IQR], 35383.3 [26698.6, 58324.0] versus 58183.5 [34663.3, 106735.3] pg/mL, p = 0.002), with no difference in MMP-3 (median [IQR], 16437.9 [8787.3, 21036.9] versus 14541.8 [7842.1, 20193.9] pg/mL, p = 0.290). In CAA patients, serum MMP-2 level was negatively correlated with MMP-9 (unadjusted coefficient –0.332, 95%CI [–0.563, –0.100], p = 0.006), while MMP-3 was positively correlated with MMP-9 (unadjusted coefficient 0.361, 95%CI 0.132–0.591, p = 0.002) (Supplementary Table 1).

Table 1

Clinical and imaging data and serum MMPs levels of healthy controls and first-ever probable CAA-ICH patients (subdivided by lobar ICH recurrence)

Variables	Healthy controls	First-ever probable CAA-ICH	p	CAA with lobar ICH recurrence		p
	(n = 69)	(n = 68)		No (n = 49)	Yes (n = 19)
Clinical and imaging factors
Age of onset, y	67.9±7.1	70.1±9.1	0.122	68.5±9.5	74.1±6.9	0.024
Sex, male (%)	47 (68.1)	50 (73.5)	0.486	35 (71.4)	15 (78.9)	0.528
Hypertension (%)	32 (46.4)	48 (70.6)	0.004	36 (73.5)	12 (63.2)	0.402
Diabetes mellitus (%)	12 (17.4)	15 (22.1)	0.492	11 (22.4)	4 (21.1)	>0.999
Dyslipidemia (%)	13 (18.8)	7 (10.3)	0.157	7 (14.3)	0	0.178
Smoking (%)	24 (34.8)	16 (23.5)	0.147	11 (22.4)	5 (26.3)	0.757
Antiplatelet drug use (%)	6 (8.7)	21 (30.9)	0.001	16 (32.7)	5 (26.3)	0.612
Anticoagulant drug use (%)	1 (1.4)	4 (5.9)	0.208	3 (6.1)	1 (5.3)	>0.999
Pre-existing dementia (%)	0	10 (14.7)	0.001	4 (8.2)	6 (31.6)	0.023
ICH volume (mL)	0	14.3 (5.0, 29.8)	<0.001	9.6 (5.0, 26.2)	25.0 (11.7, 50.6)	0.008
Lobar CMBs count	0 (0, 0)	13.0 (5, 65)	<0.001	22 (5, 73.5)	10 (5, 43)	0.480
WMH Fazekas score	2 (2, 3)	4 (4, 6)	<0.001	4 (4, 5.5)	4 (4, 6)	0.927
cSS score	0 (0, 0)	0 (0, 1)	<0.001	0 (0, 1)	1 (0, 2)	0.085
CSO-PVS score	1 (1, 2.5)	2 (2, 3)	0.010	2 (1.5, 2.5)	3 (2, 3)	0.038
Total MRI burden of SVD score	1 (0, 2)	3.5 (3, 4)	<0.001	3 (3, 4)	4 (4, 5)	0.001
Blood biomarkers
MMP-2 (pg/mL)	182796.6 (172364.1, 192363.4)	167868.0 (158537.5, 180195.3)	<0.001	169808.9±18162.6	160619.3±24090.4	0.093
MMP-3 (pg/mL)	16437.9 (8787.3, 21036.9)	14541.8 (7842.1, 20193.9)	0.290	14230.7 (7615.5, 18507.4)	15166.8 (8425.4, 27208.6)	0.561
MMP-9 (pg/mL)	35383.3 (26698.6, 58324.0)	58183.5 (34663.3, 106735.3)	0.002	52304.4 (31342.1, 92042.0)	77762.6 (38183.0, 192440.0)	0.108

CAA, cerebral amyloid angiopathy; ICH, intracerebral hemorrhage; MMP, matrix metalloproteinase, CMBs, cerebral microbleeds; cSS, cerebral superficial siderosis; WMH, white matter hyperintensities; CSO-PVS, centrum semiovale perivascular space; SVD, small vessel disease.

Associations of serum MMPs levels and lobar CMBs count in CAA patients

Linear regression analysis was performed to ascertain associations between serum MMPs levels and lobar CMBs count. In Table 2, MMP-3 serum level was significantly associated with lobar CMBs count after adjusted for age, sex, and hypertension (adjusted coefficient 0.368, 95%CI 0.099–0.637, p = 0.008). Figure 1 showed significantly positive correlation of MMP-3 serum level with lobar CMBs count. However, the associations of MMP-2, MMP-9 serum levels, and lobar CMBs count did not reach statistical significance. Besides, serum MMPs levels were not significantly correlated with other neuroimaging markers, including ICH volume, WMH volume, cSS score, CSO-PVS score, and total MRI burden of SVD score (data were not shown).

Table 2

Associations of serum MMPs levels and lobar CMBs count in CAA patients

Variables ^*	Unadjusted Coefficient (95%CI	p	Multivariable adjusted Coefficient^# (95%CI)	p
Lobar CMBs count and
MMP-2	–0.009 (–0.255–0.237)	0.941
MMP-3	0.316 (0.083–0.549)	0.009	0.368 (0.099-0.637)	0.008
MMP-9	0.059 (–0.187–0.304)	0.635

CAA, cerebral amyloid angiopathy; MMP, matrix metalloproteinase; CMBs, cerebral microbleeds. ^*The serum levels of MMPs and lobar CMBs count were continuous and Log-transformed. ^#Adjusted for age, sex, and hypertension.

Fig. 1

Significantly positive correlation of MMP-3 serum levels with lobar CMBs count in CAA patients. Data were shown as β (standardized coefficient) and p value from univariable linear regression. Region surrounded by dotted lines indicate linear regression line with 95%confidence interval.

Serum MMPs levels in predicting CAA-ICH recurrence

Then, we used Cox regression model to explore the predictive ability of potential clinical and imaging risk factors as well as MMPs for ICH recurrence. As shown in Table 3, age (HR = 1.057 per 1-year increase, 95%CI 1.005–1.112, p = 0.032), ICH volume (HR = 1.017 per 1 mL increase, 95%CI 1.003–1.031, p = 0.014), CSO-PVS score (≥3 versus≤2, HR 4.451, 95%CI 1.666–11.890, p = 0.003), total MRI burden of SVD score (≥4 versus≤3, HR 6.446, 95%CI 1.857–22.379, p = 0.003), MMP-2 (Log [MMP-2]≥5.217 versus < 5.217, HR 0.326, 95%CI 0.122–0.871, p = 0.025), MMP-9 (Log [MMP-9]≥5.000 versus < 5.000, HR 3.520, 95%CI 1.392–8.903, p = 0.008) were significantly correlated with CAA-ICH recurrence in the univariable analyses. Since there were only 19 ICH recurrences in our cohort, a maximum of two predictors were allowed in the multivariable analyses. MMP-2 was still significantly associated with CAA-ICH recurrence after adjusting age (adjusted HR 0.326, 95%CI 0.122–0.871, p = 0.025), ICH volume (adjusted HR 0.259, 95%CI 0.094–0.715, p = 0.009), CSO-PVS score (adjusted HR 0.327, 95%CI 0.122–0.875, p = 0.026), total MRI burden of SVD score (adjusted HR 0.350, 95%CI 0.131–0.936, p = 0.037) with forward selection respectively. However, MMP-9 did not reach the significance after adjustment for other confounders. In Kaplan–Meier analysis, cases with higher serum MMP-2 level (Log [MMP-2]≥5.217) had longer ICH-free survival time than those below 5.217 (log-rank test, p = 0.017, Fig. 2).

Table 3

Cox regression models for ICH recurrence after first-ever probable CAA-ICH

Variables ^*	Univariable analyses		Age-adjusted		ICH volume- adjusted		CSO-PVS score- adjusted		SVD score- adjusted
	HR (95%CI)	p	HR (95%CI)	p	HR (95%CI)	p	HR (95%CI)	p	HR (95%CI)	p
Clinical and imaging risk factors for CAA-ICH recurrence
Age (per 1-year increase)	1.057 (1.005–1.112)	0.032
Sex (female versus male)	1.271 (0.417–3.871)	0.673
Hypertension (yes versus no)	0.774 (0.300–1.996)	0.596
Pre-existing dementia (yes versus no)	2.190 (0.813–5.901)	0.121
ICH volume (per 1 mL increase)	1.017 (1.003–1.031)	0.014
Lobar CMBs count (per 1-CMB increase)	0.991 (0.978–1.004)	0.174
WMH volume (per 1 cm³ increase)	1.005 (0.995–1.016)	0.329
cSS score (= 2 versus≤1)	2.147 (0.829–5.562)	0.116
CSO-PVS score (≥3 versus≤2)	4.451 (1.666–11.890)	0.003
Total MRI burden of SVD score (≥4 versus≤3)	6.446 (1.857–22.379)	0.003
Blood biomarkers for CAA-ICH recurrence
MMP-2 (≥5.217 versus < 5.217)	0.326 (0.122–0.871)	0.025	0.326 (0.122–0.871)	0.025	0.259 (0.094–0.715)	0.009	0.327 (0.122–0.875)	0.026	0.350 (0.131–0.936)	0.037
MMP-3 (≥4.380 versus < 4.380)	2.487 (0.936–6.606)	0.068
MMP-9 (≥5.000 versus < 5.000)	3.520 (1.392–8.903)	0.008	3.520 (1.392–8.903)	0.008	3.119 (1.214–8.010)	0.018	–	–	–	–

Fig. 2

Kaplan–Meier curve showing the ICH-free survival probability in CAA-ICH cases with serum MMP-2 after Log-transformed above versus below 5.217. MMP, matrix metalloproteinase; CAA, cerebral amyloid angiopathy; ICH, intracerebral hemorrhage.

Associations of serum MMPs levels and cognitive status in CAA patients

Cognitive status was assessed in 47 CAA-ICH patients by MMSE within 10–14 days after ICH onset. According to the education-based cut-off points of MMSE, 31 patients were considered as cognitive impairment. The demographic data, imaging characteristics and MMPs levels according to cognitive status were summarized in Table 4. Since the majority of hemorrhages localized on the posterior side in our study, ICH location was classified as left, right, or bilateral. We used binary logistic regression analysis to find out whether these factors were significantly associated with cognitive impairment in univariable models. As shown in Table 5, ICH volume was a significant risk factor for cognitive impairment; however, other neuroimaging markers, including lobar CMBs count, WMH volume, cSS score, CSO-PVS score, and total MRI burden of SVD score, were not. In addition, MMPs were not significant factors associated with cognitive function when analyzed as continuous variables. However, when MMPs were changed into dichotomous variable, higher level of MMP-2 was a protective factor for cognitive impairment (Log [MMP-2]≥5.241 versus < 5.241, OR 0.187, 95%CI 0.049–0.706, p = 0.013). Since there were only 31 cases considered as cognitive impairment, a maximum of three predictors were allowed in the multivariable analyses. Higher level of MMP-2 was still significant after adjustment for age and ICH volume (adjusted OR 0.054, 95%CI 0.005–0.570, p = 0.015), pre-existing dementia and ICH volume (adjusted OR 0.054, 95%CI 0.005–0.570, p = 0.015), and ICH location and ICH volume (adjusted OR 0.054, 95%CI 0.005–0.570, p = 0.015).

Table 4

The demographic data, clinical characteristics and MMPs levels of cases with probable CAA-ICH (subdivided according to cognitive impairment)

Variables	CAA with cognitive impairment		p
	No (n = 16)	Yes (n = 31)
Clinical and imaging factors
Age of onset, y	68.0±8.2	70.3±10.5	0.451
Sex, male (%)	11 (68.8)	24 (77.4)	0.725
Hypertension (%)	14 (87.5)	21 (67.7)	0.176
Diabetes mellitus (%)	6 (37.5)	6 (19.4)	0.289
Dyslipidemia (%)	4 (25.0)	1 (3.2)	0.040
Smoking (%)	6 (37.5)	7 (22.6)	0.318
Antiplatelet drug use (%)	6 (37.5)	9 (29.0)	0.555
Anticoagulant drug use (%)	1 (6.3)	2 (6.5)	>0.999
Pre-existing dementia (%)	2 (12.5)	6 (19.4)	0.697
Education level			0.620
Illiteracy (%)	0 (0)	2 (6.5)
Primary school (%)	4 (25.0)	5 (16.1)
Middle school and higher (%)	12 (75.0)	24 (77.4)
ICH location			0.571
Left (%)	8 (50)	20 (64.5)
Right (%)	6 (37.5)	9 (29.0)
Bilateral (%)	2 (12.5)	2 (6.5)
ICH volume (mL)	5.0 (3.1, 9.2)	27.4 (9.0, 50.0)	<0.001
Lobar CMBs count	21.5 (4.5, 82)	11.0 (5.0, 56.0)	0.830
WMH volume (cm³)	40.82 (14.18, 50.67)	30.12 (20.46, 40.91)	0.728
cSS score	0 (0, 0)	0 (0, 2)	0.063
CSO-PVS score	2 (2, 3)	2 (2, 3)	0.943
Total MRI burden of SVD score	3 (3, 4)	3 (3, 5)	0.075
Blood biomarkers
MMP-2 (pg/mL)	173451.7±12971.3	167984.7±15344.2	0.230
MMP-3 (pg/mL)	13342.2 (7641.2, 16857.2)	16202.8 (8297.0, 22043.7)	0.312
MMP-9 (pg/mL)	47142.1 (31480.5, 79784.2)	59602.8 (32794.5, 166155.0)	0.252

Table 5

Logistic regression models for cognitive impairment of first-ever probable CAA-ICH

Variables ^*	Univariable analyses		Age- and ICH volume-adjusted		Pre-existing dementia- and ICH volume-adjusted		ICH location- and ICH volume-adjusted
	OR (95%CI)	p	OR (95%CI)	p	OR (95%CI)	p	OR (95%CI)	p
Clinical and imaging factors
Age (per 1-year increase)	1.025 (0.962–1.092)	0.442
Sex (female versus male)	1.558 (0.403–6.020)	0.520
Hypertension (yes versus no)	0.300 (0.057–1.581)	0.156
Pre-existing dementia (yes versus no)	1.680 (0.298–9.466)	0.556
ICH location (right versus left)	0.600 (0.160–2.243)	0.448
ICH volume (per 1 mL increase)	1.136 (1.031–1.250)	0.010
Lobar CMBs count (per 1-CMB increase)	0.999 (0.988–1.009)	0.808
WMH volume (per 1 cm³ increase)	0.997 (0.982–1.013)	0.727
cSS score (per 1-point increase)	2.127 (0.850–5.321)	0.107
cSS score (= 2 versus≤1)	2.000 (0.174–22.949)	0.578
CSO-PVS score (per 1-point increase)	1.010 (0.520–1.959)	0.977
CSO-PVS score (≥3 versus≤2)	1.053 (0.304–3.651)	0.936
Total MRI burden of SVD score	1.819 (0.955–3.465)	0.069
(per 1-point increase)
Total MRI burden of SVD score (≥4 versus≤3)	2.062 (0.579–7.347)	0.264
Blood biomarkers
MMP-2 (per 1-unit increase)	0.000 (0.000–548.707)	0.213
MMP-2 (≥5.241 versus < 5.241)	0.187 (0.049–0.706)	0.013	0.054 (0.005–0.570)	0.015	0.054 (0.005–0.570)	0.015	0.054 (0.005–0.570)	0.015
MMP-3 (per 1-unit increase)	3.104 (0.287–33.564)	0.351
MMP-3 (≥4.190 versus < 4.190)	2.347 (0.659–8.359)	0.188
MMP-9 (per 1-unit increase)	2.744 (0.465–16.184)	0.265
MMP-9 (≥5.000 versus < 5.000)	8.250 (0.957–71.095)	0.055

DISCUSSION

We established a prospective cohort of first probable CAA-ICH in China and investigated the possible roles of serum MMP-2, MMP-3, and MMP-9 in CAA-related ICH and cognitive function. Main results were showed as follows: firstly, serum MMP-2 level was significantly lower in CAA-ICH patients than controls while MMP-9 was significantly higher; secondly, serum MMP-3 level was positively correlated with lobar CMBs counts in CAA-ICH patients; thirdly, higher MMP-2 serum level was a beneficial predictor for CAA-ICH recurrence and was a protective factor for cognitive impairment; fourthly, MMP-9 presented a positive association with recurrent CAA-ICH in the univariable analysis, but did not reach significance after adjustment for other confounders. Our data provided a hint that blood MMPs may be potential biomarkers in predicting recurrent ICH and evaluating the risk of cognitive impairment in CAA.

In this study, we found a significant decrease in serum MMP-2 and a significant increase in MMP-9 during the acute phase of ICH than healthy controls. Previous research demonstrated that plasma MMP-2 level was lower, but MMP-9 was slightly higher within 24 h after the CAA-associated hemorrhagic event than in controls [8]. Other studies on spontaneous intracerebral hemorrhage showed the similar results that plasma MMP-2 was robustly decreased while MMP-9 was dramatically increased after the ICH onset than baseline [20, 21]. Our data were in agreement with those reported previously. This potentially indicated different roles for MMP-2 and MMP-9 in CAA-ICH. As hypothesized for ischemic stroke [22], MMP-9 may be detrimental in blood-brain barrier (BBB) disruption to exacerbate perihematomal edema. On the contrary, MMP-2 might play a protective role in alleviating neuroinflammation and favoring tissue repair.

CMB is one of the characteristic features of CAA. Though the exact underlying mechanisms are still poorly understood, capillary abnormalities involving BBB have been implicated in the development of CMB [23]. MMPs, acting at several sites in the neurovascular unit, regulate the permeability of the BBB. As proteolytic enzymes, MMPs degrade basal lamina around the blood vessels as well as tight junction proteins. MMP-2 is normally secreted by astrocytes from the foot processes which are intimately connected to the endothelial cells [24]. MMP-3 and MMP-9 are primarily produced by microglia, macrophages, and infiltrating neutrophils and pericytes are another major source of MMP-3 [24]. In our study, we found a significant association between MMP-3 serum levels in acute phase after ICH and lobar CMBs counts in CAA patients, but not for MMP-2 or MMP-9. This may be partially attributed to the different spatial distributions of MMPs. Besides, previous studies of ICH patients have presented time-dependent changes in MMPs [20, 25], and this may be another explanation for our results. It highlights the need for serial sampling at different timepoints after ICH ictus.

An interesting finding of our study was that the high serum level of MMP-2 in acute phase after ICH was a beneficial predictor of ICH recurrence in CAA patients. Previous research found MMP-2 was highly expressed in reactive astrocytes surrounding Aβ-damaged vessels and brain scar of chronic bleedings in CAA-ICH patients [8]. MMP-2 was also found in reactive astrocytes near cerebral microvascular amyloid deposits in TgSwDI mice brains [26]. The presence of reactive astrocytes around the fragile cerebral vessels might be a protective barrier to maintain the vessel integrity and MMP-2 expression in astrocytes might digest vascular Aβ deposition [27]. In contrast with MMP-2, we found high serum level of MMP-9 in acute phase after ICH was a predictor of CAA-ICH recurrence, though the efficacy did not reach significance after adjusting for other confounders. In CAA-ICH patients, MMP-9 staining was mainly observed in some isolated macrophages and neutrophils around Aβ-affected vessels [8]. In an animal model of CAA, the majority of cerebrovascular vessels with evidence of prior microhemorrhage showed MMP-9 immunostaining, indicating an association between increased vascular MMP-9 expression and microhemorrhage [28]. Although these results could not demonstrate an alteration of MMP-9 within or near fragile CAA vessels before rupture, one possibility is that the damaging effect of MMP-9 could be the second attack after initial bleeding, being part of the inflammation cascade [29], to cause ICH recurrence.

In addition to hemorrhagic events, MMPs have also been implicated in cognitive impairment. In the brain, MMP-3 can be derived from neurons under stress, which participates in microglia activation and neuroinflammation [30, 31]. Previous studies showed that individuals with mild cognitive impairment or AD had increased levels of MMP-3 in plasma or CSF [32, 33]. In AD brains, MMP-2 and MMP-9 expressions were particularly increased in the astrocytes surrounding amyloid plaques [34, 35]. Gamze et al. concluded that MMP-2 plasma level was significantly lower in AD patients than in the age-matched healthy controls while MMP-9 did not show significant change between two groups [36]. Also, MMP-2 and MMP-9 levels were not correlated with MMSE scores [36]. However, Lim et al. found low MMP-2 activity rather than MMP-2 level in plasma of AD subjects and a positive correlation between MMP-2 plasma activity and MMSE scores [37]. Another study also analyzed the activities and levels of MMP-2 and MMP-9 in the plasma of AD patients and showed that MMP-9 activity as well as MMP-9 level were elevated in AD patients and no correlation was observed between MMPs levels and MMSE scores [38, 39]. However, elevated MMP-9 zymogenic activity was observed in postmortem mild cognitive impairment and AD brains compared with no cognitive impairment brains and significant inverse correlation between MMP-9 activity and cognitive scores at time of death was detected [40]. These inconsistent results might partially be attributed to the truly tiny changes in MMPs concentrations of AD patients. It is possible that the activities of MMPs rather than their levels are better predictors of cognition. In our cohort, higher serum levels of MMP-2 were found to be protective factors for cognitive impairment independent of age, pre-existing dementia, ICH location, and ICH volume, indicating a role of MMP-2 in the cognitive status in CAA-ICH patients.

There are several limitations of this study. Firstly, the size of our prospective cohort is small. Therefore, a maximum of two predictors were allowed in the multivariable Cox regression analyses. A larger cohort is needed to further confirm the existing results. Secondly, high MMPs expression levels do not necessarily lead to increased enzymatic activities, which are more meaningful in the pathogenesis of disease. Our study only focused on MMPs serum levels rather than their enzymatic activities. Thirdly, we did not recruit patients with other types of ICH as comparisons. Hence, the specificity of the predictive ability of MMPs in CAA-ICH recurrence cannot be evaluated. Fourthly, since some patients could not cooperate in the MMSE measurement, only 47 cases were assessed cognitive status. Though the comparisons of demographic and neuroimaging characteristics as well as serum MMPs levels between these two groups showed no significant differences (Supplementary Table 2), bias might still exist.

In conclusion, serum MMP-2 level in acute phase might be a promising biomarker to predict CAA-ICH recurrence and to evaluate the risk of cognitive impairment.

Footnotes

ACKNOWLEDGMENTS

This study is funded by National Key R & D Program of China (2016YFC1300500-3, 2017YFC1308201), National Natural Science Foundation of China (81971123), the Chinese Academy of Sciences (QYZDY-SSW-SMC012 and XDB39000000), the Fundamental Research Funds for the Central Universities (YD2070002003), Clinical Research Plan of Shanghai Hospital Development Center (SHDC2020CR4016), Shanghai Rising-Star Program (15QA1400900), Shanghai Municipal Science and Technology Major Project (2018SHZDZX01) and ZJLab.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Jellinger

(2002) Alzheimer disease and cerebrovascular pathology: An update. J Neural Transm (Vienna) 109, 813–836.

Charidimou

, Boulouis

, Roongpiboonsopit

, Xiong

, Pasi

, Schwab

, Rosand

, Gurol

, Greenberg

, Viswanathan

(2019) Cortical superficial siderosis and recurrent intracerebral hemorrhage risk in cerebral amyloid angiopathy: Large prospective cohort and preliminary meta-analysis. Int J Stroke 14, 723–733.

Charidimou

, Boulouis

, Greenberg

, Viswanathan

(2019) Cortical superficial siderosis and bleeding risk in cerebral amyloid angiopathy: A meta-analysis. Neurology 93, e2192–e2202.

Biffi

, Anderson

, Battey

, Ayres

, Greenberg

, Viswanathan

, Rosand

(2015) Association between blood pressure control and risk of recurrent intracerebral hemorrhage. JAMA 314, 904–912.

Zhao

, Arbel-Ornath

, Wang

, Betensky

, Greenberg

, Frosch

, Bacskai

(2015) Matrix metalloproteinase 9-mediated intracerebral hemorrhage induced by cerebral amyloid angiopathy. Neurobiol Aging 36, 2963–2971.

Chang

, Emanuel

, Mack

, Tsivgoulis

, Alexandrov

(2014) Matrix metalloproteinase-9: Dual role and temporal profile in intracerebral hemorrhage. J Stroke Cerebrovasc Dis 23, 2498–2505.

Jung

, Zhang

, Van Nostrand

(2003) Pathogenic A beta induces the expression and activation of matrix metalloproteinase-2 in human cerebrovascular smooth muscle cells. J Neurochem 85, 1208–1215.

Hernandez-Guillamon

, Martinez-Saez

, Delgado

, Domingues-Montanari

, Boada

, Penalba

, Boada

, Pagola

, Maisterra

, Rodriguez-Luna

, Molina

, Rovira

, Alvarez-Sabin

, Ortega-Aznar

, Montaner

(2012) MMP-2/MMP-9 plasma level and brain expression in cerebral amyloid angiopathy-associated hemorrhagic stroke. Brain Pathol 22, 133–141.

Tanskanen

, Myllykangas

, Saarialho-Kere

, Paetau

(2011) Matrix metalloproteinase-beta19 expressed in cerebral amyloid angiopathy. Amyloid 18, 3–9.

10.

Xia

, Shen

, Wang

, Yang

, Wang

, Li

, Liang

, Zhao

, Wu

, Chu

, Liang

, Wang

, Qiu

, Ding

, Cheng

, Dong

(2020) White matter hyperintensities associated with progression of cerebral small vessel disease: A 7-year Chinese urban community study. Aging (Albany NY) 12, 8506–8522.

11.

Cheng

, Su

, Wang

, Gao

, Ye

, Wang

, Xia

, Fu

, Shen

, Salman

, Dong

(2020) Neurofilament light chain predicts risk of recurrence in cerebral amyloid angiopathy-related intracerebral hemorrhage. Aging (Albany NY) 12, 23727–23738.

12.

Charidimou

, Martinez-Ramirez

, Reijmer

, Oliveira-Filho

, Lauer

, Roongpiboonsopit

, Frosch

, Vashkevich

, Ayres

, Rosand

, Gurol

, Greenberg

, Viswanathan

(2016) Total magnetic resonance imaging burden of small vessel disease in cerebral amyloid angiopathy: An imaging-pathologic study of concept validation. JAMA Neurol 73, 994–1001.

13.

Yushkevich

, Yang

, Gerig

(2016) ITK-SNAP: An interactive tool for semi-automatic segmentation of multi-modality biomedical images. Annu Int Conf IEEE Eng Med Biol Soc 2016, 3342–3345.

14.

Zhang

, Katzman

, Salmon

, Jin

, Cai

, Wang

, Qu

, Grant

, Yu

, Levy

, et al. (1990) The prevalence of dementia and Alzheimer’s disease in Shanghai, China: Impact of age, gender, and education. Ann Neurol 27, 428–437.

15.

Bravo

, Hebert

(1997) Age- and education-specific reference values for the Mini-Mental and modified Mini-Mental State Examinations derived from a non-demented elderly population. Int J Geriatr Psychiatry 12, 1008–1018.

16.

Cui

, Yao

, Xu

, Tang

, Jiang

, Wang

, Chen

, Cheng

(2011) Cognitive impairment using education-based cutoff points for CMMSE scores in elderly Chinese people of agricultural and rural Shanghai China. Acta Neurol Scand 124, 361–367.

17.

Yin

, Yan

, Liang

, Jiang

, Cai

, Song

, Feng

, Qiu

(2016) Interactive effects of diabetes and impaired kidney function on cognitive performance in old age: A population-based study. BMC Geriatr 16, 7.

18.

Yin

, Fei

, Qiu

, Brasher

, Kraus

, Zhao

, Shi

, Zeng

(2017) Dietary diversity and cognitive function among elderly people: A population-based study. J Nutr Health Aging 21, 1089–1094.

19.

Camp

, Dolled-Filhart

, Rimm

(2004) X-tile: A new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res 10, 7252–7259.

20.

Alvarez-Sabin

, Delgado

, Abilleira

, Molina

, Arenillas

, Ribo

, Santamarina

, Quintana

, Monasterio

, Montaner

(2004) Temporal profile of matrix metalloproteinases and their inhibitors after spontaneous intracerebral hemorrhage: Relationship to clinical and radiological outcome. Stroke 35, 1316–1322.

21.

Castellazzi

, Tamborino

, De Santis

, Garofano

, Lupato

, Ramponi

, Trentini

, Casetta

, Bellini

, Fainardi

(2010) Timing of serum active MMP-9 and MMP-2 levels in acute and subacute phases after spontaneous intracerebral hemorrhage. Acta Neurochir Suppl 106, 137–140.

22.

Zhao

, Tejima

, Lo

(2007) Neurovascular proteases in brain injury, hemorrhage and remodeling after stroke. Stroke 38, 748–752.

23.

Fisher

(2016) Cerebral microbleeds and thrombolysis: Clinical consequences and mechanistic implications. JAMA Neurol 73, 632–635.

24.

Yang

, Rosenberg

(2015) Matrix metalloproteinases as therapeutic targets for stroke. Brain Res 1623, 30–38.

25.

Howe

, Zhu

, Sansing

, Gonzales

, McCullough

, Edwards

(2018) Serum markers of blood-brain barrier remodeling and fibrosis as predictors of etiology and clinicoradiologic outcome in intracerebral hemorrhage. Front Neurol 9, 746.

26.

Liao

, Van Nostrand

(2010) Degradation of soluble and fibrillar amyloid beta-protein by matrix metalloproteinase (MT1-MMP) in vitro. Biochemistry 49, 1127–1136.

27.

Hernandez-Guillamon

, Mawhirt

, Fossati

, Blais

, Pares

, Penalba

, Boada

, Couraud

, Neubert

, Montaner

, Ghiso

, Rostagno

(2010) Matrix metalloproteinase 2 (MMP-2) degrades soluble vasculotropic amyloid-beta E22Q and L34V mutants, delaying their toxicity for human brain microvascular endothelial cells. J Biol Chem 285, 27144–27158.

28.

Lee

, Yin

, Hsin

, Chen

, Fryer

, Holtzman

, Hsu

, Xu

(2003) Matrix metalloproteinase-9 and spontaneous hemorrhage in an animal model of cerebral amyloid angiopathy. Ann Neurol 54, 379–382.

29.

Rosell

, Ortega-Aznar

, Alvarez-Sabin

, Fernandez-Cadenas

, Ribo

, Molina

, Lo

, Montaner

(2006) Increased brain expression of matrix metalloproteinase-9 after ischemic and hemorrhagic human stroke. Stroke 37, 1399–1406.

30.

Yoshiyama

, Asahina

, Hattori

(2000) Selective distribution of matrix metalloproteinase-3 (MMP-3) in Alzheimer’s disease brain. Acta Neuropathol 99, 91–95.

31.

Kim

, Choi

, Block

, Lorenzl

, Yang

, Kim

, Sugama

, Cho

, Hwang

, Browne

, Kim

, Hong

, Beal

, Joh

(2007) A pivotal role of matrix metalloproteinase-3 activity in dopaminergic neuronal degeneration via microglial activation. FASEB J 21, 179–187.

32.

Hanzel

, Iulita

, Eyjolfsdottir

, Hjorth

, Schultzberg

, Eriksdotter

, Cuello

(2014) Analysis of matrix metallo-proteases and the plasminogen system in mild cognitive impairment and Alzheimer’s disease cerebrospinal fluid. J Alzheimers Dis 40, 667–678.

33.

Iulita

, Ganesh

, Pentz

, Flores Aguilar

, Gubert

, Ducatenzeiler

, Christie

, Wilcock

, Cuello

(2019) Identification and preliminary validation of a plasma profile associated with cognitive decline in dementia and at-risk individuals: A retrospective cohort analysis. J Alzheimers Dis 67, 327–341.

34.

, Poteet

, Xie

, Liu

, Wen

, Yang

(2011) Regulation of matrix metalloproteinase 2 by oligomeric amyloid beta protein. Brain Res 1387, 141–148.

35.

Yin

, Cirrito

, Yan

, Hu

, Xiao

, Pan

, Bateman

, Song

, Hsu

, Turk

, Xu

, Hsu

, Mills

, Holtzman

, Lee

(2006) Matrix metalloproteinases expressed by astrocytes mediate extracellular amyloid-beta peptide catabolism. J Neurosci 26, 10939–10948.

36.

Tuna

, Yener

, Oktay

, Islekel

, FG

(2018) Evaluation of matrix metalloproteinase-2 (MMP-2) and -9 (MMP-9) and their tissue inhibitors (TIMP-1 and TIMP-2) in plasma from patients with neurodegenerative dementia. J Alzheimers Dis 66, 1265–1273.

37.

Lim

, Villemagne

, Soon

, Laughton

, Rowe

, McLean

, Masters

, Evin

, Li

(2011) Investigation of matrix metalloproteinases, MMP-2 and MMP-9, in plasma reveals a decrease of MMP-2 in Alzheimer’s disease. J Alzheimers Dis 26, 779–786.

38.

Lorenzl

, Albers

, Relkin

, Ngyuen

, Hilgenberg

, Chirichigno

, Cudkowicz

, Beal

(2003) Increased plasma levels of matrix metalloproteinase-9 in patients with Alzheimer’s disease. Neurochem Int 43, 191–196.

39.

Lorenzl

, Buerger

, Hampel

, Beal

(2008) Profiles of matrix metalloproteinases and their inhibitors in plasma of patients with dementia. Int Psychogeriatr 20, 67–76.

40.

Bruno

, Mufson

, Wuu

, Cuello

(2009) Increased matrix metalloproteinase 9 activity in mild cognitive impairment. J Neuropathol Exp Neurol 68, 1309–1318.