Cerebral small vessel disease (CSVD) is caused by the impairment of small perforating arteries and venules in the brain (1). Hypertensive angiopathy (HA) and cerebral amyloid angiopathy (CAA) are the two most common pathologies of CSVD (1, 2), and are the main etiologies of primary intracerebral hemorrhage (ICH) (3). In general, HA causes deep ICH or cerebral microbleeds (CMBs), while CAA results in lobar ICH/CMBs (4).

Despite the different pathogeneses of HA and CAA, hypertension is frequently observed in CAA patients (5). Previous experimental research has suggested that hypertension could accelerate CAA accumulation (6) and promote ICH occurrence in Tg2576 mice (a CAA mouse model) (7). Furthermore, histopathological studies have shown that HA-related arteriolosclerosis and CAA pathological changes often coexist (8, 9). Therefore, there might be an association between HA and CAA.

Recent researches have suggested that CSVD makers, such as CMBs, cortical superficial siderosis (cSS), enlarged perivascular spaces (EPVS), and white matter hyperintensities (WMH), could help to determine the etiology of primary ICH (3, 10). These features have been selected as MRI features in the Boston criteria version 2.0 (11). In addition, the CAA-CSVD score, which incorporates CAA CSVD markers such as lobar CMBs, cSS, centrum semiovale-EPVS (CSO-EPVS), and WMH into an ordinal scale, has been found to be associated with the presence of CAA-related pathological changes (12). Hence, using CSVD markers to investigate the relationship between HA and CAA might be a practical approach. In this study, we aimed to investigate the association between the burden of CSVD due to HA and the severity of CAA on MRI in patients with primary ICH.

After screening 301 consecutive ICH patients for potential eligibility, 70 patients were excluded due to lack of CTA or MRA to confirm the absence of structural lesions, and nine patients with prior ischemic stroke or primary intraventricular hemorrhage. Finally, 222 cases of primary ICH were included in the analysis. HA-ICH (42.8%, 95/222) accounted the most among the included patients, followed by mixed-ICH (40.5%, 90/222), and CAA-ICH (13.1%, 29/222). Eight patients (3.6%) were diagnosed with an undetermined etiology. Specifically for CAA-ICH, the age of patients with CAA ranged from 56 to 88, with a median of 70 and interquartile range of 64–77.

Table 2 presents the clinical characteristics and CSVD markers of the general population and subgroups stratified by lobar CMB severity or the presence/absence of high CAA-CSVD score. Among patients with different level of lobar CMB severity, age (p = 0.044) and ICH etiology (p < 0.001) differed among those with lobar CMB 0–1, lobar CMBs 2–4, and lobar CMBs ≥5. There were no significant differences in vascular risk factors and laboratory examination variables. Lobar CMB severity was positively associated with a high prevalence of HA related CSVD markers such as the presence of lacune, the presence of deep CMB, deep CMBs ≥5, periventricular WMH scored 3, deep WMH scored ≥2, and the presence of WMH, as well as higher number of deep CMBs and a higher HA-CSVD score (all p < 0.001). Also, patients with high CAA-CSVD score were older (p < 0.001) and were more likely to have a high burden of HA (p < 0.001 for each of the HA-related CSVD markers). ICH etiology differed between patients with and without high CAA-CSVD score (p < 0.001). Concerning laboratory examination variables, blood glucose (p = 0.030), total cholesterol (p = 0.011), and LDL-C levels (p = 0.049) on admission were lower in patients with high CAA-CSVD score as compared with those without high CAA-CSVD score.

In order to explore the association between HA burden and vascular risk factors, we generated a Supplementary Table 1 to show the distribution of vascular risk factors across different deep CMB numbers and between HA-CSVD score ≥ 2 and HA-CSVD score < 2. The results indicated that as the number of deep CMB increased, the proportion of hypertension also increased (p < 0.001). Also, patients with HA-CSVD score ≥ 2 exhibited a higher proportion of hypertension compared to those with HA-CSVD score < 2 (p = 0.002). In the multivariate analyses, we adjusted for age, male sex, vascular risk factors, including hypertension, diabetes mellitus, hyperlipidemia, smoking, and alcohol, as well as ICH etiology, to examine the association between each of the HA-related CSVD markers and CAA (Table 3). The data revealed that the presence of lacune (adjusted odds ratio [ad OR] 3.909, 95% CI 1.587–9.625, p = 0.003), deep CMBs ≥5 (ad OR 4.600, 95% CI 1.811–11.687, p = 0.001), the number of deep CMBs (ad OR 1.057, 95% CI 1.002–1.116), the presence of WMH (ad OR 5.425, 95% CI 1.911–15.398), and HA-CSVD score (ad OR 2.317, 95% CI 1.483–3.621) were independently associated with the presence of lobar CMBs ≥5. In addition, high CAA-CSVD score was found to be significantly associated with each of the HA-related CSVD markers and the HA-CSVD score, although the association between a higher CAA-CSVD score and the number of deep CMBs was borderline significant.

Due to the significant differences in blood glucose, total cholesterol, and LDL-C levels between groups with and without a high CAA-CSVD score as observed in the univariate analysis, we additionally conducted a multivariate binary logistic regression analysis to assess the association between each of the HA-related CSVD markers and CAA by adjusting for age, male sex, hypertension, blood glucose, total cholesterol, smoking, alcohol, and ICH etiology (see Supplementary Table 2). The results were similar to that in Table 3. Specifically, the presence of lobar CMBs ≥5 or a high CAA-CSVD score was independently related with the presence of HA-related markers and a higher HA-CSVD score, with the exception of the presence of deep CMB.

Figure 1 illustrates the distribution of CAA-CSVD score in patients with and without deep CMBs ≥5 (ad OR 2.351, 95% CI 1.242–4.455, p = 0.009, as determined by ordinal analysis after correcting for age, sex, ICH etiology, and vascular risk factors). The heatmap (Figure 2) displays a correlation between CAA-CSVD score and HA-CSVD score (Spearman correlation coefficients 0.631, p < 0.001). Furthermore, when we stratified the patients based on age and ICH etiology, HA-CSVD score was correlated with CAA-CSVD score or lobar CMB severity in each subgroup (Table 4).

Our study demonstrates that the MRI markers of HA, for instance, the presence of lacune, deep CMBs ≥5, and the presence of WMH, are strongly associated with lobar CMB severity, and further reveals that HA-CSVD score and CAA-CSVD score are closely correlated. These associations are independent of age, sex, ICH etiology, and potential vascular risk factors. Notably, our results supported that the definition of HA based on imaging makers reflected the modifiable vascular risk factors, particularly hypertension, which is the most important risk factor for ICH (21). Specifically for the etiology of CAA, it was diagnosed in 13.1% (29/222) of the included patients, falling within the prevalence spectrum of 8.8% ~ 33% for primary ICH (22–24).

To the best of our knowledge, our study firstly systematically examined the association between HA and CAA based on the severity of their CSVD markers in ICH patients. Although Pasi et al. (15) previously reported that lobar CMBs were correlated with deep CMBs, periventricular WMH, and deep WMH in patients with ICH determined by Spearman correlation analysis. Those associations were not adjusted for confounding factors in this study (15). In our study, we not only confirmed the association between lobar CMBs and HA-related CSVD markers by adjusting for age, sex and potential vascular risk factors, but also observed a correlation between HA-CSVD score and lobar CMB severity, as well as CAA-CSVD score in patients with primary ICH. Interestingly, Ii et al. (20) have reported a positive correlation between HA-CSVD score and CAA-CSVD score in patients undergoing cognitive impairment workup. Meanwhile, the other two studies (25, 26) reported that WMH volume and lacune were positively associated with both lobar and deep CMBs, where lacune was considered as a marker of hypertensive CSVD. Taken together with our findings, the overlap between HA and CAA is common.

Recently, we performed a systematic review and meta-analysis (27) to explore the association between arteriolosclerosis [also named hypertensive small vessel disease (4)] and CAA on pathology by including studies that enrolled patients with primary ICH with pathological examination. And this study (27) found that severe CAA pathological changes were associated with arteriolosclerosis; however, this association was not independent of age and sex. This weak association might be attributed to the small sample size of patient with pathology-proven etiology, or might be attributed to increasing age, a shared risk factor for both HA and CAA (4). Accordingly, we performed a subgroup analysis by stratifying patients into different age groups in our present study. The data demonstrated the association between HA-CSVD score and CAA-CSVD score in each age group. In addition, Kim et al. (25) performed a 3-year longitudinal positron emission topography (PET) study and found that baseline and longitudinal lacune number were associated with Pittsburgh compound retention on longitudinal lobar CMBs in patients with mild cognitive impairment, suggesting that HA had a synergistic effect on the progression of lobar CMBs. Therefore, CAA and HA might have a interact link, which needs to be investigated further in future studies.

Another important finding was that blood glucose, total cholesterol, and LDL-C levels on admission were different between patients with and without high CAA score. After correcting for age and sex, high CAA score was associated with lower total cholesterol (ad OR 0.721, 95% CI 0.527–0.988, p = 0.042, as determined by multivariate binary regression analysis), but not blood glucose and LDL-C. MUCH-Italy study (17) and one previous meta-analysis (21) have reported that low total cholesterol is associated with ICH occurrence, including lobar ICH which is mainly due to CAA. Also, total cholesterol in CAA-ICH is reported to be lower than that in Alzheimer’s disease or controls (age and sex matched healthy subjects) (12). Therefore, lipid levels might be involved in the progression of CAA-CSVD markers in patients with CAA-ICH, which warrants further investigation.

The strength of our study is the systematic assessment of CSVD markers associated with HA and CAA in ICH patients, as well as the use of validated CSVD scores to quantify the overall burden of HA-CSVD and CAA-CSVD. Our findings may help to guide efforts in understanding if hypertensive small vessel disease plays a role in CAA in patients with primary ICH. However, there are several limitations in this study. First, although CSVD markers and the combined CSVD score could help to identify the etiology of CSVD, the lack of neuropathological examination is a major limitation of the present study. Fazekas et al. (9) performed a pathologic examination and postmortem MR imaging in ICH patients, and found that patients with lipofibrohyalinosis but without amyloid angiopathy on pathology could also have cortico-subcortical microbleeds and WMH. Therefore, the component of CAA-CSVD score, such as WMH and lobar CMB could also potentially reflect a severe hypertensive arteriosclerosis. Future studies with pathological evidence of CAA or amyloid-PET imaging are warranted to validate our findings. Second, we only included patients with brain MRI for CSVD study, potentially limiting the generalizability of our results to those with large hematoma volume or severe stroke conditions precluding MRI acquisition. Finally, According to both the classic Boston criteria and modified Boston criteria (19), the presence of multiple hemorrhages in lobes (cerebellar hemorrhage allowed), in together with an age of ≥55 years could be diagnosed with probable CAA. In addition, findings by Pasi et al. (9) suggest that superficial cerebellar ICH might be associated with strict lobar CMB, which is an established marker of CAA. However, cerebellum is also a cite of hypertensive hemorrhage (26). Pasi et al. (25) found that pathologically confirmed CAA was identified in 96% of patients with superficial cerebellar CMBs and in 25% of those with deep/mixed cerebellar CMBs. Therefore, there might be a distinction in etiology between superficial and deep cerebellar CMBs. Future studies with a specific focus on lobar CMBs or with an investigation that assesses lobar CMBs along with superficial cerebellar CMB are needed to confirm our results.

In summary, our study identified a significant association between HA-CSVD burden and CAA-CSVD burden in ICH patients. In addition, HA-CSVD score was correlated with CAA-CSVD score or lobar CMB severity when stratified by age and ICH etiology. These findings suggest that there might be a synergistic effect between HA and CAA in primary ICH, which needs further investigation in future studies.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by The Ethics Committee on Biomedical Research at West China Hospital of Sichuan University. The studies were conducted in accordance with the local legislation and institutional requirements. The ethics committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants’ legal guardians/next of kin because of the retrospective chart review.

YZ: Writing – original draft, Writing – review & editing, Formal analysis, and Data curation. LL: Writing – original draft, Writing – review & editing, and Data curation. LZ: Writing – review & editing and Data curation. YC: Writing – review & editing and Data curation. SZ: Writing – review & editing. BW: Writing – review & editing. DW: Writing – review & editing, Funding acquisition, Conceptualization, and Supervision. MX: Writing – review & editing, Funding acquisition, Conceptualization, Supervision, Methodology, and Project administration.