Cerebral Microbleeds and Long‐Term Affective Symptoms in Survivors of Spontaneous Intracerebral Hemorrhage

Abstract

Background

Spontaneous intracerebral hemorrhage (ICH) is strongly associated with the development of long‐term cognitive impairment and neuropsychiatric symptoms (NPS), particularly affective disturbances including depression and anxiety. Although small vessel disease (SVD) underlies most cases of ICH, the contribution of SVD to persistent affective symptoms and the effects of cognitive dysfunction on NPS remain unclear.

Methods

We prospectively recruited 95 survivors of spontaneous ICH (median, 5.7 years post‐ICH) between 2021 and 2025 from stroke clinics using standardized clinical protocols. NPS were assessed using the Neuropsychiatric Inventory Questionnaire. Cognitive status was evaluated; brain magnetic resonance imaging was performed to quantify SVD markers. Logistic regression and mediation analyses were used to examine the associations between SVD burden, affective symptoms, and cognition.

Results

Affective symptoms were the most prevalent NPS subsyndrome (37.9%), followed by apathy/vegetative symptoms (32.6%). Patients with affective symptoms showed greater cognitive impairment and higher hypertensive SVD burden, including higher numbers of deep cerebral microbleeds (4.0±5.9 versus 1.2±1.9, P=0.010) and lacunes (47.2% versus 23.7%, P=0.024) than those without affective symptoms. In the multivariable model, deep cerebral microbleeds (adjusted odds ratio, 1.215 [95% CI, 1.03–1.43], P=0.021), but not lacunes, were independently associated with affective symptoms. Mediation analysis revealed that cognitive impairment partially mediated this relationship (indirect effect P=0.012; 35.9% of total effect).

Conclusions

Affective symptoms are common long‐term outcomes after ICH; deep perforator arteriopathy, especially with the form of deep cerebral microbleeds, contribute to these symptoms; cognitive dysfunction only partly mediates this association. There is a need to screen NPS and evaluate microvascular pathology to assess emotional health post ICH.

Nonstandard Abbreviations and Acronyms

CAA

cerebral amyloid angiopathy

CMBs

cerebral microbleeds

ICH

intracerebral hemorrhage

NPS

neuropsychiatric symptoms

PVS

perivascular spaces

SVD

small vessel disease

Clinical Perspective.

What Is New?

• This study demonstrates that deep cerebral microbleeds—a key marker of hypertensive small vessel disease—are significantly associated with long‐term affective symptoms in survivors of intracerebral hemorrhage, with cognitive impairment acting as a partial mediator of this relationship.

What Are the Clinical Implications?

• These findings underscore the critical importance of evaluating microvascular pathology and cognitive dysfunction in post‐intracerebral hemorrhage affective outcomes and support the integration of neuropsychiatric screening into standard long‐term stroke care strategies.

Spontaneous intracerebral hemorrhage (ICH) accounts for approximately 6.5% to 19.6% of all strokes worldwide. ¹ , ² , ³ According to previous studies, survivors of ICH have a high risk of recurrent hemorrhage, ischemic stroke, and other vascular events, which contribute to substantially reduced long‐term survival. ⁴ , ⁵ Given these clinical risks, cognitive and psychological sequelae among this population often remain underrecognized, especially in the chronic stage. Emerging evidence indicates that nearly half of survivors of ICH continue to experience neuropsychiatric symptoms (NPS) beyond 6 months post event. ⁶ Affective symptoms—particularly depression and anxiety—are frequently reported and may substantially hinder functional recovery and long‐term outcomes. ⁶ , ⁷

Most cases of spontaneous ICH are attributable to underlying cerebral small vessel disease (SVD). ⁸ , ⁹ Two primary subtypes of SVD contribute to ICH: hypertensive arteriopathy, typically associated with deep hemorrhages, and cerebral amyloid angiopathy (CAA), which predominantly causes lobar or cortical hemorrhages. ¹⁰ Both of these processes represent chronic, progressive vasculopathies that cause diffuse brain injury beyond the acute hemorrhage insult. Neuroimaging markers of SVD—including cerebral microbleeds (CMBs), white matter hyperintensities, lacunar infarcts, and enlarged perivascular spaces—accumulate over time and have been linked to affective symptoms and cognitive dysfunction. ¹¹ , ¹²

Despite the high prevalence of NPS and cognitive dysfunction after ICH, ⁶ , ¹³ , ¹⁴ the relationship between chronic SVD pathology and these long‐term neurobehavioral outcomes remains poorly understood. Prior studies have reported that the prevalence of NPS increases with the severity of cognitive impairment, ¹⁵ suggesting that neuropsychiatric manifestations could occur as a consequence of impaired cognition. These findings raise the possibility that cognitive deficits may not only coexist with—but also mediate—the emergence of affective symptoms. Due to this uncertainty, integrative models need to be developed to clarify how intrinsic SVD, cognitive dysfunction, and affective symptoms jointly influence long‐term recovery after ICH.

In the current study, we investigated the long‐term prevalence of NPS and its subsyndromes in survivors of spontaneous ICH. Neuroimaging biomarkers of SVD were explored to assess which subtypes of SVD are associated with long‐term affective symptoms. The secondary aim of this study was to elucidate whether the presence of affective symptoms is mediated by cognitive dysfunction after ICH.

METHODS

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request. The reporting of this study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Ethical Approval

This study was approved by the Institutional Review Board of National Taiwan University Hospital (202109120RINC, 202210006RIND, and 202310051RIND) and was conducted in accordance with the hospital’s ethical guidelines. Written informed consent was obtained from all participants or their legal representatives.

Patient Enrollment

We prospectively recruited patients with supratentorial ICH aged 20 years or older between 2021 and 2025 from the stroke clinics at National Taiwan University Hospital and the hospital’s Bei‐Hu Branch. Participants were eligible if they were in the chronic phase of ICH, defined as at 6 months post‐ICH onset. Exclusion criteria included visual impairment, hearing loss, aphasia, disturbances of consciousness, bedridden status, and inability to complete a magnetic resonance imaging (MRI) examination. A total of 95 patients were enrolled and underwent global cognitive screening and neuropsychological assessment. All patients also underwent brain MRI and apolipoprotein E genotyping. Moreover, a subgroup of participants (n=58) additionally participated in another clinical study that investigated the underlying amyloid pathology in ICH and therefore underwent 11‐C Pittsburgh Compound B positron emission tomography. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ The flow chart of patient enrollment is presented in Figure 1.

Figure 1. Patient enrollment flow chart.

ICH indicates intracerebral hemorrhage; and MRI, magnetic resonance imaging.

Cognitive and Neuropsychological Assessments

Cognitive screening was conducted using the Mini‐Mental State Examination. ²¹ In this study, a cutoff score of <24 was applied to define cognitive impairment. To assess NPS, we used the Neuropsychiatric Inventory Questionnaire, ²² which is a concise screening instrument designed to assess clinically significant NPS and its impact on caregivers. The Neuropsychiatric Inventory Questionnaire evaluates 12 neuropsychiatric domains, including delusions, hallucinations, agitation/aggression, dysphoria/depression, anxiety, euphoria/elation, apathy/indifference, disinhibition, irritability/lability, aberrant motor behaviors, nighttime behavioral disturbances, and appetite/eating disturbances. For the purposes of this study, participants or their caregivers were required to respond “yes” or “no” to each symptom domain. We categorized the NPS into 4 distinct subsyndromes ²³ : hyperactivity (agitation, disinhibition, or irritability), psychosis (delusions or hallucinations), affective symptoms (depression or anxiety), and apathy/vegetative symptoms (apathy, sleep disturbances, or appetite changes).

Neuroimaging

For participant without available MRI in 1 recent year, we arranged a brain MRI upon study enrollment. All brain MRI was performed using Siemens Verio, TIM, or mMR systems (Siemens Medical Solutions, Malvern, PA, USA) using protocols that included T1‐weighted imaging, T2‐weighted imaging, fluid‐attenuated inversion recovery imaging, susceptibility‐weighted imaging, diffusion‐weighted imaging, apparent diffusion coefficient maps, and 3‐dimensional T1‐weighted magnetization‐prepared rapid acquisition gradient echo imaging at 1‐mm slice thicknesses. MRI was used to assess the locations of the ICH and markers of SVD according to the Standards for ReportIng Vascular Changes on Neuroimaging 2.0 criteria. ²⁴ Briefly, the number of CMBs and the presence of cortical superficial siderosis were evaluated using axial susceptibility‐weighted imaging sequences. The number of CMBs in the lobar (ie, the frontal, temporal, parietal, occipital, and insular cortices) and deep regions (ie, the brainstem, basal ganglia, thalamus, internal capsule, external capsule, corpus callosum, and deep periventricular white matter) were counted. Lacunes were evaluated in the supratentorial region and defined as “round or ovoid, subcortical, fluid‐filled cavities, ranging from 3 to 15 mm in diameter.” White matter hyperintensity volume was calculated based on fluid‐attenuated inversion recovery imaging of the ICH‐free hemisphere and multiplied by 2, as we previously proposed. ¹⁹ MRI‐visible perivascular spaces (PVSs) were evaluated on T2‐weighted imaging, and defined as sharply delineated structures measuring <3 mm following the course of perforating or medullary vessels. High‐degree PVS was defined as >20 visible PVS in the centrum semiovale or basal ganglia on the side of the brain with more severe involvement. Two composite scores, the total SVD score (range 0–4) ²⁵ and total CAA scores (range 0–6) ²⁶ were calculated to assess the total SVD burden attributed to hypertensive arteriopathy and CAA, respectively.

For patients for whom 11‐C Pittsburgh Compound B positron emission tomography scans were available, amyloid burden was quantified using the whole cortex standardized uptake value ratio, with the cerebellar gray matter serving as the reference region, as we previously described. ²⁰

Statistical Analysis

We analyzed the presence of NPS to determine their prevalence and to characterize symptom subtypes. Demographic and neuroimaging variables were compared between patients with and without affective subsyndrome. For categorical variables, Pearson’s χ² test were used when all expected cell counts were ≥ 5 and Fisher’s exact test were used otherwise. The independent samples t test were used for normally distributed continuous variables, and the Mann–Whitney U test were used for nonnormally distributed continuous variables. Categorical variables are reported as percentages, and continuous variables are presented as means ± SD or medians with interquartile ranges, as appropriate.

We examined the associations between the affective subsyndrome and neuroimaging variables using univariate and multivariate logistic regression analyses. Variables that showed significant differences between the 2 groups were included as independent variables, which were entered into separate regression models to avoid collinearity and model overfitting, with the presence of affective subsyndrome serving as the dependent variable. The models were adjusted for age, sex, years of education, and ICH location to account for potential confounders.

Additionally, mediation analyses were conducted to explore whether specific variables mediated the relationship between neuroimaging biomarkers and affective subsyndrome. Indirect effects were estimated using the delta method (Sobel test).

All statistical analyses were performed using SPSS software, version 18.0 (SPSS Inc., Chicago, IL, USA). A 2‐tailed P value <0.05 was considered statistically significant.

RESULTS

Prevalence of Long‐Term Neuropsychiatric Subsyndromes in Survivors of ICH

A total of 95 survivors of ICH were enrolled in this study, and the median time since their ICH event was 5.7 years (interquartile range, 2.7–8.7). The prevalence of each NPS subsyndrome is shown in Figure 2A. Affective symptoms were most prevalent (37.9%), followed by apathy/vegetative symptoms (32.6%) and hyperactivity symptoms (22.1%). Psychotic symptoms were rare, observed in only 3.2% of participants. Detailed distributions of symptom severity across NPS domains are provided in Table S1. The NPS profiles were similar between the groups with lobar versus deep ICH (Figure 2B). The frequency of each of the 4 NPS subsyndromes was significantly higher in patients with impaired cognition compared with patients with normal cognition, as follows: affective symptoms (84.6% versus 30.5%, P < 0.001), apathy/vegetative symptoms (69.2% versus 26.8%, P=0.004), hyperactivity (61.5% versus 15.9%, P=0.001), and psychosis (23.1% versus 0%, P=0.002; Figure 2C).

Figure 2. Long‐term distribution of subsyndromes of NPS in survivors of ICH.

A, Radar plot illustrating the distribution of the four subsyndromes of NPS. B, The distribution of NPS subsyndromes was similar between patients with lobar and deep ICHs. C, The prevalence of all subsyndromes of NPS was higher among patients with impaired cognition compared with patients with normal cognition (affective symptoms, 84.6% vs 30.5%, P<0.001; apathy/vegetative symptoms, 69.2% vs 26.8%, P=0.004; hyperactivity, 61.5% vs 15.9%, P=0.001; psychosis, 23.1% vs 0%, P=0.002). ICH indicates intracerebral hemorrhage; and NPS, neuropsychiatric symptoms.

Clinical and Neuroimaging Characteristics Associated With Affective Symptoms

We next compared the clinicopathological features of survivors of ICH with and without affective symptoms (Table 1). Participants with affective symptoms had similar age and sex distributions to those without affective symptoms but had significantly fewer years of education (12.1 ± 3.8 versus 14.3 ± 3.1 years, P=0.003) and a higher proportion of global cognitive impairment (defined as a Mini‐Mental State Examination score <24; 30.6% versus 3.4%, P < 0.001), indicating greater overall cognitive impairment. There was no significant difference in the frequency of ApoE4 carriage between the groups of patients with and without affective symptoms (8.3% versus 11.9%, P=0.737).

Table 1.

Comparison of Clinical and Imaging Features Between Patients With and Without Affective Symptoms

	With affect (n=36)	Without affect (n=59)	P value
Male sex, n (%)	28 (77.8%)	40 (67.8%)	0.295
Age, y, mean±SD	63.2 (11.9)	60.6 (10.0)	0.328
Education, y (SD)	12.1 (3.8)	14.3 (3.1)	0.003*
Follow‐up time after ICH, mo (IQR)	77 (22–98)	86 (36–109)	0.380
Prior ischemic stroke, n (%)	6 (16.7%)	8 (13.6%)	0.679
≥2 prior strokes, n (%)	9 (25.0%)	9 (15.3%)	0.240
Hypertension, n (%)	34 (94.4%)	52 (88.1%)	0.308
Diabetes, n (%)	12 (33.3%)	19 (32.2%)	0.909
Hyperlipidemia, n (%)	18 (50.0%)	29 (49.2%)	0.936
Mini‐Mental State Examination score <cutoff 24, n (%)	11 (30.6%)	2 (3.4%)	<0.001*
ApoE4 carrier, n (%)	3 (8.3%)	7 (11.9%)	0.737
Probable CAA, n (%)	2 (5.6%)	5 (8.5%)	0.706
Lobar ICH, n (%)	15 (41.7%)	23 (39.0%)	0.796
Deep ICH, n (%)
Thalamic ICH	7 (19.4%)	7 (11.9%)	0.312
Basal ganglia ICH	11 (30.6%)	26 (44.1%)	0.190
Presence of lacunes, n (%)	17 (47.2%)	14 (23.7%)	0.018*
Presence of cortical superficial siderosis, n (%)	0 (0%)	2 (3.4%)	0.524
Presence of CMBs, n (%)	24 (66.7%)	34 (57.6%)	0.381
Number of CMBs, mean±SD
Lobar CMBs	4.8 (8.6)	4.1 (13.8)	0.782
Deep CMBs	4.0 (5.9)	1.2 (1.9)	0.010*
Cerebellar CMB	0.3 (1.4)	1.0 (5.1)	0.440
White matter hyperintensity volume, mL (IQR)	7.4 (2.6–17.7)	3.8 (1.9–9.3)	0.062
Magnetic resonance imaging‐visible enlarged perivascular spaces, n (%)
Basal ganglia (>20)	17 (47.2%)	17 (28.8%)	0.069
Centrum semiovale (>20)	12 (33.3%)	16 (27.1%)	0.519
Total small vessel disease score, mean±SD	1.9 (1.3)	1.4 (1.3)	0.062
CAA score, mean±SD	1.4 (1.1)	1.1 (1.4)	0.268
11‐C Pittsburgh Compound B whole cortex standardized uptake value ratio, mean±SD1	1.09 (0.13)	1.07 (0.09)	0.505

CAA indicates cerebral amyloid angiopathy; CMBs, cerebral microbleeds; ICH, intracerebral hemorrhage; IQR, interquartile range.

¹ n=58 was available; *P < 0.05.

Based on the neuroimaging assessment, patients with affective symptoms exhibited more severe hypertensive SVD, including a greater burden of deep CMBs (4.0 ± 5.9 versus 1.2 ± 1.9, P=0.010) and a higher prevalence of lacunes (47.2% versus 23.7%, P=0.018) than patients without affective symptoms. Additionally, there was a trend toward a higher white matter hyperintensity volume (7.4 [2.6–17.7] versus 3.8 [1.9–9.3] mL, P=0.062), more severe PVS in the basal ganglia (47.2% versus 28.8%, P=0.069), and a higher total SVD score (1.9 ± 1.3 versus 1.4 ± 1.3, P=0.062) in patients with affective symptoms compared with those without. We did not observe significant differences in the proportions of patients with amyloid pathology, specifically the prevalence of probable CAA, the severity of total CAA score or the cerebral amyloid burden measured by positron emission tomography, between the groups of patients with and without affective symptoms (both P > 0.05), suggesting that amyloid‐related neurodegeneration was unlikely to account for the observed associations. A post hoc sensitivity analysis based on the observed group difference in deep CMB burden (Cohen’s d=0.70; n₁=36, n₂=59; α=0.05) yielded a power of 0.905, indicating adequate sensitivity to detect effects.

Associations Between Hypertensive SVD, Cognitive Dysfunction, and Affective Symptoms in Survivors of ICH

To further investigate the associations between hypertensive SVD markers and affective symptoms, we constructed logistic regression models to predict the occurrence of affective symptoms based on relevant neuroimaging markers (lacunes, deep CMBs, white matter hyperintensities volume, PVS in the basal ganglia, total SVD score). In univariable logistic regression, the presence of lacunes (odds ratio [OR], 2.88 [95% CI, 1.18–6.99], P=0.020) and a higher number of deep CMBs (OR, 1.27 [95% CI, 1.07–1.50], P=0.007) were significantly associated with affective symptoms (Table 2, Model 1). After adjustment for age, sex, years of education, and ICH location, deep CMBs remained a significant risk factor for the affective subsyndrome (OR, 1.22 [95% CI, 1.03–1.43], P=0.021); however, lacunes did not remain significant after adjustment (P=0.175, Table 2, Model 2). In a sensitivity analysis further adjusting for hypertension and diabetes, the association between deep CMBs and affective symptoms persisted (OR, 1.21 [95% CI, 1.03–1.44], P=0.024).

Table 2.

Logistic Regression Models of Imaging Parameters Associated With Affective Symptoms

Parameter	Model 1 (univariable)		Model 2 (age, sex, y of education, and ICH location‐adjusted)#
	OR (95% CI)	P value	OR (95% CI)	P value
Presence of lacunes	2.88 (1.18–6.99)	0.020*	1.97 (0.74–5.23)	0.175
Deep cerebral microbleeds	1.27 (1.07–1.50)	0.007*	1.22 (1.03–1.43)	0.021*
White matter hyperintensities volume	1.03 (0.99–1.06)	0.128	1.00 (0.97–1.05)	0.692
Perivascular spaces‐basal ganglia	2.21 (0.93–5.24)	0.072	2.04 (0.71–5.85)	0.186
Total small vessel disease score	1.36 (0.98–1.88)	0.065	1.21 (0.83–1.78)	0.319

ICH indicates intracerebral hemorrhage; and OR, odds ratio.

^* P < 0.05.

^# Each imaging variable was entered into a separate multivariable logistic regression model to avoid potential multicollinearity.

As cognitive impairment was more prevalent in patients with affective symptoms and has been reported to be related to intrinsic SVD, ²⁷ we investigated whether impaired cognition mediates the relationship between deep CMBs and affective symptoms (Figure 3). Deep CMBs were significantly associated with the presence of cognitive impairment (Path a: β=0.026, SE=0.008 [95% CI, 0.010–0.043], P=0.002). Additionally, impaired cognition was significantly associated with affective symptoms (Path b: β=0.541, SE=0.135 [95% CI, 0.273–0.810], P < 0.001), indicating a robust link exists between cognitive impairment and the presence of affective symptoms.

Figure 3. Mediation analysis of the associations between deep CMBs, cognitive impairment, and affective symptoms.

Mediation analysis showed that the association between deep CMBs and affective symptoms was partially mediated by cognitive impairment. Paths a and b were both significant (P=0.002 and <0.001, respectively), as was the direct effect (c’: β=0.027, SE=0.021 [95% CI, 0.005–0.050], P=0.021). The indirect effect (a×b=0.014) was also significant (Sobel z=2.52, P=0.012), supporting a partial mediation effect of 35.9%. CMB indicates cerebral microbleed.

We next investigated whether the relationship between deep CMBs and affective symptoms was mediated by the presence of cognitive impairment. The direct effect of deep CMBs on affective symptoms was statistically significant (Path c’: β=0.027, SE=0.012 [95% CI, 0.004–0.050], P=0.021), suggesting that deep CMBs independently contribute to affective symptoms. Our mediation analysis indicated the indirect effect of deep CMBs on the affective symptoms through impaired cognition (Path a × b) was statistically significant (β=0.014, Sobel z=2.52, P=0.012) and supported a partial mediation effect of 35.9%.

DISCUSSION

We investigated the prevalence and neuroimaging correlates of long‐term NPS in survivors of spontaneous ICH. Among the 4 NPS subsyndromes, affective symptoms were the most prevalent, observed in 38% of patients. Affective symptoms, including depression and anxiety, were significantly associated with the presence of deep CMBs, indicating a close link to hypertensive SVD—but not to amyloid pathology. Additionally, cognitive impairment was more frequent in survivors of ICH with affective symptoms than those without affective symptoms. Cognitive impairment partially mediated, but did not fully account for, the association between deep CMBs and long‐term affective symptoms. Collectively, our findings suggest that NPS profiles need to be more clearly characterized in survivors of ICH considering the high prevalence of these symptoms and the diverse subsyndromes. A significant association was found between hypertensive SVD, especially deep CMBs, and affective symptoms, and this association was only partly mediated by cognitive impairment. Thus, neuropsychiatric manifestations should be considered as a separate entity among the long‐term neuropsychological consequences of the intrinsic SVD in survivors of ICH.

Few studies have investigated the long‐term emotional disturbances in survivors of ICH. In our cohort, which was assessed at a median of 5.7 years post ICH, nearly half (49.5%) of participants exhibited at least 1 NPS, with affective symptoms being the most prevalent, followed by apathy/vegetative symptoms. These findings closely align with the results of a previous study, ⁶ which assessed NPS at 6 months after ICH and reported a similar overall prevalence of NPS (55%) and an identical prevalence of affective symptoms (38%). Another recently published large cohort study from China reported long‐term affective disturbances in survivors of minor ICH. After a median follow‐up of approximately 3.5 years post ICH, more than half (53%) of subjects had depression and 39% of subjects experienced anxiety. ¹³ Overall, our study and these reports consistently found that affective symptoms are prevalent in survivors of ICH and may persist for years after the ICH.

Our study identified an association between a higher SVD burden, particularly deep CMBs, among the group of survivors ICH with affective symptoms. Similar results were observed in a previous study, ¹² which showed that the presence of CMBs was significantly associated with an increased risk of developing new‐onset depression in survivors of ICH. Specifically, the presence of ≥1 deep CMB was associated with an 18% increased risk of developing depression. This association between deep CMBs and depression was not only noted in patients who had suffered a stroke but also in a stroke‐free cohort. ²⁸ Collectively, these findings support the “vascular depression” hypothesis, ²⁹ , ³⁰ which suggests there is an interplay between vascular pathology and affective symptoms. Our study further expands this hypothesis to survivors of ICH, as these patients tend to exhibit severe, intrinsic SVD, and long‐term depression after ICH was mainly driven by deep perforator arteriopathy pathologies rather than CAA. However, this pattern contrasts with a previous study, ¹² which reported stronger associations between CAA‐related markers—particularly lobar CMBs—and depression in survivors of ICH. One potential explanation for this difference may be the systematic differences between the study populations. In the previous cohort, a higher proportion (56%) of participants met the criteria for probable CAA, whereas only 7.4% of participants in this study were classified as having probable CAA. The underrepresentation of CAA in our study may have resulted in different associations between lobar CMBs and affective symptoms.

There is a close relationship between impaired cognitive function and mood alterations, as demonstrated by our studies and others’ research. ⁶ , ¹³ , ³¹ The group of survivors of ICH without affective symptoms had higher levels of education and had lower rates of cognitive impairment post ICH. This observation is consistent with the cognitive reserve hypothesis, ³² , ³³ which suggests that individuals with greater cognitive reserve are better able to use compensatory neural mechanisms to withstand brain injury, leading to more favorable clinical outcomes. ³⁴ As intrinsic SVD is linked to the risk of cognitive dysfunction after ICH, ²⁷ , ³³ one interesting finding of our study is the partial mediation effect of cognitive impairment on the relationship between SVD and affective symptoms. The mediation effect size of cognition (35.9%) was modest, yet it suggests that mood disturbances and cognitive decline are partly intertwined, which may be especially relevant in patients with deep CMBs. This finding highlights the clinical importance of addressing both domains together rather than treating them in isolation. Another important finding of our study is that the independent association between SVD and affective symptoms was not mediated by cognitive impairment. The presentation of mood disorders should therefore be considered as a separate long‐term sequela of ICH. Depressive disorders are often associated with reduced metabolic activity in the frontal cortex (particularly the orbitofrontal cortex and dorsolateral prefrontal cortex) and the caudate nucleus. ³⁵ , ³⁶ , ³⁷ CMBs can cause localized tissue damage within this circuitry and thereby may disrupt the normal functioning of the fronto‐subcortical circuits and potentially lead to neuropsychiatric disorders. Although this “vascular depression” pathway provides a compelling explanation, affective symptoms after ICH are also likely influenced by a broader neuropsychosocial context. In addition to vascular pathology, reduced cognitive reserve, lack of social and environmental support, other medical comorbidity or medication effect could also play an important role leading to the long‐term affective symptoms.

There are several limitations to this study. First, the sample was not fully representative; most participants had mild cognitive impairment (42%) or subjective cognitive decline (40%), whereas survivors who progressed to moderate‐to‐severe dementia were not included. In addition, 11‐C Pittsburgh Compound B positron emission tomography data were available only for a subset of participants, and the assessment of amyloid pathology was not available in the remaining patients. These factors limit the generalizability of our findings, which are most applicable to long‐term survivors without overt dementia and may also lead to underestimation of NPS prevalence and their associations with SVD markers. Second, the cross‐sectional design allows identification of associations between neuroimaging markers and NPS but does not establish causality. The interval between ICH onset and subsequent neuropsychiatric assessment also varied across participants, although the follow‐up time was comparable between those with and without affective symptoms, suggesting timing was unlikely to drive group differences. Third, cognition was primarily assessed with Mini‐Mental State Examination, which may not capture the domain‐specific cognitive deficits; the NPS were classified in a binary manner, whereas the symptom severity or functional impact was not considered in the current analyses. A more comprehensive phenotyping of cognition and neuropsychiatric symptoms, together with advanced neuroimaging biomarkers are needed in future studies. We did not collect volumetric data, including the total intracranial volume, or assess the severity of cortical atrophy, which may provide further insights on how neurodegeneration and its related markers affect the presentation of NPS. Finally, as already mentioned, of the various SVD pathologies in ICH, CAA was underrepresented in our cohort and most participants had hypertensive SVD. The sample size was modest and was determined by clinical feasibility; therefore, no a priori power calculation was performed and the analyses were mainly exploratory. Accordingly, no formal adjustment for multiple comparisons or interaction testing was performed, and nonsignificant findings for some markers (eg, lacunes, PVS‐basal ganglia, total SVD score) should be interpreted with caution. Future cohorts with larger sample sizes and longitudinal design are warranted to clarify causal relationships and the long‐term trajectory of NPS in survivors of ICH.

Conclusions

In conclusion, our findings confirm that survivors of ICH have a high risk of long‐term NPS, especially affective symptoms, even years after the index stroke. We identified a significant relationship between deep perforator arteriopathy and affective symptoms, which reinforces the importance of assessing microvascular contributions to neuropsychiatric outcomes. In addition, we found that the association between affective symptoms and deep CMBs was partly mediated by cognitive dysfunction. Thus, future longitudinal studies of NPS outcomes in survivors of ICH are essential to better elucidate the longitudinal trajectory and the underlying mechanisms that link cerebrovascular pathology and neuropsychiatric disorders. Moreover, NPS should also be considered as an important long‐term outcome measure when introducing new treatments for patients who have suffered an ICH.

Sources of Funding

This work was supported by grants from Taiwan National Science and Technology Council (Tsai HH, 112‐2923‐B‐002‐001‐MY3 and 113‐2628‐B‐002‐013‐MY3), National Taiwan University Hosptial (Tsai HH, 114‐SS0015; Lee BC, 114‐S0123) and National Taiwan University Hospital, Bei‐Hu Branch (Tsai HH, 11106; Lin SS, 11205 and 11 306).

Disclosures

None.

Supporting information

Table S1