The Global Burden of Disease 2019 study ranked stroke as the primary cause of disability-adjusted life years in those aged 50 years and older (1). Intracerebral hemorrhage (ICH) is one of the most devastating and fatal subtypes of cerebrovascular accidents, and is associated with mortality rates of up to 40% (2). Approximately 30% of patients with ICH will experience hematoma expansion (HE) within 6 hours after the onset of neurological deficits (3), and HE has long been deemed as an independent predictor for worse outcomes (4). Every 1 mL increase in a hematoma has been estimated to elevate the risk of death by 5% (5). Additionally, attempts to prevent HE are crucial in managing the early phase of ICH, and this is a potential target for hemostasis treatment (6, 7). The timely detection of patients with ICH who have a high risk of hemorrhagic progression is therefore of great importance.

HE is associated with multiple factors, including Glasgow Coma Scale (GCS) score at admission, time from symptom onset to the first computed tomography (CT) scan, hemorrhage location, and initial hematoma volume (8). However, the relationship between hemorrhage location and HE deserves further investigation. Several studies have indicated that patient with deep ICH are more susceptible to HE than those with lobar ICH (9–11). Nevertheless, the inclusion of hemorrhage location may have restricted the applicability of HE differences between lobar and deep groups, which have a different etiology. Spontaneous ICH is overwhelmingly driven by hypertensive angiopathy, especially in deep zones, whereas cerebral amyloid angiopathy is generally cited as the primary reason for ICH at lobar sites (12). However, the heterogeneity between these sites means that there is no sufficient evidence to suggest that HE is more frequently observed in deep ICH than in lobar ICH. Additionally, the basal ganglia-thalamic region has long been viewed as the most frequent site of hemorrhage (7); focusing on deep ICH may help to determine the mechanisms underlying HE.

Thus, additional exploration of the relationship between deep ICH location and HE is warranted. We hypothesized that specific sites of deep ICH affect HE. The current study was performed to test this hypothesis.

In this relatively large retrospective multi-center study, we demonstrated a strong relationship between anatomic zones and HE in patients with deep ICH. Previously identified vital factors related to HE are larger baseline hematoma volumes, shorter time from onset to initial CT, lower GCS score at admission, and IVH (3, 16, 17). When controlling for these potential confounders in the multivariable logistic regression, we found that any deep ICH that affects the PLIC is still a vigorous driver of HE alone. Similarly, subgroup analyses including relevant variables revealed a similar result.

Many previous studies have focused on the occurrence of hematoma growth, with an emphasis on the ability to predict HE (3, 18, 19). That is to say, differences between deep and lobar hemorrhages, which may affect HE, were not taken into consideration. Deep hemorrhages were found to be overwhelmingly driven by hypertension, whereas cerebral amyloid angiopathy has been commonly cited as the primary reason for lobar locations, although hypertension was also a pivotal cause (12). In most previous studies (9–11), the implications of hematoma heterogeneity have typically been disregarded, which means that there has been no sufficient evidence to suggest that HE is more frequently observed in deep than in lobar groups. Therefore, we did not include patients with lobar hemorrhages in the present study.

It is increasingly urgent and imperative to identify HE, as growing evidence has revealed that deep sites contribute to worse clinical outcomes after ICH (7, 9, 11). The influence of deep HE on the clinical severity is dependent on the functional integrity of impaired fiber bundles and the specific anatomic regions of lesions. Damage to pyramidal tracts caused by HE of the internal capsule not only triggers physical disability but also gives rise to subjective discomfort and distress (20, 21). Once thalamic ICH occurs, there is a higher probability that blood will leak into the ventricles and that massive bleeding will add pressure to the brainstem, resulting in acute obstructive hydrocephalus and even rapid death (22). However, previous investigations into the latent association between deep locations and HE have been limited. Consequently, identification of the risk of HE based on the involvement of specific areas is critical for stratifying patients for therapeutic interventions.

The precise mechanism underlying the correlation between PLIC involvement and HE, however, remains unclear. It possible that specific anatomic characteristics provide the foundation for hemorrhagic progression. Since projection fibers are primarily located in the internal capsule, and are most abundant in the PLIC (23), blood to the PLIC region is supplied by branches from multiple arteries (24). One large study has also indicated that hematoma involving the PLIC is strongly correlated with a worse prognosis (13), and that HE itself is a critical predictor of ICH outcome (3), which underlines the potential link between the PLIC and HE. An additional possible explanation for our novel finding is that the comparatively larger contact area between the PLIC and lateral ventricle increases the risk of HE. Hematoma involving the PLIC might result in higher pressures, with effects spreading to the lateral ventricle. Deng et al. (25) also proposed that hematoma ventricle distance can predict poor outcomes in ICH, which is well-aligned with the current findings. It is puzzling that the thalamus, which is closer to the lateral ventricles, should theoretically be more prone to HE, but we found no evidence for this. Neisewander et al. (26) divided thalamic hemorrhage into six parts, and they found that posterior and lateral bleeds were less likely to cause neurological deterioration than global bleeds. However, our study is only a crude analysis of the involvement of thalamic hemorrhage, which may have obscured the natural conditions. As such, future studies focusing on the influence of specific regions of the thalamus on hematoma growth are warranted.

Our study has several strengths. We used data from three centers, which better exemplifies the real-world heterogeneity, with much more diverse and representative cases of ICH than single-center studies. Another advantage of our research lies in the accurate measurement of ICH volumes based on three-dimensional analysis, which reduces measurement errors caused by irregular hematoma and intraventricular hemorrhage.

However, the current investigation has some limitations. First, The patients were recruited retrospectively from the existing database, and so selection bias cannot be ruled out. Second, it is difficult to estimate the precise time of onset since neurological symptoms are found suddenly by eyewitness statements. If the actual onset was not witnessed, the time of onset of ICH symptoms was defined as the moment at which the patient was last seen to be normal, which may have resulted in recall bias. Finally, follow-up CT scans for a small portion of patients are determined by the actual clinical situation, which may underestimate the detection rate of real HE.

For patient with deep ICH, we suggest that more attention be paid to the hematoma location, because there is a strong association between hemorrhage involving the posterior limb of the internal capsule and HE.

The datasets presented in this article are not readily available because requests to access them must first be approved by the stroke centers that provided them. Requests to access the datasets should be directed to Yong-Qiang Yu, cjr.yuyongqiang@vip.163.com.

The Ethical Review Committees of Xiangyang Central Hospital, the First Affiliated Hospital of Anhui Medical University, and the Second Hospital of Anhui Medical University reviewed and approved this study. The data from three centers were anonymous and processed in a confidential manner. The request for written informed consent was waived due to the retrospective design of the present study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

LS and Y-QY involved in the study design, interpretation of data, and study supervision. LS contributed to data collection, data analysis, data management, and manuscript writing. T-TG contributed to the study design, manuscript review, and proofreading. X-MQ contributed to the data analysis and proofreading. HZ, L-SW, D-FT, HC, and H-BW responsible for data collection. Y-FF contributed to data analysis. H-QM involved in the data analysis and grammar. All authors contributed to the article and approved the submitted version.

This study was supported by grants from the National Natural Science Foundation of China, Grant/Award Numbers: 81771817, 81801679, and 82071905.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.