The institutional ethics committee of our institution approved (No. KYLLK20240133) this study; we retrospectively reviewed the database of the stroke center in Suining Central Hospital, Sichuan, China, and the data of patients undergoing MT for LAO from January 2021 to May 2024 were screened. Patient selection was based on the following inclusion criteria: (1) patients were admitted within 24 h after disease onset; (2) patients with age more than 18 years; and (3) patients suffered from LAO in anterior circulation (internal carotid artery [ICA] or middle cerebral artery [MCA]). The exclusion criteria were as follows: (1) Patients with LAO in the posterior circulation (vertebral artery or basilar artery); (2) patients with apparent infection or immune disease at admission; (3) patients who had comorbidities that required hormone or immunosuppressive agent therapy; (4) patients with progressive or decompensated comorbidities before AIS onset; (5) patients without SII data at admission or outcomes at follow-up; and (6) patients with a history of stroke had residual apparent dysfunction before AIS onset (modified Rankin Scale [mRS] score ≥ 2). In addition, patients with multiple times of MT for re-occlusion of ICA or MCA within the same hospital stay were also excluded, because this may interfere with the predictive value of SII for outcomes.

A comprehensive evaluation was deployed immediately after patients were admitted to our emergency, including blood sample collection, consciousness assessment by Glasgow Coma Scale (GCS) score, neurological deficits assessment by the National Institutes of Health Stroke Scale (NIHSS) score, and neuroimaging test. The non-enhanced computed tomography (CT) of the brain was performed to rule out ICH, and CT angiography to confirm the presence of LAO. Intravenous thrombolysis with alteplase was administered within 4.5 h after AIS onset, and MT was carried out for patients with contraindication or futile recanalization of thrombolysis. For LAO with a time window of 4.5–24 h from onset, MT was performed after confirming the mismatch of ischemic and infarction areas evaluated by CT perfusion or MRI/DWI. All MT procedures were performed under general anesthesia using a modern stent retriever and/or direct aspiration technique. Stent implantation or balloon angioplasty was performed for stenosis or dissection at the surgeon’s discretion. The vessel recanalization was measured by modified thrombolysis in cerebral infarction (mTICI) score ≥ 2b was regarded as successful recanalization. Time from onset to recanalization was defined as the time from symptom appearance to successful recanalization or abortion of procedure if successful recanalization was not achieved (14). Planned cranial CT scans were conducted immediately after surgery, 24- and 48-h postprocedure. A CT was also taken whenever there was a change in neurological symptoms.

The following clinical and radiological data were collected: demographics, medical history, admission GCS score and NIHSS score, baseline Alberta Stroke Program Early Computed Tomography Score (ASPECTS) on non-enhanced CT before MT, intravenous thrombolysis, laboratory examinations, and procedure details. History of hypertension, diabetes mellitus, and heart disease was determined by self-report, current medication, or prior medical records. The definition of a smoker was smoking at least one cigarette per day for 1 year or more, and the definition of alcoholism was consuming 80 g of liquor per day. The etiology of stroke was assessed using the Trial of OGR 10172 in Acute Stroke Treatment (TOAST) classification (15). Collateral circulation was evaluated based on DSA at the beginning of the MT, categorized as follows: Grade 0: no significant or little filling of the occluded region or less than one-third of the occluded territory; grade 1: collateralization fills less than two-thirds of the occluded territory; and grade 2: collaterals reach more than two-thirds of the occluded territory or the proximal main stem (16). The HT was indicated as the presence of postoperative hemorrhage in the cerebral infarction area. The MBE was defined as follows: (1) More than 50% of middle cerebral artery supply territory had parenchymal hypodensity, accompanied by local brain edema, like lateral ventricle compression and sulcal effacement; and (2) midline shift of >5 mm at septum pellucidum or pineal gland with basal cistern occlusion (17). Two neuroradiologists independently evaluated the perioperative imaging characteristics blind to clinical data, including baseline ASPECTS, hyperdense middle cerebral artery sign (HMCAS), occluded vessel region, collateral score, and postoperative HT and MBE. Any discrepancies were figured out through discussion. The laboratory test data included peripheral blood cell counts, high-sensitivity C-reactive protein (hsCRP), coagulation function, serum electrolyte, liver and kidney function, blood lipid, cholesterol, and glucose. SII was calculated by the following formula: Neutrophil × Platelet/Lymphocyte. Qualified personnel or physicians evaluated the 90-day outcome measures through in-person interviews at the clinic or by telephone. The mRS score was used to assess the 90-day functional outcome, of which mRS score 0–2 was defined as favorable function, and mRS score 3–6 as adverse function (17).

From July 2023, a therapeutic method of mild hypothermia (MH) was conducted for patients with LAO treated by MT in our institution. Patients were included if (1) baseline ASPECTS ≤ 7 with or without core infarct volume > 70 mL measured by CT perfusion; and (2) their family members endorsed MH therapy. There was no different preoperative evaluation and surgical procedure strategy for these patients. After the MT procedure, patients were cooled using two cool blankets (Richeng, Changchun, Jilin, China) with a sandwiched method to reach the targeted temperature of 35°C at the posterior pharyngeal wall. All patients were endotracheally intubated with assisted ventilation. Pharmacological sedation with midazolam and analgesia with sufentanil were employed to prevent shivering during hypothermia. Arterial blood gas analysis was performed to monitor PH value, and arterial partial pressure of oxygen and carbon dioxide every 3–6 h. Hematological examinations for peripheral blood cells, hepatorenal function, electrolyte, coagulation function, and myocardial enzyme were scrutinized every 12–24 h. When the target temperature was achieved, the blanket temperature was adjusted to maintain the core temperature at 35.0°C. Slowly rewarming to 36.5°C of core temperature with a controlling rate of no faster than 0.5°C/h was initiated at the final phase of MH therapy by turning off the cooling blankets (18).

The patients (without mild hypothermia) were classified according to quartiles of SII levels (Q1–Q4). The categorical variables were presented as numbers (percentage) and were compared using chi-square test or Fisher’s exact test. The continuous variables with normal distribution were expressed as mean ± standard deviation (SD) and compared using analysis of variance (ANOVA) or Student’s t-test, otherwise, presented as median with interquartile range, and compared using Kruskal–Wallis test or Mann–Whitney U-test. Univariate analysis was performed for outcomes (HT, MBE, adverse function, and mortality). To avoid overfitting, variables that reached variate significance in all four outcomes were selected to build the “regression model”; in addition, the variable of age was also enrolled. All multivariate analyses in this study were adjusted for this model. The relationships between SII and four outcomes were quantified using multivariate logistic regression, generating odds ratio (OR) and 95% confidence interval (95% CI) after adjusting for multiple covariates. The median SII of each quartile was also treated as a continuous variable for the linear trend test. Weighted linear regression was employed to calculate multicollinearity among variables in the multivariable model, defined as variance inflation factor (VIF) >5. Correlations among variables in the “regression model” were assessed using the Spearman correlation test. Kaplan–Meier curve was depicted to reveal the cumulative incidence of mortality in quartiles of SII using a log–rank test for comparison. The receiver operator characteristic (ROC) curves were plotted to show the diagnostic effect of SII on outcomes. Meanwhile, the AUC values of SII and other composite inflammatory indices that have been verified to correlate with stroke prognosis, including systemic inflammation response index (SIRI; Neutrophil × Monocyte/Lymphocyte), neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), neutrophil percentage-to-albumin ratio (NPAR), aggregate inflammation systemic index (AISI; Neutrophil × Monocyte × Platelet/Lymphocyte), were calculated and compared using Delong test to explore whether SII possessed better diagnostic effect. Subgroup analysis was performed stratifying sex, age (either below or above 70 years), history of hypertension, TOAST classification (large-artery atherosclerosis [LAA] or others), occluded vessel region (ICA or MCA), collateral score (grades 0 + 1, or 2), and baseline ASPECTS (either below or above 7). The median value was determined as the threshold of SII in the subgroup analysis, using the L50 as a reference. In subgroup analysis, the likelihood ratio tests were employed to probe for interactions between SII and stratified variables. The Sobel test was used for mediation analysis to explore whether HT/MBE mediated elevated SII to 90-day adverse function and mortality.

Since the baseline ASPECTS was the primary indicator for MH, patients underwent mere MT with ASPECTS ≤ 7, and those with MH were selected to form a new cohort. The multivariate logistic regression was performed to explore the role of SII and MH in predicting the four outcomes after controlling for the “regression model.” All statistical analyses were carried out using Statistical Package for the Social Sciences (SPSS) version 25.0 (IBM, Armonk, NY, USA) and R-program version 4.4.1;¹⁰⁰¹ p < 0.05 was considered statistically significant.

A total of 445 patients with LAO have been surgically treated in our stroke center, and 323 patients treated with mere MT were eligible for this study. The process of patient selection is depicted in Figure 1. The mean age of patients was 70.73 ± 11.87 years, and 46.7% (151) were women. The ASPECTS, GCS, NIHSS, and NIHSS scores at admission were 8.20 ± 1.27, 12 (IQR: 10–14), and 15 (IQR: 12–19), respectively. One hundred and eighteen (36.5%) patients presented HMCAS. The occluded vessel region was categorized as ICA (124, 38.4%) and MCA (199, 61.6%). The number of patients with collateral scores of grades 0, 1, and 2 was 72 (22.3%), 133 (41.2%), and 118 (36.5%), respectively. A total of 103 (31.9%) patients experienced HT for the early complications and 82 (25.4%) patients suffered from MBE. At 90-day follow-up, 192 (59.4%) patients had adverse function (mRS score: 3–6) and 90 (27.9%) died.

The baseline characteristics of patients classified by quartiles of SII are presented in Table 1. There were 80, 81, 81, and 81 patients in Q1–Q4 Compared to patients with lower SII values, those with higher SII values were more likely to have lower baseline ASEPECTS, longer time from onset to recanalization, and higher serum glucose at admission, while less likely to experience intravenous thrombolysis, and successful revascularization. Critically, these patients possessed a higher proportion of HT, MBE, adverse function, and mortality (all p < 0.05).

The Kaplan–Meier curve demonstrated that individuals with elevated SII exhibited an increased mortality risk at 90-day follow-up (Figure 2). Univariate analysis for HT, MBE, adverse function, and mortality indicated that sex, GCS score at admission, initial NIHSS score, baseline ASPECTS, present HMCAS, occluded vessel region, collateral score, and successful revascularization were the standard significant variables that selected to build up the “regression model” (Supplementary Tables S1–S4), and age was also enrolled. The detailed relationships between SII and outcomes were summarized in Table 2. Multivariate analyses, controlled for sex, age, GCS score at admission, initial NIHSS score, baseline ASPECTS, present HMCAS, occluded vessel region, collateral score, and successful revascularization, demonstrated that patients in the highest quartile had significantly increased risk of HT (OR: 3.691; 95% CI: 1.634–8.340; p = 0.002), MBE (OR: 6.755; 95% CI: 2.288–19.940; p = 0.001), 90-day adverse function (OR: 4.969; 95% CI: 2.326–10.614; p < 0.001), and 90-day mortality (OR: 3.021; 95% CI: 1.159–7.876; p = 0.024), compared to those in the lowest quartile, after categorizing the SII as an ordinary variable. When presented as a continuous variable, the SII was also found as a significant predictor of HT (OR: 1.061; 95% CI: 1.035–1.086; p < 0.001), MBE (OR: 1.074; 95% CI: 1.045–1.103; p < 0.001), 90-day adverse function (OR: 1.061; 95% CI: 1.031–1.092; p < 0.001), and 90-day mortality (OR: 1.044; 95% CI: 1.018–1.070; p = 0.001). No obvious multicollinearity among variables was detected (VIF from 1.033 to 2.866).

The correlations among variables in the multivariate model are depicted in Figure 3. The SII was significantly correlated with the initial NIHSS score (r_s = 0.167, p = 0.003), baseline ASPECTS (r_s = −0.170, p = 0.002), collateral score (r_s = −0.197, p < 0.001), and mTICI score (r_s = −0.201, p < 0.001).

The ROC curve suggested the moderate diagnostic performance of SII for HT, MBE, 90-day adverse function, and mortality (AUC 0.6294–0.7512, Figure 4). Importantly, SII showed a better diagnostic effect for these four outcomes than the other inflammatory indices, including SIRI, NLR, LMR, NPAR, and AISI (Table 3).

The subgroup analysis was performed to meticulously evaluate the prognostic utility of SII for predicting the above-mentioned four outcomes, stratifying age, sex, history of hypertension, TOAST classification, occluded vessel region, collateral score, and baseline ASPECTS. No interaction was detected between SII and these variables (all p > 0.05). The SII was identified as a substantial predictor of HT, MBE, and 90-day adverse functional outcomes across majority of the patient subgroups. However, no significant correlation was observed between SII and 90-day mortality in all subgroups (all p > 0.05, Figure 5).

The mediation analysis was conducted to explore whether the relationships between elevated SII and increased risk of 90-day adverse function/mortality were mediated by HT or MBE. Using HT or MBE as a mediator respectively, we revealed the mediating effect of HT and MBE in the effect of the SII on 90-day adverse function and mortality (Figure 6).

A total of 42 patients were treated with MT + MH, and no severe cardiac arrhythmia, coagulopathy, and hypotension were encountered in these patients. The mean time of inducing hypothermia was 3.4 ± 1.8 h, the duration of maintenance was 53.7 ± 18.1 h, and the period of rewarming was 7.2 ± 2.6 h. Seventy-eight patients who underwent mere MT with ASPECTS ≤ 7 were selected with these 42 patients to form a new cohort. Compared to remaining patients with baseline ASPECTS 8–10, those in the new cohort possessed a significantly high value of SII (759 [IQR: 497–1,467] vs. 1,137 [IQR: 714–2,060], p < 0.001). The baseline characteristics of patients treated with or without MH were presented in Supplementary Table S5. A similar value of SII was found in these two groups (MT vs. MT + MH = 1,283 [IQR 706–2,223] vs. 1,066 [IQR 694–1,802], p = 0.314). Compared to patients with mere MT, those treated with MT + MH experienced a significantly lower incidence of postoperative HT (48.7% vs. 28.6%, p = 0.033), MBE (53.8% vs. 23.8%, 0.002), and 90-day mortality (42.3% vs. 19.0%, p = 0.01). In comparison, the rate of 90-day adverse function was similar (82.1% vs. 85.7%, p = 0.608). As shown in Table 4, after adjusting for sex, age, GCS score at admission, initial NIHSS score, baseline ASPECTS, present HMCAS, occluded vessel region, collateral score, and successful revascularization, the risk effect of SII and protective effect of MH were observed in HT (SII—OR: 1.037, 95% CI: 1.001–1.074, p = 0.047; MH—OR: 0.361, 95% CI: 0.136–0.957, p = 0.04), MBE (SII—OR: 1.063, 95% CI: 1.019–1.109, p = 0.005; MH—OR: 0.231, 95% CI: 0.081–0.653, p = 0.006), and 90-day mortality (SII—OR: 1.048, 95% CI: 1.011–1.087, p = 0.011; MH—OR: 0.343, 95% CI: 0.118–0.994, p = 0.049). However, the effects were not found in 90-day adverse function (SII—OR: 1.048, 95% CI: 0.983–1.118, p = 0.153; MH—OR” 1.214, 95% CI: 0.319–4.622, p = 0.776). In addition, other common complications are listed in Table 5. Patients treated with MH + MT developed a significantly higher rate of pneumonia than those treated with only MT (MH + MT vs. MT = 54.8% vs. 30.8%, p = 0.010).