Integrated analysis of the association between methionine cycle and risk of moyamoya disease

Abstract Objective The role of methionine (Met) cycle in the pathogenesis and progression of cardiovascular and cerebrovascular diseases has been established, but its association with moyamoya disease (MMD) has rarely been studied. This study aimed to analyze the levels of Met cycle‐related metabolites and constructed a risk model to explore its association with the risk of MMD. Methods In this prospective study, a total of 302 adult MMD patients and 88 age‐matched healthy individuals were consecutively recruited. The serum levels of Met cycle‐related metabolites were quantified by liquid chromatography‐mass spectrometry (LC–MS). Participants were randomly divided into training set and testing set at a ratio of 1:1. The training set was used to construct the risk score model by LASSO regression. The association between Met cycle‐related risk score and the risk of MMD was analyzed using logistic regression and assessed by ROC curves. The testing set was used for validation. Results The levels of methionine sulfoxide and homocysteine were significantly increased, while the levels of betaine and choline were significantly decreased in MMD and its subtypes compared to healthy controls (p < 0.05 for all). The training set was used to construct the risk model and the risk score of each participant has been calculated. After adjusting for potential confounders, the risk score was independently associated with the risk of MMD and its subtypes (p < 0.05 for all). We then divided the participants into low‐risk and high‐risk groups, the high‐risk score was significantly associated with the risk of MMD and its subtypes (p < 0.05 for all). The risk scores were further assessed as tertiles, the highest tertile was significantly associated with a higher risk of MMD and its subtypes compared to the lowest (p < 0.05 for all). The results were validated in the testing set. Conclusion This study has constructed and validated a risk model based on Met cycle‐related metabolites, which was independently associated with the risk of MMD and its subtypes. The findings provided a new perspective on the risk evaluation and prevention of MMD.


| INTRODUC TI ON
Moyamoya disease (MMD) was a rare cerebrovascular disorder characterized by progressive stenosis or occlusion of the internal carotid arteries, leading to the formation of abnormal collateral vessels at the base of the brain. 1 Patients with MMD commonly presented with a wide range of ischemic or hemorrhagic symptoms, including headache, seizure, cognitive decline, transient ischemic attack (TIA), infarction, and intracranial hemorrhage. 2 Epidemiology showed that MMD was more prevalent in East Asian countries and exhibited a familial aggregation pattern, suggesting the genetic predisposition. 3,4Although various risk factors have been identified, including RNF213 variants, immune response, and environmental factors, the underlying pathophysiology of MMD remained poorly understood. 5,6Current therapeutic approaches for MMD focus on revascularization procedures using extracranial vessels to restore the blood supply of the affected hemispheres. 7However, the outcomes of surgical interventions and the long-term prognosis of patients exhibit a considerable inter-individual variability.Hence, there has been a critical need to identify the potential factors associated with the risk of MMD and develop targeted prevention strategies for at-risk populations.Further studies are needed to improve the understanding of this complex disease and advance the development of more effective treatments.
Metabolomics provided a comprehensive understanding of the metabolic state of organisms. 8Analyses of the pattern and level of metabolites contributed to identifying the disease-related changes, which provided a unique insight into the underlying biological processes. 9,10In recent years, metabolomics has become an increasingly powerful tool in the identification and validation of biomarkers for various diseases.2][13] The altered levels of Met cycle-related metabolites played a crucial role in the pathogenesis and progression of cardiovascular and cerebrovascular diseases. 14,15The correlation between homocysteine (Hcy) and MMD has been discussed in previous studies. 16,17Furthermore, the association between decreased levels of choline and betaine and the risk of MMD has been recently identified. 18

| Participants and study design
In this prospective study, we consecutively recruited patients diagnosed with MMD using digital subtraction angiography (DSA) according to the 2012 Japanese diagnostic criteria from September 2020 to December 2021 at Beijing Tiantan Hospital. 19Minors and and its subtypes.The findings provided a new perspective on the risk evaluation and prevention of MMD.

K E Y W O R D S
biomarker, metabolite, methionine cycle, Moyamoya disease, risk model unilateral MMD patients were excluded from analyses.We also included a control group of 89 age-matched individuals without any underlying diseases or cerebrovascular diseases.After obtaining written informed consent, we collected the peripheral blood samples and information on demographics, clinical characteristics, and laboratory examination of the participants.Blood samples were tested using liquid chromatography-mass spectrometry (LC-MS) to quantify levels of methionine cycle-related metabolites (Figure 1).Participants without complete LC-MS data were excluded from the study.We then randomly divided all participants into training set and testing set at a ratio of 1:1.The participant recruitment process has been illustrated (Figure 2).The Ethics Committee of Beijing Tiantan Hospital has reviewed and approved this study (KY2022-051-02).

| Data collection
We collected the baseline patient characteristics, including age, gender, past medical history, heart rate, systolic blood pressure (SBP), diastolic blood pressure (DBP), body mass index (BMI), RNF213 variants, and laboratory tests.Fasting peripheral venous blood samples were obtained with vacuum biochemical tubes on the second morning after admission.All samples were centrifuged, aliquoted, and stored at −80°C within 1 h after collection.In order to avoid bias in the test results, all information of participants was blinded to the laboratory technicians.RNF213 p.R4810K variants were confirmed with the following primer sequences: Forward 5′-GCCCT CCA TTT CTA GCACAC-3′ and Reverse 5′-AGCTG TGG CGA AAG CTTCTA-3′.Neurological function evaluation at admission was assessed using the modified Rankin Scale (mRS) scores, which were divided into two levels (0-2 and 3-5).The severity of MMD patients was determined based on the Suzuki stage of the more severe side.

| Construction of Met cycle-related risk model
The levels of Met cycle-related metabolites in the training set were incorporated into LASSO regression to construct the risk model.The lambda that minimized the deviations has been selected.Based on the levels of Met cycle-related metabolites, a risk score was calculated for each participant using the following formula (n, metabolite quantity; level i , metabolite level; coefficient i , metabolite regression coefficient)

| Statistical analysis
The testing set was used for validation.Receiver-operating characteristic (ROC) curves and the area under ROC curve (AUC) were used to assess the predictive abilities of models for the risk of MMD and its subtypes.A two-sided p value less than 0.05 was considered significant.

| Study participants and baseline information
A total of 390 individuals were eventually enrolled in this study, consisting of 88 HCs and 302 MMD patients.Baseline characteristics between the two groups were compared, revealing that MMD patients had a higher prevalence of hypertension, diabetes, hyperlipidemia, smoking, and drinking, as well as higher levels of SBP, DBP, and BMI than HCs (p < 0.05 for all, Table 1).Additionally, it showed significantly higher levels of WBC count, NEUT count, and TG and lower levels of TC, HDL-C, LDL-C, and ApoA in the MMD group compared to the HC group.Furthermore, levels of Met cycle-related metabolites were also analyzed, and found that the levels of methionine sulfoxide (MetO) and Hcy were significantly increased in the MMD group, while betaine and choline levels were significantly decreased (p < 0.05 for all).
Baseline characteristics between HCs and MMD subtypes (ischemic-type and hemorrhagic-type) were also compared.Both subtypes showed higher rates of hypertension, hyperlipidemia, smoking, and drinking, as well as higher levels of SBP compared to HCs (p < 0.05 for all, Table 2).In the ischemic MMD group, there were higher levels of WBC count, LY count, NEUT count, GLU and TG, and lower levels of TC, HDL-C, LDL-C, and ApoA compared to HCs (p < 0.05).In the hemorrhagic MMD group, there were higher levels of NEUT count, TG, and ApoB, and lower levels of HDL-C and ApoA compared to HCs (p < 0.05 for all).Consistent with MMD overall, both ischemic and hemorrhagic subtypes showed significantly increased MetO and Hcy levels, and decreased betaine and choline levels (p < 0.05 for all).

| Construction of Met cycle-related risk model
All participants were randomly divided into training and testing sets, with equal numbers of 44 healthy individuals and 151 MMD patients in each set.The data of Met cycle-related metabolites in training set was used to construct the risk score system by LASSO regression (Figure 3).Four metabolites were identified as the most significant contributors to the risk score, including MetO, Hcy, betaine, and choline (Table 3).The Met cycle-related risk score was then calculated for all participants, and it was found to be significantly higher in MMD group and its subtype groups compared to the HC group (p < 0.001 for all, Figure 4).
Based on the median risk score value, the MMD patients in training set were divided into low-risk and high-risk groups (Table 4).We compared the clinical characteristics of the two groups.It showed that there was a significantly higher proportion of hyperlipidemia in the high-risk group (p = 0.013).The levels of WBC count (p = 0.026) and MONO count (p = 0.033) were also significantly higher in the high-risk group than in the low-risk group.Further division of MMD patients into low-risk, intermediate-risk, and high-risk groups based on the tertiles of the risk score showed a significantly positive correlation between the risk score and hyperlipidemia (r = 0.204, p = 0.012, Table 5), levels of WBC count (r = 0.172, p = 0.035), and MONO count (r = 0.176, p = 0.031).Conversely, the level of HDL-C was significantly negatively correlated with the risk score (r = −0.184,p = 0.024).

| Role of the risk score in MMD and its subtypes
In the training set, logistic regression analyses showed a significant positive association between the Met cycle-related risk score and the risk of MMD in Crude model (OR = 5.904, 95%CI = 3.153-11.057,p < 0.001, Table 6).After adjusting potential confounders, it showed that the risk of MMD enhanced with the increase in the risk score  (AUC = 0.840) and Model 1 (AUC = 0.879).In addition, the risk score was also found to be independently associated with ischemic and hemorrhagic subtypes in all three models (Table 6).The ROC curves confirmed the predictive ability of these models (Figure 5B,C).
We then compared the risk of MMD between individuals with low and high-risk scores and found that the high risk score was We calculated the Met cycle-related risk score of the individuals in the testing set and verified the association between the risk score and MMD.Subsequently, the participants in the testing set were then stratified into different risk levels based on the median value and tertiles of the risk score.The results confirmed that the risk score was independently associated with MMD and its subtypes (Table 7).The AUC values of the ROC curves confirmed the predictive accuracy of the models (Figure 6A-I).

| DISCUSS ION
In this prospective study, we first compared the difference in clinical characteristics between MMD patients and HCs.We further  After adjusting for other risk factors, it showed an improvement in risk prediction.In this study, we provided a novel perspective on the role of metabolism dysregulation in the pathogenesis of MMD and suggested potential therapeutic targets.
One of the Met cycle-related metabolites, MetO, was produced from Met by reactive oxygen species (ROS) and had the potential to serve as a biomarker of oxidative stress. 21MetO could be specifically reversed to Met by MetO reductase system. 22MetO has been shown to influence protein folding and stability, as well as protein-protein interactions, thereby impacting cellular functions such as oxidative stress, inflammation, and apoptosis.4][25] Previous studies have discussed the relationship between MetO and cerebrovascular disease.A recent large-scale prospective metabolomics study revealed that MetO was significantly associated with incident stroke, and its related metabolite score improved the prediction of the risk of incident ischemic stroke. 26Another study confirmed the protective role of MetO reductase system to neurovascular inflammation in ischemic stroke, which attenuated ROS-augmented NF-κB activation in endothelial cells and protected against the oxidation of Met residues in the regulatory domain of calcium/calmodulindependent protein kinase II (CaMKII). 27Furthermore, the deficiency of MetO reductase A (MsrA) has been shown to promote vascular smooth muscle cell (VSMC) proliferation and neointimal hyperplasia through the extracellular signal-regulated kinase 1/2 signaling pathway, 28 which were also pathological changes associated with MMD.
It suggested the role of MetO in the risk of MMD and MetO reductase system might be the potential therapeutic targets. 29y, a sulfur-containing amino acid, was a key determinant of Met cycle.Folate and vitamin B12 were involved in the remethylation of Hcy back into Met. 30Deficiency of these cofactors and mutation in methylenetetrahydrofolate reductase (MTHFR) and methionine synthase reductase (MTRR) could lead to excessive accumulation of plasma Hcy called hyperhomocysteinemia (HHcy). 31Previous studies found that Hcy was associated with cardiovascular diseases, ischemic stroke, neurological disorders, cognitive impairment, and cancer. 32In previous genome-wide association study, researchers identified MTHFR and TCN2, two Hcy metabolism regulatory genes, as novel susceptibility genes for MMD, which were significantly associated with high-serum Hcy levels in MMD. 33Previous case-control studies identified the positive association between Hcy, HHcy, and the risk of MMD. 16,17Furthermore, HHcy has been found correlated to postoperative ischemia and poor postoperative angiogenesis in MMD patients. 34,35Growing evidence has shown that MMD was associated with oxidative stress and chronic inflammation.Hcy could induce cellular and molecular oxidative injury via ROS and increase the level of MMP-9 in the vascular wall. 36,37In addition, the blocking effect of Hcy on nitric oxide (NO) synthesis has been shown to disrupt NO-mediated endothelial-dependent vasodilation, and the mitogenic effect of Hcy could lead to the proliferation of VSMCs. 38y has also been linked to the promotion of coagulation and alterations in lipid metabolism, leading to thrombosis. 39These studies indicated the potential role of Hcy in the pathogenesis of MMD.
Choline has been an essential molecule involved in several key physiological processes, including the normal function of cell membranes, regulation of neurotransmitter function, transportation of lipids, and metabolism of one-carbon units in body. 40Recent studies have shown a relationship between DNA methylation and the pathogenesis of MMD, with the aberrant promoter hypomethylation of Sortilin 1 as a novel biomarker. 41,42We considered that the change in  epigenetic patterns and regulation of gene expression by epigenetic modifications might be associated with the low intake of choline. 43rthermore, plasma choline has been found inversely associated with cardiovascular events, recurrent stroke, and a lower risk of cognitive impairment after ischemic stroke. 44,45Betaine, also known as trimethylglycine, was obtained directly from the diet or converted from choline by choline dehydrogenase and betaine aldehyde dehydrogenase.
Betaine was an important methyl donor involved in the methylation process of Hcy into Met. 46An inadequate level of betaine could result in the accumulation of Hcy. 47Betaine played an important role in reducing oxidative stress, inhibiting NF-κB activity, and NLRP3 inflammasome activation, and attenuating endoplasmic reticulum stress and apoptosis. 48Additionally, betaine has been shown to protect against coagulation events and reverse platelet aggregation. 49Interestingly, there seemed to be a potential link between betaine and MetO.A recent study has found that betaine could increase the expression of MetO reductases b1 (Msrb1) and b2 (Msrb2) to exert a neuroprotective effect in ischemia/reperfusion injury-induced brain damage. 50 It indicated the complex interactions between the Met cycle and the risk of MMD.However, the integrated analysis of the role of Met cycle-related metabolites in MMD has hardly been studied.Therefore, we analyzed the levels of Met cycle-related metabolites in MMD and constructed a Met cycle-related risk model, which helped to demonstrate the correlation between Met cycle and the risk of MMD.Our study suggested that interventions targeting the Met cycle might have therapeutic potential in the prevention and treatment of MMD.
explored the differences in ischemic and hemorrhagic MMD groups.It showed that the levels of four Met cycle-related metabolites were significantly different in the MMD group and its subtypes from those in the HC group.The training set was used to construct the risk model by LASSO regression and analyze the role of Met cyclerelated risk score in MMD.We found that the risk score was independently associated with an increased risk of MMD and its subtypes.

F I G U R E 4
Levels of risk score between different groups.ns, not significant; ***p < 0.001.A. Risk score levels between MMD patients and HCs; B. Risk score levels between MMD subtypes and HCs.

F I G U R E 5 7
ROC curves of Met cycle-related risk score in different models for the risk of MMD and its subtypes in training set.(A-C) ROC curves of Met cycle-related risk score in different models for the risk of MMD and its subtypes.A. MMD overall; B. Ischemic-type; C. Hemorrhagic-type.(D-F) ROC curves of low and high levels of Met cycle-related risk score in different models for the risk of MMD and its subtypes.D. MMD overall; E. Ischemic-type; F. Hemorrhagic-type.(G-I) ROC curves of Met cycle-related risk score tertiles in different models for the risk of MMD and its subtypes.G. MMD overall; H. Ischemic-type; I. Hemorrhagic-type.Association between Met cycle-related risk score and risk of MMD and its subtypes in the testing set.
The etiology of MMD remains unclear, which has been considered as a combined effect of immune, genetic, environmental, metabolic, and other factors.MMD has long been considered an irreversible condition with high disability and fatality rate, causing a huge burden to families and society.Direct and indirect revascularization procedures have been widely used in the treatment of MMD.However, the bypass surgery aimed to increase blood supply to the affected hemispheres, improve ischemic symptoms, and reduce the risk of recurrent hemorrhage, rather than reversing the progression of the disease.Therefore, it has been necessary to investigate novel risk factors associated with this disease, particularly modifiable risk factors, and explore potential therapeutic targets.Our study provided a new point of view in the exploration of metabolism dysregulation in the pathogenesis of MMD.In this study, we identified four different Met cycle-related metabolites in MMD patients compared with HCs, including MetO, Hcy, choline, and F I G U R E 6 ROC curves of Met cycle-related risk score in different models for the risk of MMD and its subtypes in testing set.(A-C) ROC curves of Met cycle-related risk score in different models for the risk of MMD and its subtypes.A. MMD overall; B. Ischemic-type; C. Hemorrhagic-type.(D-F) ROC curves of low and high levels of Met cycle-related risk score in different models for the risk of MMD and its subtypes.D. MMD overall; E. Ischemic-type; F. Hemorrhagic-type.(G-I) ROC curves of Met cycle-related risk score tertiles in different models for the risk of MMD and its subtypes.G. MMD overall; H. Ischemic-type; I. Hemorrhagic-type.
3)were used for the statistical analyses in this study.Continuous variables were analyzed for normality of the distribution using the Kolmogorov-Smirnov test.Normally distributed continuous variables were compared by t-test and analysis of variance (ANOVA).
Baseline characteristics between HCs and MMD subtypes.
Construction of Met cycle-related risk model.A. LASSO deviance profiles; B. LASSO coefficient profiles.
TA B L E 3Coefficients in LASSO regression model of the Met cycle-related metabolites.
Baseline characteristics of MMD patients in low and high-risk groups.Baseline characteristics of MMD patients according to tertiles of the risk score.Association between Met cycle-related risk score and risk of MMD and its subtypes in the training set.
Note: Suzuki's stage was defined on the more severe side.*p < 0.05, significant difference.TA B L E 4 TA B L E 5 TA B L E 6