More than 5 million events of incident hemorrhagic stroke occur annually worldwide (1), and less than 50% of patients with intracerebral hemorrhage (ICH) survive to 1 year (2). Theoretically, surgical hematoma evacuation is intuitively appealing to reduce the mass effect and secondary cytotoxic injury of ICH (3, 4). However, positive effects may be offset by surgical trauma. Randomized controlled trials (RCTs) have shown that standard surgical procedures (open craniotomy) have not demonstrated improved functional outcomes or mortality in patients with ICH (5, 6). In recent years, surgical and radiologic technology have made minimally invasive surgery (MIS) possible (7). A meta-analysis has demonstrated that MIS for hematoma evacuation could decrease the rate of moderate to severe functional impairment and mortality in patients with supratentorial ICH (8). In addition, the results of minimally invasive catheter evacuation followed by thrombolysis (MISTIE) III trial showed that MISTIE group had a significant lower serious adverse events rate than standard medical care group, while the occurrence rates of symptomatic bleeding and brain bacterial infections were similar between two groups (9). Therefore, the bright prospect of MIS in ICH patients has renewed interest in ICH surgery.

Seizures are a common complication after intracerebral hemorrhage (ICH), affecting up to 6%–15% of patients with ICH in the acute and long-term periods (10, 11). Seizures after ICH can hurt the poststroke outcome and quality of life (12, 13); therefore, identifying individuals at high risk of seizures after a stroke and preventing epilepsy in a timely manner are crucial.

Previous studies have shown several risk factors for late seizures after ICH, such as cortical involvement, early seizures, a larger hemorrhage volume, a younger age of ICH onset, conventional surgical interventions, intraventricular hemorrhage, and the NIHSS score (10, 14–18). However, the association between minimally invasive hematoma aspiration and thrombolysis and late seizures after ICH remains uncertain. Therefore, in this study, we aimed to investigate whether MIS increases the risk of late seizures after ICH and use propensity score matching to eliminate interference from other risk factors. Additionally, a case–control study of MIS-treated patients with and without late seizures was performed to identify the risk factors for late seizures after ICH in this patient subgroup.

Figure 1 shows the flowchart of the selection of the subjects based on the inclusion/exclusion criteria. A total of 1,055 and 875 patients eligible for this study were derived from TJH and WZH, respectively. Additionally, 241 patients received MIS for hematoma evacuation (MIS group) and 1,689 patients chose conservative medication (conservative treatment group). The median follow-up time for all subjects was 48 months (interquartile range [IQR] = 31–68 months). The rates of late seizures were 12.9% (31/241) and 4.1% (69/1689) in the MIS group and conservative treatment group, respectively.

Propensity score matching was applied to balance the demographic and clinically relevant variables between the MIS and conservative treatment groups; 161 ICH patients from the MIS group were successfully matched with 161 ICH patients from the conservative treatment group (1:1). The demographic and clinically relevant variables before and after matching between the groups are shown in Table 1. Before matching, significant differences (p < 0.05) were found between the MIS and conservative treatment groups in the age of ICH onset, comorbidities (hypertension, diabetes mellitus, hyperlipidemia, hyperuricemia and cardiac disorders), NIHSS score, GCS score, volume of ICH, cortical involvement, multilobar involvement, intraventricular hemorrhage and follow-up duration. After matching, no significant differences (p > 0.05) were found in the covariates of propensity score matching (age of ICH onset, history of alcohol abuse, NIHSS score, volume of ICH, cortical involvement, intraventricular hemorrhage, and early seizures) or the remaining variables.

Figure 2 compares the late seizure rates between the MIS and conservative treatment groups. Before matching, significant differences (p < 0.001) were found between the MIS group (31/241; 12.9%) and conservative treatment group (69/1689; 4.1%) in the incidence of late seizures. However, after matching, no significant differences (p = 0.854) were found between the MIS group (17/161; 10.6%) and conservative treatment group (16/161; 9.9%).

Among the participants in TJH, 96 patients (9.1%) received regular blood thinning/antiplatelet drugs during the follow-up. The rates of late seizures were 3.1% (3/96) in the patients receiving blood thinning/antiplatelet drugs, while 5.5% (53/959) in the patients without taking those drugs, and no significant differences (p = 0.317) were found between the two groups. In the MIS group, 31 patients (12.9%) had one or more seizures beyond 7 days post-ICH. Table 2 provides the baseline characteristics of the MIS group with and without late seizures. Univariate analysis showed that cortical involvement, NIHSS scores at admission and the volume of ICH were associated with late seizures (p < 0.1). Multivariate stepwise logistic regression analysis revealed that only cortical involvement (OR = 2.547; 95% CI = 1.137–5.705; p value = 0.023) and higher NIHSS scores at admission (OR = 1.050; 95% CI = 1.008–1.094; p value = 0.019) were independent risk factors for the occurrence of late seizures.

In the present study, the incidence of late seizures in ICH patients who received MIS was significantly higher than that in ICH patients who received conservative medication (12.9% vs. 4.1%; p value < 0.001). However, substantial differences were found between the two patient subgroups, including those with proven risk factors for late seizures after ICH. To eliminate these interferences, propensity score matching was performed to balance relevant risk factors between the MIS and conservative treatment groups. After matching, no significant differences were found in those with proven risk factors between the subgroups (Table 1), and no significant differences were found between the patients with MIS and those with conservative medication in the incidence of late seizures in the ICH patients (10.6% vs. 9.9%; p value = 0.854). Additionally, our study showed that a higher NIHSS score at admission and cortical involvement were independent risk factors for late seizures in ICH patients with MIS.

In recent years, several trials have shown that craniotomy does not improve the functional outcome or mortality of ICH, likely because the surgical approach creates sufficient trauma to the surrounding brain for hematoma evacuation (5, 6). Previous studies have explored the relationship between conventional surgical hematoma evacuation and late seizures (15, 17). A prospective study of spontaneous intracerebral hemorrhage found that conventional surgical hematoma evacuation was an independent risk factor for late seizures (OR = 2.6; 95% CI = 1.4–4.8; p value = 0.0027) (15). Another study showed that hematoma evacuation, which included standard craniotomy and MIS, increased the risk of late seizure after ICH (OR = 2.47; 95% CI = 1.12–5.45; p value < 0.05) (17). Additionally, other frequent complications reported in ICH patients with craniotomy were pulmonary (30.4%), renal (3.2%), and thromboembolic (2.9%) complications (23).

The failure of conventional surgical interventions and common complications leaves MIS as the most promising surgical strategy for ICH patients. MIS for ICH aims to reduce the manipulation of healthy brain tissue, decrease the procedure time and achieve rapid hematoma evacuation (24). A meta-analysis demonstrated that MIS is associated with lower rates of moderate-to-severe functional impairment and lower mortality than medical management or conventional surgery (25). Nevertheless, previous studies were mostly small, unblinded, and none assessed the association between surgical performance and clinical outcome. So additional evidence is needed to recommend use of minimally invasive surgery plus thrombolysis in routine care. MISTIE III was an open-label, blinded endpoint, phase 3 trial and the trial centrally randomized patients to MISTIE treatment group or standard medical care group (9). The result of MISTIE III trial showed that MISTIE did not improve functional outcomes compared with standard medical care in patients with large intracerebral hemorrhage. However, assessment of the surgical data indicates that when MISTIE achieves the protocol-defined surgical aim, which include large reduction of hematoma size to 15 ml or less, a benefit in mortality and functional improvement seems possible. More importantly, the MISTIE procedure was safe with regard to serious bleeding and infection, and MISTIE group had a significant lower serious adverse events rate than standard medical care group. These findings suggest MISTIE has few negative consequences. Our study showed that receiving MISTIE did not increase the incidence of late seizures after ICH, providing another evidence for safety of MISTIE.

The reported incidence of late seizures after ICH ranges from 3.1% to 10.8% (14–18, 26, 27). In this study, the incidence of late seizures in ICH patients with MIS was up to 12.9% (31/241); in patients with conservative medication, it was 4.1% (69/1689). The main reason for the high incidence of late seizures was that MIS-treated patients had more risk factors for late seizures (10, 14, 16), including significantly younger ages, higher NIHSS scores, a larger volume of ICH, and more frequent cortical involvement and intraventricular hemorrhage than those treated conservatively (Table 1). After eliminating interferences by propensity score matching, MIS for hematoma evacuation showed no significantly higher risk of late seizures than conservative medication (Figure 2). This result may be attributed to the reduction in excitotoxic glutamate concentrations and improvement in blood brain barrier integrity within the perihematoma region by MIS (28, 29).

In our analysis of the patients with MIS, the risk of developing late seizures after ICH is increased with cortical involvement and higher NIHSS scores. Cortical involvement has been consistently reported as a risk factor for post-ICH seizures (14, 15, 17, 26). Gliosis, neuronal loss, hemosiderin deposition, neurodegeneration and altered synaptic plasticity in neurons near the cortical area lead to greater hyperexcitability of synaptic activities than that in the deep part of the brain parenchyma (30). The association between cortical involvement and seizures may be attributed to the increase in neuronal excitability and disinhibition by altering the cortical structure and cortical compression (31). Likewise, several previous studies have already established higher NIHSS scores as a risk factor for late seizure development (16, 22, 32). In the current era of personalized medicine, the diagnosis or prognosis of disease has become increasingly significant (33). The CAVE score was the first tool to estimate the risk of late seizures after ICH in individual patients, and the score was derived using a 0–4 point scoring system with 1 point each for cortical involvement, age <65 years, volume >10 ml, and early seizures within 7 days of ICH (14). The c-statistic of the CAVE score was 0.81 (0.76–0.86) and 0.69 (0.59–0.78) in the internal and external validation cohorts, respectively. Among the clinical scores currently available, the CAVS score was the only score that included surgical variables, improving the CAVE score by substituting surgical evacuation for early seizures (15). Our study showed that MIS did not significantly increase the risk of late seizures; therefore, MIS should be separated from conventional surgery in terms of clinical scores instead of being a risk factor.

Our study has several limitations. First, as our study design was retrospective, identifying patients with early seizures relied on inpatient medical records. Although motor seizures could be correctly identified by the treating clinician and the researchers, there was less certainty around the recognition for patients with subtle, non-motor seizures. The latter events were excluded from the study, therefore, it inevitably led to the under-reporting early seizure events. In addition, registering seizure events retrospectively is prone to misdiagnosis and underdiagnosis, although the medical reports were collected based on a standardized questionnaire. Second, as the recognition of late seizures rested largely on patient and caregiver reports in the tele-interview, we did not have data on subclinical electrographic seizures. Therefore, the incidence of late seizures may have been underestimated. Given the high early mortality associated with ICH, competing risk from death not related to the outcome of interest is a potential problem in this study and represents an inevitable bias. Furthermore, the conservative treatment group consists of the patients who were not deemed eligible for surgery, so the potential bias of the propensity score matching is inevitable due to the absence of choosing patients randomly for surgery.Despite these limitations, our study provided an independent cohort including MIS-treated patients, as well as a large two-center control cohort including medication-treated patients for strict matching. To our knowledge, this study is the first on the association between MIS and late seizures after ICH.

In summary, our study revealed that receiving MIS did not increase the incidence of late seizures after ICH. Additionally, cortical involvement and NIHSS scores were independent risk factors for late seizures in MIS-treated patients.