During 1 July to 31 December 2019, 138 adult participants with MMD were enrolled prospectively. Pediatric patients were not included in this study. In this study, we excluded 9 and 3 participants who did not accept surgical therapy or died due to comorbidities unrelated to MMD progression, respectively. Furthermore, three participants were lost to follow-up. Ultimately, 123 patients with MMD were enrolled in this research. Digital subtraction angiography (DSA) was used to diagnose MMD, according to the Japanese guidelines published in 2012 (17). The detailed study design and procedures have been described previously (18). The processes of this study were conducted on the basis of the guidelines of the Declaration of Helsinki. Moreover, this research was approved by the Ethics Committees of Beijing Tiantan Hospital, Capital Medical University. All the participants received and signed the informed consents.

The following clinical data were included: demographic information (age and sex), comorbidities, personal history (smoking and drinking), clinical characteristics (primary symptoms, heart rate, and blood pressure), laboratory examinations, and manifestations of the neuroimaging. Comorbidities were summarized into hypertension, diabetes mellitus, and thyroid disease. Suzuki stage and posterior cerebral artery involvement on the operative side was measured blindly through DSA by two neurosurgeons. Any divergence on the radiological manifestations was re-assessed by a third neurosurgeon. We evaluated the preoperative cerebral hemodynamic status of the participants by CT perfusion. Our previous study divided the hemodynamic status in the pre-stroke period into the following four stages (19): Stage I, time to peak (TTP) was delayed, mean transit time (MTT), regional cerebral blood flow (rCBF), and regional cerebral blood volume (rCBV) were normal; Stage II, TTP, and MTT were delayed, rCBF was normal, and rCBV was normal or slightly increased; Stage III, TTP, and MTT were delayed, cCBF was decreased, and rCBV was normal or slightly decreased; Stage IV, TTP, and MTT were delayed, rCBF and rCBV were decreased. The laboratory examinations concerning peripheral blood samples included glucose, creatinine (Cr), uric acid (UA), albumin (ALB), triglyceride (TG), total cholesterol (TC), HDL, low-density lipoprotein (LDL), apolipoprotein A (ApoA), apolipoprotein B (ApoB), and homocysteine (Hcy). Peripheral blood samples were gathered at 8:00 a.m. when patients with MMD had fasted for over 12 h. Serum Hcy levels above 15 μmol/L were defined as hyperhomocysteinemia (HHcy) (20). Serum HDL levels below 0.9 mmol/L were described as HHDL (21).

As described previously, three types of surgical treatment were conducted by neurosurgeons, namely, direct bypass, indirect bypass, and combined bypass. Direct bypass and combined bypass were performed for the most participants in our institution. When the superficial temporal artery or middle cerebral artery were too fragile to conduct artery anastomosis, we performed the indirect surgical bypass (7). As for the indication of revascularization, the clinical manifestation of patients should be considered first. The symptomatic hemisphere was given priority for the revascularization surgery. CT perfusion was taken into account for the patients who had no obvious symptom. The hemisphere with lower perfusion was treated first (22). The long-term outcome was determined through clinic visits and telephone interviews 24–35 months after discharge. Follow-up accidents concluded transient ischemia attack (TIA), ischemic stroke, hemorrhagic stroke, and loss of life. Recurrent stroke was regarded as a newly neurological deficiency that was persistent over 24 h, which was related to a new infarct or hemorrhage by using MRI or CT. The site and characteristics of the ischemic lesion were measured using the Oxfordshire Community Stroke Project (OCSP) system (23). OCSP subtypes included total anterior circulation infarcts (TACIs), partial anterior circulation infarcts (PACIs), posterior circulation infarcts (POCIs), and lacunar infarcts (LACIs). TACI refers to large anterior circulation infarcts with both the cortical and subcortical involvement, PACI refers to more restricted and predominantly cortical infarcts, POCI refers to infarcts associated with vertebrobasilar arterial territory, and LACI refers to infarcts confirmed to the territory of deep perforating arteries (24). The modified Rankin Scale (mRS) was to assess the neurological status at follow-up, with scores of 0–2 indicating a favorable neurological status and 3–6 representing an unfavorable neurological status. Assessment of the mRS score and stroke-free rate was performed by the two neurosurgical residents.

All the statistical analysis was executed by using the IBM SPSS Statistics (version 22.0; IBM Corp.). The counts (percentages) were evaluated by the Pearson Chi-square test or the Fisher exact test. The continuous variables were assessed by t-test and the Mann–Whitney U test. The Kaplan–Meier (KM) survival analysis was conducted to compare the stroke-free survival rates between participants with HHDL or diabetes and those without. The identification of independent factors of the neurological status were evaluated by univariate and multivariate logistic regression analyses. Univariate and multivariate Cox regression analyses were performed to identify the predictors of the recurrent stroke. Statistical significance was regarded as p < 0.05 in a two-tailed test. Variables with p < 0.05 in univariate analysis were included in the multivariate analysis. Multivariate Cox regression analyses were adjusted for age, gender, glucose, Suzuki stage, primary symptom, and LDL.

In total, 123 eligible patients with MMD, aged 19–61 years, were enrolled in the present study (39.02% males). The subjects were aged 39.11 ± 9.69 years (mean ± SD). The clinical and laboratory characteristics of the participants with and without stroke recurrence are shown in Table 1. Concerning the comorbidities, the incidence of diabetes was remarkably higher in the recurrent stroke group than in the non-recurrent stroke group (p = 0.022). No statistical differences in the primary symptom or personal history were presented between the two groups. Heart rate, blood pressure, BMI, surgical option, and the Suzuki stage also showed similarities between the recurrent group and the non-recurrent stroke group. Regarding laboratory examination, participants with the recurrent stroke group showed a lower level of HDL (p = 0.030). Moreover, more participants in the recurrent stroke group had HHDL (p = 0.045).

The clinical features of participants with and without HHDL are presented in Table 2. HHDL was observed in 28 participants (19 men and 9 women). The mean age of these participants was 42.18 ± 9.53 years. Compared with the normal HDL group, participants with HHDL were more likely to be men (p = 0.000). In the terms of primary symptom, 20 (71.43 %) participants were infarction and 8 (28.57 %) participants were non-infarction in HHDL group. As for comorbidities and personal history, participants with HHDL had higher incidences of diabetes (p = 0.000), smoking (p = 0.003), and drinking (p = 0.003) compared with the normal HDL group. Comparing to the normal HDL group, participants with HHDL showed a higher level of BMI (p = 0.050). In the terms of laboratory examinations, the incidences of UA (p = 0.013) and TG (p = 0.000) were significantly higher in participants with HHDL than those with normal HDL. Contrastingly, the incidences of TC (p = 0.002), LDL (p = 0.026), and ApoA (p = 0.000) were remarkably higher in the normal HDL group than in the HHDL group. Suzuki stage between the two groups also exhibited differences (p = 0.018).

The results of the long-term clinical outcomes analysis according to HHDL are shown in Table 3. After discharge, follow-up events during 28 months occurred more frequently in the HHDL group than in the normal HDL group (p = 0.025). However, no significant difference was displayed in the ischemic stroke, hemorrhagic stroke, and TIA (p > 0.05 for all). Functional outcomes (mRS score) did not show statistical differences between the two groups. What's more, there was statistical significance in the KM curve of stroke incidence between the normal HDL group and the HHDL group (log-rank test, p = 0.034) (Figure 1). As illustrated in Figure 2, the KM survival curve showed significant difference in the stroke incidence between the participants with diabetes and those without.

Predictive factors of the neurological status were also evaluated in this study. As demonstrated in Table 4, univariate logistic regression indicated that diabetes (OR 4.444, 95% CI 1.158–17.063; p = 0.030) and Suzuki stage (OR 1.853, 95% CI 1.134–3.028; p = 0.014) were related to the neurological status. It also could be found from the univariate logistic regression that HHDL (OR 0.916, 95% CI 0.237–3.543; p = 0.899) and HDL (OR 0.729, 95% CI 0.094–5.648; p = 0.763) were not related to the neurological status. Multivariate logistic regression analysis showed that the Suzuki stage (OR 1.779, 95% CI 1.085–2.917; p = 0.022) was significantly related to the neurological status. Diabetes (OR 4.036, 95% CI 0.974–16.732; p = 0.054) was not a predictive factor of the neurological status in the multivariate regression analysis.

As shown in Table 5, we identified diabetes (HR 3.078, 95% CI 1.116–8.492; p = 0.030), HDL (HR 0.121, 95% CI 0.018–0.815; p = 0.030), HHDL (HR 2.526, 95% CI 1.031–6.187; p = 0.043) were predictive for the recurrent stroke in the univariate Cox regression analysis. In the multivariate analysis, diabetes (HR 4.195, 95% CI 1.041–16.899; p = 0.044), HDL (HR 0.061, 95% CI 0.006–0.626; p = 0.019), HHDL (HR 3.341, 95% CI 1.110–10.051; p = 0.032) were independent risk factors for the recurrent stroke. Multivariate analyses were adjusted for age, gender, glucose, Suzuki stage, primary symptom, and LDL.

We utilized the OCSP classification to classify 17 participants who developed the recurrent ischemic stroke. As illustrated in Table 6, 6 cases in the HHDL group and 4 cases in the diabetes group were classified as LACI among the 17 participants with the recurrent ischemic stroke. Only one participant in each of the two groups was classified as PACI.