Cortical Microinfarcts Are Related to Amyloid Pathology and Cognitive Decline in Spontaneous Intracerebral Hemorrhage

Article information

J Stroke. 2026;28(2):326-329

Publication date (electronic) : 2026 May 28

doi : https://doi.org/10.5853/jos.2025.04308

Pu-Tien Chiang^1,2,3, Hsin-Hsi Tsai^,3, Bo-Ching Lee⁴, Chia-Ju Liu⁵, Syu-Jyun Peng⁶, Sheng-Sian Lin², Jiann-Shing Jeng³, Li-Kai Tsai³

¹Graduate Institute of Clinical Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan

²Department of Neurology, National Taiwan University Hospital Bei-Hu Branch, Taipei, Taiwan

³Department of Neurology, National Taiwan University Hospital, Taipei, Taiwan

⁴Department of Medical Imaging, National Taiwan University Hospital, Taipei, Taiwan

⁵Department of Nuclear Medicine, National Taiwan University Hospital, Taipei, Taiwan

⁶In-Service Master Program in Artificial Intelligence in Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan

Correspondence: Hsin-Hsi Tsai Department of Neurology, National Taiwan University Hospital, No.7, Zhongshan S. Rd., Zhongzheng Dist., Taipei 100225, Taiwan (R.O.C.) Tel: +886-2-23123456 ext. 262144 E-mail: hsinhsi@ntu.edu.tw

Received 2025 September 1; Revised 2025 December 23; Accepted 2026 March 26.

Dear Sir:

Cognitive impairment is common after intracerebral hemorrhage (ICH) (approximately 40% at 3–6 mon), with many cases declining over subsequent years [1]. The risk is the highest in lobar ICH from cerebral amyloid angiopathy (CAA). The heavy burden of lobar cerebral microbleeds (CMBs) and disseminated cortical superficial siderosis (cSS) predict cognitive decline [2], indicating a crucial role of small-vessel pathology, especially CAA, in longterm cognitive outcomes and demonstrating the need for imaging markers. Cortical cerebral microinfarcts (CMIs) represent an important yet under-recognized form of small-vessel brain injury [3,4]; evidence links CMIs with CAA and cognitive decline [5], yet prospective in vivo data integrating amyloid positron emission tomography (PET) with longitudinal outcomes in ICH survivors remain scarce. Therefore, we evaluated whether CMIs are related to amyloid pathology and independently predict long-term cognitive decline after spontaneous ICH.

We prospectively enrolled 169 ICH survivors who underwent standardized brain magnetic resonance imaging (MRI) and ¹¹CPittsburgh compound B (PiB) PET [6], with the global cortical standardized uptake value ratio (SUVR) calculated using the cerebellar cortex as a reference. PET was performed within 3 months of MRI, and the median interval from ICH onset to PET was 6.8 months. CMIs were identified by experienced readers blinded to PET results and cognitive outcomes, and were defined as lesions ≤4 mm that were T1 hypointense, fluid-attenuated inversion recovery (FLAIR) hyper- or isointense, and T2 hyperintense, visible on ≥2 orthogonal planes (Supplementary Figure 1) [7]. Baseline demographics, vascular risk factors, ICH characteristics, and small vessel disease (SVD) markers were systematically collected. Cognitive status was assessed at enrollment and during follow-up clinic visits; decline was defined as conversion to dementia, a ≥1-point increase in Clinical Dementia Rating score (for those already demented at baseline), or a ≥2-point decrease in Mini-Mental State Examination score [8]. Of 169 participants, 99 had ≥1 follow-up cognitive assessment and were included in longitudinal analyses (median follow-up, 2.8 yr). Details of the participant assessments, imaging acquisition and analysis, and statistical analysis are provided in Supplementary Methods.

Among the 169 survivors of spontaneous ICH (mean age, 66.6 ±11.5 yr; 62.7% male), 59.2% had lobar ICH and 30.8% met the Boston 2.0 criteria for probable CAA. CMIs were present in 63 participants (37.3%, termed “CMI(+)”): 34 (54.0%) had a single CMI, 27 (42.9%) had two to four, and two (3.2%) had more than five. The lobar distribution of the CMIs did not differ between the CAA and non-CAA groups (P=0.504) (Supplementary Figure 2). Inter- and intra-rater agreement for CMI identification were good (κ=0.70 and 0.68, respectively). Among the 19 participants scanned using 3D FLAIR, CMI detection was 57.9% compared with 52.6% detected using standard 2D sequences.

Participants with CMIs showed a higher global ¹¹C-PiB SUVR (median, 1.19 [interquartile range, 1.04–1.52] vs. 1.08 [1.02–1.22]; P=0.008), increased probable CAA frequency (41.3% vs. 24.5%, P=0.026), and a greater cSS prevalence (27.0% vs. 13.2%, P=0.039) than their CMI(-) counterparts. Other clinical and neuroimaging features (age, sex, vascular risk factors, antithrombotic use, hematoma volume, lobar/deep CMBs, white matter hyperintensity volume, lacunes, perivascular space burden, total SVD score, and cortical thickness) did not significantly differ (all P>0.05) (Table 1). The global ¹¹C-PiB SUVR increased with CMI burden: CMI(-), 1.08 [1.02–1.22]; single CMI, 1.11 [1.03–1.50]; multiple CMIs, 1.23 [1.09–1.53] (P=0.002 for multiple CMIs vs. CMI(–)) (Figure 1).

Table 1.

Demographic, clinical, and imaging characteristics of the survivors of ICH with and without CMIs

Figure 1.

Elevated amyloid burden in participants with CMIs. (A) Representative PiB positron emission tomography scan revealing cortical amyloid deposition (global SUVR=1.54). (B) Global PiB SUVR was significantly higher in participants with multiple CMIs compared with those without CMIs (P=0.002). PiB, Pittsburgh compound B; SUVR, standardized uptake value ratio; CMI, cortical cerebral microinfarct.

In multivariable logistic regression adjusted for age, sex, and usage of 3T MRI, CMI presence was associated with probable CAA (adjusted odds ratio, aOR, 2.26 [95% confidence interval, CI, 1.07–4.75], P=0.032), cSS (aOR, 2.40 [1.03–5.61], P=0.043), and higher global PiB SUVR (aOR, 1.22 [1.05–1.42] per 0.1-unit increase, P=0.010). For multiple CMIs, cSS (aOR, 2.78 [1.03–7.56], P=0.045) and PiB SUVR (aOR, 1.26 [1.05–1.52] per 0.1-unit increase, P=0.014) remained significant (Supplementary Table 1).

Among 99 participants with follow-up (median, 2.8 yr), cognitive decline occurred at 21.6 (95% CI, 13.8–33.8) versus 10.4 (6.8–15.9) per 100 person-years in CMI(+) versus CMI(-) groups (P=0.024) (Figure 2). CMIs independently predicted cognitive decline in Cox models adjusted for basic variables (age, sex, education, hematoma volume, and ICH location [lobar vs. nonlobar]) and additional covariates. Hazard ratio (HR) was 2.77 (95% CI, 1.30–5.86; P=0.008) when the model included probable CAA, mean cortical thickness, and total SVD score, and it was 2.58 (1.19–5.59, P=0.016) when global PiB SUVR replaced CAA (Supplementary Table 2). In models including presence of lobar CMB, both CMI presence (HR, 2.73; 95% CI, 1.21–6.18) and lobar CMB presence (HR, 6.39; 95% CI, 1.97–20.69) independently predicted cognitive decline, with no evidence of interaction (P=0.823).

Figure 2.

Kaplan–Meier analysis of cognitive decline in survivors of spontaneous ICH. Kaplan–Meier analysis of cognitive decline over 5 years stratified by the presence or absence of CMI. Participants with one or more CMI experienced a higher rate of cognitive decline than those without (P=0.024). CMI, cortical cerebral microinfarct; CMI(+), participants with one or more CMI(s); CMI(-), participants without CMIs.

This prospective cohort study shows that CMIs are common after spontaneous ICH, occurring in more than one-third of the survivors. CMIs are closely linked to amyloid-related small-vessel pathology, and their presence was associated with a higher likelihood of probable CAA, a greater cSS prevalence, and higher cortical amyloid deposition on ¹¹C-PiB PET. Importantly, CMIs independently predicted subsequent cognitive decline even after adjusting for amyloid markers, cortical atrophy, overall SVD burden, and hemorrhagic markers, including lobar CMBs.

Clinically, MRI-visible CMIs may help flag CAA in patients with mixed or uncertain ICH etiology [5,9]. In this cohort, probable CAA and cSS conferred more than double the odds of having CMIs, and every 0.1-unit increase in global PiB SUVR was associated with 22% higher odds of CMIs, suggesting that CMIs can expose an amyloid component even when hemorrhage-based criteria are inconclusive, which help stratify patients with mixed etiologies and clarify the often under-recognized ischemic CAA component.

Mechanistically, the cumulative burden of microscopic CMIs is associated with disruption of cortical network integrity and reduced cognitive reserve [3,10]. The independence of their association with the decline suggests unique tissue- and circuit-level damage (e.g., cortical–subcortical disconnection) that is not fully captured by conventional imaging markers and amyloid pathology [8]. To the best of our knowledge, this is the first study to directly link CMIs with the cortical amyloid burden using in vivo PET in a spontaneous ICH cohort, providing novel evidence that amyloid-mediated neuronal injury and microischemic damage are intertwined drivers of long-term cognitive outcomes.

The strengths of this study includes the use of in vivo amyloid PET, whereas its limitations include the inability of PiB to distinguish parenchymal from vascular amyloid, scanner heterogeneity (1.5 T/3 T), partial 3D FLAIR use, enrichment for lobar ICH, modest sample size, incomplete follow-up with worse baseline cognition among non-followed participants (potential selection bias), and lack of detailed cognitive phenotyping.

In conclusion, our findings demonstrate that CMIs are common in ICH survivors and are closely associated with amyloid pathology, particularly CAA. CMIs were independently associated with an elevated risk of cognitive decline after ICH and thus, may represent markers of both underlying SVD and future cognitive deterioration.