Introduction

Primary intracerebral hemorrhage (ICH) results in high morbidity and disability. Cerebral small vessel disease (CSVD) is the major cause of primary ICH [1]: arteriolosclerosis, which is closely associated with hypertension [2], mainly accounts for deep ICH, and cerebral amyloid angiopathy (CAA) accounts for lobar ICH. Notably, CAA and arteriolosclerosis coexist frequently. Previous studies have found that lobar ICH/microbleed and deep ICH/microbleed could coexist in a patient [3]. Additionally, animal studies have found that an increase in systolic blood pressure promotes the occurrence of ICH in Tg2576 mice characterized by amyloid beta (Aβ) deposition [4]. And in a clinical study, subgroup analysis of the PROGRESS study has shown that antihypertensive treatment (vs. placebo) reduced the risk of ICH in CAA patients [5]. However, there is currently a lack of systematic reviews that assess the association between CAA and arteriolosclerosis in ICH patients with pathology-proven evidence.

On the other hand, increasing evidence has demonstrated that imaging markers of CSVD are related to the presence and prognosis of primary ICH [6, 7]. However, the characteristics of CSVD markers in patients with pathological evidence for arteriolosclerosis- and CAA-related ICH, and the underlying pathological changes of theses markers, remain unclear. Although a previous study [8] reviewed the pathological substrates of CSVD such as cerebral microbleeds (CMB) and white matter changes, the majority of included studies investigated Alzheimer’s disease. Despite 80~97% of patients with Alzheimer’s disease have CAA at autopsy, most of them do not suffer from symptomatic ICH [9, 10]. We, therefore, performed a systematic review and meta-analysis to elucidate (1) the association between arteriolosclerosis and CAA and their imaging characteristics, and (2) the imaging and pathological correlations of CSVD markers in primary ICH patients to better understand the mechanisms underlying primary ICH.

Methods

Search Strategy

Our study was approved by the Ethics Committee on Biomedical Research, West China Hospital of Sichuan University, and was registered in PROSPERO (CRD42022343347). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [11] guidelines. We systematically searched PubMed, Ovid Medline, and Web of Science for studies investigating the neuropathology of CSVD in patients with primary ICH until 8 June 2022. The following terms were used: “(cerebral hemorrhage) OR (intracranial hemorrhage)” AND “(patholog*) OR (postmortem) OR (autopsy) OR (biopsy)” AND “(cerebral small vessel diseases) OR (microvessel) OR (white matter) OR (microinfarct) OR (lacune) OR (perivascular space) OR (microbleed) OR (microhemorrhage) OR (atrophy).” Additionally, references to relevant studies were searched. The details of the study selection process are presented in Fig. 1.

Fig. 1
figure 1

Flow chart of study selection

Full size image

Eligibility Criteria

The included criteria were as follows: investigating adult patients (age≥18 years) with primary ICH; having autopsy or biopsy to confirm the etiology for arteriolosclerosis and CAA; with or without CSVD markers, including white matter changes, CMB, cerebral microinfarcts (CMI), lacunes, enlarged perivascular spaces (EPVS), brain atrophy, and the total CSVD burden. Abstract, review, editorial, and basic research were excluded. This systematic review only focused on primary ICH, not on hemorrhage into the subarachnoid space, epidural space, or subdural space. Pure intraventricular hemorrhage, ICH secondary to trauma, structural vascular lesions, inflammation, infections, inherited or genetic small vessel diseases, and hemorrhagic transformation after ischemic stroke were beyond the scope of this review. We only included studies that were published in English.

Data Extraction

Two authors (Mangmang Xu and Xindi Song) independently searched the literature and extracted data. Any disagreement was resolved by consensus. A data extraction form was developed using SPSS. For each included study, we extracted the family name of the first author, publication year, country of origin, the sample size of primary ICH, examination method (autopsy or biopsy), patient demographics (male sex, age at death or biopsy), pathology diagnosis (CAA or arteriolosclerosis), ICH location (lobe, deep, cerebellum), and CSVD imaging markers including white matter changes, lacunes, CMB, CMI, EPVS, atrophy, and total CSVD burden. For studies that investigated ICH with CAA evidence, we additionally extracted information on the diagnostic method of CAA, CAA-related small vessel wall changes (SVWCs), the coexistence of arteriolosclerosis, senile plaques, and blood vessel changes in or around the hematoma.

Definitions

Lobar ICH included hemorrhages originating from the frontal, temporal, parietal, or occipital lobes of the brain. For the two studies [12, 13] that enrolled clinically diagnosed CAA-ICH according to the modified Boston criteria without reporting ICH location, we regarded those cases as having lobe ICH. Deep ICH was considered when originating from the basal ganglia, thalamus, internal capsule, external capsule, or brain stem. Biology included brain biopsy and biopsy at hematoma evacuation in this study.

Severe CAA in histopathology was defined as follows: (1) a CAA score of 9–12 for the study [12] that used a cumulative cortical CAA score with a range of 0–12 by calculating the basis of Aβ stained sections from four areas, (2) a CAA score of 3–4 for those studies [14,15,16,17] that used the grading scale by Greenberg and Vonsattel with a range of 0–4, (3) a CAA score of 3+ indicating that many lesions were visible [18] in a scale of 1+~3+, or as per the original study authors’ definition [19,20,21,22].

Arteriolosclerosis was defined if there existed lipofibrohyalinosis [20], lipohyalinosis [21, 23], arteriolosclerosis [13, 18, 19, 24,25,26,27,28], or hypertensive vasculopathy [12, 13, 17] on pathology. For the study [21] that performed a pathological examination and concluded that there were no other possible causes of ICH except CAA, we considered those cases as strict CAA.

Statistical Analysis

Most of the included studies were case reports, so we extracted individual patient data from those cases. The demographic data in the study by Ter Telgte [12] were the same as that in the study by van Veluw [13]; therefore, we only extracted data from the latter. For CMB analysis, we extracted CMB number in ex vivo MRI in the study by van Veluw 2016 [17], under the other two studies [13, 20]. Independent-samples T test, one-way ANOVA, Mann-Whitney U test, or Kruskal-Wallis H test were used for continuous variables, and Pearson chi-squared test or Fisher’s exact test for categorical variables, when appropriate. Multivariate analysis was performed using binary logistic regression. For the statistical analysis in our present study, hemorrhage in the cerebellum was categorized as lobar ICH. All statistical analyses were performed using IBM SPSS Statistics (version 23), and a p value of ≤0.05 was considered statistically significant.

Results

Study Characteristics

A total of 4155 citations and 509 full texts were screened (Fig. 1). Of these, 28 studies with 456 participants were included [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39]. The characteristics of included studies are shown in Table 1. Most of the studies were conducted in the USA (39.3%, 11/28), followed by Japan (28.6%, 8/28). Three hundred twenty patients had CAA with or without arteriolosclerosis; 48 patients had strict CAA while 19 had CAA + arteriolosclerosis. CAA-related SVWCs were investigated in 17 studies [14,15,16,17,18,19,20,21,22, 24, 29,30,31,32,33, 38, 39], and their detailed pathological changes are summarized in the Supplementary table.

Table 1 Characteristics of included studies which investigating primary ICH
Full size table

The Association Between Arteriolosclerosis and CAA

Ten studies [13, 17,18,19,20,21, 23,24,25, 33] with 79 cases assessed both CAA and arteriolosclerosis histopathologically. CAA + arteriolosclerosis accounted for 21.5% (17/79) of cases. The frequency of lobar ICH differed among CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis (82.4%, 100%, 38.9%, respectively; p<0.001), so did the total number of CMB (median 7, 161, 0, respectively; p=0.015).

Ten studies [13,14,15,16,17,18,19,20,21,22] with 76 cases assessed the presence and severity of CAA on pathology. Data showed that severe SVWCs due to CAA were highly prevalent (67.6%, 53/76), which was found to be associated with arteriolosclerosis when compared with CAA without severe SVWCs (58.3 vs. 18.8%, p=0.050; OR 6.067, 95% CI 1.107–33.238, p=0.038). However, this association was not significant after correcting for age and sex (adjusted OR 4.959, 95% CI 0.734–33.521, p=0.101).

The Clinical and Imaging Characteristics of Arteriolosclerosis and CAA

As shown in Table 2, all cases of ICH due to strict CAA aged > 55 years and had lobar ICH. CAA presence was associated with a higher number of CMB (CAA vs. non-CAA: 15 (2–104) vs. 0 (0–3), p=0.006, calculated from 7 studies [13, 16, 17, 20, 25, 31, 34] with 29 participants), as well as a higher number of CMI (CAA vs. non-CAA: 29 (3.5–56.25) vs. 0 (0–0), p=0.064, calculated from 3 studies [13, 17, 25] with 10 participants), although without reaching a statistically significant difference for the latter. And patients with strict CAA were more likely to have a higher number of total CMB (OR 1.016, 95% CI 1.002–1.030, p=0.029, as determined by binary logistic regression) than those with strict arteriolosclerosis or CAA+ arteriolosclerosis; this association remained significant after adjusting for age and sex (adjusted OR 1.015, 95% CI 1.001–1.029, p=0.040). There existed conflicting results regarding the association between ICH etiology and CMB number by location [20, 32]. For cerebellar CMB, CAA-ICH had more superficial cerebellar CMB, while non-CAA-ICH had more deep/mixed cerebellar CMB [37].

Table 2 The clinical characteristics by CAA severity and ICH etiology
Full size table

For patients with ICH due to strict arteriolosclerosis, the majority (61.1%) of ICH were located in deep regions of the brain. The presence of arteriolosclerosis was significantly associated with hypertension (percentage of hypertension in arteriosclerosis vs. non-arteriolosclerosis: 74.6 vs. 36.4%, p=0.001). This association remained significant after correcting for age and sex (adjusted OR 3.329, 95% CI 1.056–10.495, p=0.040).

The Correlation Between MRI and Pathological Changes of CSVD Markers in Primary ICH

White Matter Changes

Overall, six studies [15, 18, 20, 30, 32, 36] investigated white matter changes on CT or MRI. Of these, three studies [18, 20, 30] examined the correlation between MRI and pathological changes in white matter lesions (Table 3). White matter changes were associated with rarefaction of myelin staining, diffuse or patchy myelin loss, diffuse white matter edema, and even advanced white matter loss [20, 30]. Histologically, the white matter appeared vacuolated and was accompanied by swollen oligodendrocytes and astrocytic proliferation [18, 30]. Amyloid plaques that were positive for Aβ were occasionally observed in the white matter [18]. However, CAA-associated vasculopathy did not always result in ischemic white matter lesions [18].

Table 3 Neuropathological changes of white matter changes on imaging
Full size table

Cerebral Microbleed

Twelve studies [13, 15,16,17, 20, 22, 25, 31, 32, 34, 35, 37] investigated CMB on MRI, and nine [13, 16, 17, 20, 22, 25, 31, 34, 35] of them retrieved CMB for histopathological analysis. The pathological changes of CMB in primary ICH are presented in Table 4. There existed inconsistencies in CAA severity around a CMB [13, 25]. The vessels involved in CMB were enlarged and fibrinoid necrotic, with extensive remodeling of the vessel wall [13]. On the other hand, a high CMB count was associated with an increased wall thickness of amyloid-positive vessels, and this increased thickness of vessel walls might be more prone to CMB formation than macrobleed formation [34].

Table 4 Neuropathological changes of other CSVD markers in ICH patients
Full size table

Cerebral Microinfarct

Four studies [12, 13, 17, 25] investigated CMI, and all four investigated pathological changes of CMI, as detailed in Table 4. The findings suggested that small DWI lesions could be matched to acute CMI histopathologically, characterized by tissue pallor consisting of eosinophilic neurons, absent or only mild reactive astrocytes, and absent or fragmented microglia. And chronic cortical CMI was associated with tissue loss or central cavitation, as well as many reactive astrocytes around the lesions and few reactive or amoeboid microglia [12]. CAA was found to be severe surrounding a CMI, particularly for CMI in those with lobar ICH.

Lacunes, EPVS, and Brain Atrophy

Each of the three imaging markers was investigated in a separate study (Table 1). Patients with brain atrophy corresponded to amyloid deposition in the parenchyma and small vessels [15]. However, the pathological changes of lacune and EPVS in primary ICH were not investigated.

Total CSVD Burden

Two studies [26, 36] investigated the total CSVD burden (Table 4). A direct histopathological-MRI study demonstrated that the total CAA burden of CSVD, as determined by summing the four most characteristic MRI markers of CAA, was associated with the presentation of symptomatic CAA-related ICH (versus CAA without ICH) and CAA-related microangiopathy on pathology [36].

Discussion

We presented here a comprehensive systematic review of published articles that investigated ICH with pathology-proven evidence for arteriolosclerosis and CAA. Our study directly compared MRI and pathology to clarify the pathological changes of CSVD imaging markers. To the best of our knowledge, this is the first review to assess the interaction between arteriolosclerosis and CAA by summarizing published cases with a pathology diagnosis of CAA or arteriolosclerosis in ICH. Our findings provide pathological evidence that there might be differences in lobar ICH and total microbleed number among CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis, and there might be an association between arteriolosclerosis and severe CAA, which needs to be investigated further in future studies.

Although the SVWCs in CAA-ICH were found to be similar to those in CAA without ICH, in terms of the location of Aβ deposition and the changes in the lumen [40], our findings suggest that severe CAA (usually characterized by double barreling and/or fibrinoid necrosis) is highly prevalent in CAA-ICH. Our results, together with the findings of Vonsattel et al. [21], indicate that a severe degree of CAA represents an important feature of CAA-ICH [21].

Regarding the CSVD markers in ICH, we found that patients with CAA had higher number of total CMB, particularly for patients with strict CAA, as compared to those without CAA on pathology. This finding was in line with previous study by Edip Gurol et al. that found that patients with CAA-ICH had higher number of total CMB as compared to those with non-CAA related ICH [41]. In addition, the total CAA burden assessed in neuroimaging was associated with the presentation of symptomatic CAA-related ICH and CAA-related microangiopathy on pathology. Therefore, it might be reasonable to speculate that higher individual CSVD markers indicate greater Aβ severity. However, our analysis of individual patient data showed that severe CAA was not associated with CMB number, which is in line with the findings of another study that found CMB to be frequently located around Aβ negative small vasculature [42]. In addition, one amyloid positron emission tomography study evaluating CAA burden in both the ICH-affected hemisphere and the ICH-free hemisphere indicated that amyloid burden was similarly distributed and ICH was unlikely to be directly linked to amyloid burden [43]. Taken together, these findings challenge the hypothesis that the ruptured vessel reflects a significant Aβ load [42].

As expected, hypertension was significantly associated with arteriolosclerosis, which was highly prevalent in CAA-ICH as well. It is widely accepted that lobar hemorrhage is required for CAA diagnosis per the Boston Criteria [44]. In this study, we found that all cases of strict CAA-ICH were located in the lobes. However, we also found that in some patients with the coexistence of CAA and arteriolosclerosis on pathology, the hemorrhage could be located in deep territories, suggesting that the presence of deep ICH could not preclude the possibility of amyloid angiopathy in the elderly patients.

Overall, the pathology of CSVD markers in ICH was under-researched, with most studies focusing on CAA-ICH. While the majority of CMBs observed on MRI could be confirmed histopathologically, several CMBs could not be confirmed, with no abnormality or small cavities at neuropathological amination [20, 35]. Compared with standard histopathological sections, MRI tends to underestimate the total burden of CMB [20], as well as CMI [13]. The lesions that could not be retrieved for histological analysis were generally too small [25]. Therefore, future studies with high-field MRI scanners are warranted to capture the total burden of these two markers [20, 35]. Of note, CMB on MRI was traditionally recognized to represent foci of past hemorrhage [20, 22]. However, this review found that CMB could also correspond to recent microhemorrhage with small bleeds in patients with ICH [35].

Another interesting finding was the inconsistency in CAA severity around CMB [13, 25]. This inconsistency might be attributed to the different methodologies used by the two studies. Jolink et al. [25] investigated CAA severity for each patient by taking samples with multiple lesions, while van Veluw et al. [13] assessed CAA for each section that contained CMB. Studies that directly correlated MRI and pathology of lacunes, EPVS, and atrophy were scarce. Thus, future studies are warranted to further elucidate the CAA pathology around CMB and CMI, and to assess the pathology of lacunes, EPVS, and atrophy in ICH.

Strength and Limitation

The present systematic review provided the most recent and comprehensive overview of the pathology of CSVD in primary ICH. Studies were comprehensively searched from databases using a systematic search strategy, as well as from handsearching of references of relevant studies. However, the majority of included studies only assessed the presence of CAA on pathology, without information on arteriolosclerosis, resulting in a relatively small number of cases with CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis. Despite this, our study has the largest number of ICH cases with pathology-proven etiology across the literature and firstly provides the clinical and imaging information for CAA+arteriolosclerosis, strict CAA, and strict arteriolosclerosis.

Conclusion and Considerations for Future Research

In summary, our study provides synthesized pathological evidence for the hypothesis that there might be an interaction between CAA and arteriolosclerosis. Future studies are needed to verify our findings. The pathological changes of CSVD features in neuroimaging are heterogeneous, and MRI-histopathological correlation studies of lacunes and EPVS remain scarce. Therefore, further investigation is needed to explore the pathological changes of CSVD markers by ICH etiology. Additionally, studies using advanced MRI techniques with a high spatial resolution to visualize structural and functional brain changes of CSVD in ICH are also warranted.