Serum Metabolomic Patterns in Young Patients with Ischemic Stroke: A Case Control Study

Background: Ischemic stroke is one of the leading causes of death and adult disability. The incidence of ischemic stroke continues to rise in young adults. This study aimed to provide a comprehensive evaluation of metabolic changes and explore possible mechanisms in young ischemic stroke patients without common risk factors. Methods: This study investigated serum metabolomics in 50 young patients with newly suffered ischemic stroke and 50 age-, sex-, and body mass index–matched healthy controls. The metabolomic data were analyzed by performing a multivariate statistical analysis. Results: The 197 metabolites, including amino acids, bile acids, free fatty acids, and lipids, were identied in all participants. Multivariate models showed signicant differences in serum metabolomic patterns between young patients with ischemic stroke and healthy controls. The stroke patients had increased L-methionine, homocysteine, glutamine, uric acid, GCDCA, and PE (18:0/20:4, 16:0/22:5), and decreased levels of L-citrulline, taurine, PC (16:2/22:6, 16:2/20:5, 15:0/18:2), and SM (d18:1/23:0, d20:0/19:1, d18:1/22:0, d16:0/26:1, d16:0/18:0, d16:0/22:1, d18:1/19:1, d16:0/17:1, d16:1/24:1, d18:1/19:0). Based on the identied metabolites, the metabolic pathways of arginine biosynthesis, glycerophospholipid metabolism, and taurine and hypotaurine metabolism were signicantly enriched in the young patients with ischemic stroke. Conclusions: Serum metabolomic patterns were signicantly different between young patients with ischemic stroke and healthy controls.

Conclusions: Serum metabolomic patterns were signi cantly different between young patients with ischemic stroke and healthy controls.

Background
Ischemic stroke is one of the leading causes of death and adult disability [1]. In recent years, the incidence of ischemic stroke has declined gradually in the general population but has continued to rise in young adults [2]. Most patients cannot completely recover even after proper treatment, so ischemic stroke has major social and economic impacts for working-age adults. Management of stroke risk factors is considered as the best strategy to decrease the incidence of ischemic stroke [2]. Notably, the etiology of ischemic stroke differs signi cantly between young and older patients [2][3][4]. Several previous studies have indicated that smoking, hypertension, and diabetes are common risk factors in young patients with ischemic stroke [5][6]. Some cardiac and vascular diseases, such as arrhythmia, cerebral artery dissection, and small vessel disease, were also common causes in young ischemic stroke patients [2][3][4].
Although several risk factors have been discovered, over a third of ischemic strokes in young adults remain cryptogenic, which hints at an inadequate understanding of the pathogenesis of ischemic stroke in young adults [2][3][4]. Therefore, identifying novel risk factors and understanding the pathogenesis of ischemic stroke in young adults is urgent and important.
Metabolomics is a novel analytical approach dedicated to speci cally identifying small-molecule metabolites, which represents the endpoint of the omics cascade and provides an explanation of the pathophysiology and metabolic changes of some diseases. Several studies have identi ed altered metabolomics in patients with ischemic stroke; however, few studies have focused on young patients [7][8][9][10]. This study aims to provide a comprehensive evaluation of metabolic changes and explore possible mechanisms in young ischemic stroke patients without common risk factors.

Study design and participants
A total of 50 patients between ages 18 and 50 who newly suffered ischemic stroke were consecutively recruited between September 2016 and October 2017 at the Beijing Chaoyang Hospital A liated with Capital Medical University [11]. Meanwhile, 50 age-, sex-, and body weight index (BMI)-matched healthy controls were enrolled from the Physical Examination Center at the same hospital. Ischemic stroke was diagnosed by clinical symptoms and imaging examination, including a cranial computed tomography scan and/or magnetic resonance imaging within 24 hours of hospital admission, according to the International Classi cation of Diseases, 10th revision. All healthy controls had a normal cranial imaging examination and no history of cerebrovascular diseases according to the medical history collection and physical examination. All participants in the control and stroke groups had no history of diabetes, hypertension, hypercholesterolemia, past or present cigarette smoking, pregnancy and puerperium, use of vasoactive or conceptive or illicit drugs, excessive drinking, heart or vascular disease, atrial brillation or utter, cervical arterial dissection, hematologic disease, antiphospholipid antibody syndrome, infectious disease, cardiac or liver or renal function impairment, thyroid dysfunction, systemic in ammatory disease, or cancer. All participants in the control and stroke groups had no family history of coronary heart disease or stroke, which was de ned as coronary artery disease, sudden death, or stroke in a rstdegree male relative younger than 55 years old or a female relative younger than 65 years old. All studies were conducted in accordance to the principles of the Declaration of Helsinki. (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG), were measured enzymatically as described previously (12). The levels of total bile acid (TBA) were measured using an enzymatic cycling method (Hitachi 747, Roche Diagnostics, Germany). Uric acid (UA) was measured using the enzymatic uricase method (Hitachi 747, Roche Diagnostics, Germany). HbA1c was measured using the HPLC method. Homocysteine was measured using the cycling enzymatic method (Hitachi 747, Roche Diagnostics, Germany). BMI was calculated as weight in kilograms divided by height in meters squared.
All stroke patients were examined to exclude common risk factors for ischemic stroke via electrocardiograph, transesophageal and transthoracic echocardiogram, brain magnetic resonance angiography, carotid and vertebral computed tomography angiography, immunological antibodies (such as antinuclear antibodies, lupus anticoagulants, anticardiolipin antibodies, etc.), and serum functional levels of antithrombin, protein C, and protein S, etc. Individuals were considered to have dyslipidemia if they had a total cholesterol level over 200 mg/dL, or if they were being already treated. Individuals were considered to have diabetes if they had HbA1c levels ≥ 6.5%, or if they were already diagnosed as diabetic. Individuals were considered to have hypertension if they had ever had diastolic blood pressure ≥ 90 mmHg and/or systolic blood pressure ≥ 140 mmHg and/or used antihypertensive medication.

Metabolomics
The detailed method for metabolomic analyses has been described previously [12]. In brief, we used liquid chromatography of Waters I-Class (Waters, Milford, MA, USA) coupled with a Waters Xevo TQ-S (Waters, Milford, MA, USA) mass spectrometer with an electrospray ionization (ESI) source to analyze each amino acid or bile acid extract. Each free fatty acid or lipid extract was analyzed by liquid chromatography using an LC-20AXR Rapid Separation LC system (Shimadzu, Kyoto, Japan) coupled with a Qtrap5500 (AB SCIEX, Redwood City, CA, USA) mass spectrometer with an ESI source. MassLynx software 4.1 (Waters, USA) and Analysis software (SCIEX, USA) were used for systemic control and data acquisition, respectively. The samples from the control and stroke groups were alternately injected into the analytic work ow at a random order. The quality control samples, which were mixed with equal aliquots of serums from the control and stroke groups, were injected into every 15 samples throughout the analytical work ow. Skyline software (MacCoss, University of Washington) was used to analyze ultrahigh-performance liquid chromatography-mass spectrometry raw data and obtain the quantitative concentration of each metabolite in the samples. The features were selected based on their coe cients of variation (CVs) with quality control (QC) samples. The features with CVs over 15% were excluded. The stability and reliability of metabolomic data were evaluated prior to data analysis, and the results showed that the method was excellent (Supplementary Figures S1-S4).

Statistical analysis
The clinical parameters of the control and stroke groups are expressed by means ± standard deviations for normally distributed data. TBA variables had a skewed distribution, and are shown as median and upper and lower quartiles. The skewed-distribution variables were log-transformed before analysis. The differences between the control and stroke groups were analyzed using an independent sample t-test or a Mann-Whitney U test. The proportions were analyzed using chi-squared tests. All statistical analyses were performed using SPSS 21.0 (Chicago, IL, USA). All tests were two-tailed, and the results were considered statistically signi cant when P < 0.05.
Metabolomic data were analyzed using SIMCA 14.0 (Umetrics AB, Umeå, Sweden) and MetaboAnalyst 4.0 (www.metaboanalyst.ca) [13]. MetaboAnalyst 4.0 was used to normalize data, reduce systematic bias, and improve consistency. The metabolites with features > 50% missing values were removed. The remaining missing values were replaced by the half of the minimum positive value in the original data. Both a principal component analysis (PCA) and an orthogonal partial least-squares discriminant analysis (OPLS-DA) were performed to reveal the global metabolic changes between the control and stroke groups, using SIMCA 14.0. A validation plot was used to assess the validity of the OPLS-DA model using sevenfold cross-validation and permutation tests (n = 200). The variable in uences on projection (VIP) values were calculated using the OPLS-DA model. The serum differential variables with VIP values > 1.5 from the OPLS-DA model were assessed using Student's t-test or the Wilcoxon (Mann-Whitney U) test to analyze signi cance. Signi cant differences in the pathway were evaluated using the hypergeometric test, and pathway topology was analyzed based on the relative betweenness centrality.

Results
Baseline characteristics of the control and stroke groups Table 1 presents the baseline characteristics of the control and stroke groups. The two groups were well matched for age, gender, and BMI. The stroke group had higher TG, UA, and homocysteine and lower HDL-C levels (all P < 0.01). There was no signi cant difference in TC, LDL-C, TBA, or HbA1c between the control and stroke groups. Metabolomic analysis of the control and stroke groups The 197 metabolites were identi ed in all participants (Table S1). PCA score plots showed a clear clustering of the control and stroke groups, and the cumulative tness (R2 value) of the PCA model was 0.767 ( Figure. 1). The OPLS-DA analysis indicated clear separations between the control (green dots) and stroke (blue dots) groups (R2Y = 0.928, Q2 = 0.814, Fig. 2A). The results of the permutation test strongly indicate that the original model was valid (R2 intercept = 0.252, Q2 intercept = -0.394, Fig. 2B).
Based on the selection criteria (VIP > 1.5 and P < 0.05), 20 metabolites were obtained (

Pathway analysis
We further performed a pathway analysis to identify the signi cantly changed metabolic pathway, according to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Based on the identi ed metabolites, the metabolic pathways of arginine biosynthesis, glycerophospholipid metabolism, and taurine and hypotaurine metabolism were signi cantly enriched in young patients with ischemic stroke (Fig. 4 and Table 3). Hit means the matched number in pathway; the P value is calculated from the enrichment analysis; Impact value is calculated from pathway topography analysis; FDR value is the false discovery rate adjusted P value.

Discussion
The altered metabolomics in ischemic stroke have been identi ed in older patients by several studies; however, until now only a few studies focused on young patients (7-10). The present study showed that serum metabolomic patterns were signi cantly different between young patients with ischemic stroke and healthy controls. The young ischemic stroke patients had increased L-methionine, homocysteine, glutamine, uric acid, GCDCA and PE ( Amino acids are an important group of metabolites that participate in multiple physiological and pathophysiological processes. Consistent with previous studies in older people, the present study showed that young patients with ischemic stroke had signi cantly increased L-methionine, homocysteine, uric acid, and glutamine levels [14][15][16][17]. As is known, both hyperhomocysteinemia and hyperuricemia are independent risk factors for stroke [14][15]. As an essential amino acid, methionine comes from dietary intake. Homocysteine is an intermediate in methionine metabolism, and a moderate methionine diet of four weeks can induce hyperhomocysteinemia [18]. The present study found that young patients with ischemic stroke had signi cantly increased L-methionine. The elevated methionine and homocysteine levels might be associated with increased methionine intake. L-glutamine has been considered a bene cial amino acid that has antioxidant and anti-in ammatory effects [19][20]. L-glutamine supplementation reduced infarct volume and promoted neurobehavioral recovery in stroke mice [20]. And brain injury increased glutamine output in the glutamate-glutamine cycle, and further protected neurons from damage [21]. Therefore, increased L-glutamine might be a compensatory reaction to brain injury. Unlike results in older people, our study found that young patients with ischemic stroke had decreased levels of L-citrulline and taurine. Nitric oxide (NO) is a gas-signal molecule with various physiological functions, including regulating the balance of blood ow and oxygen demand and neurovascular coupling in the brain [22][23]. Endogenous NO was mainly generated from the citrulline-arginine-NO pathway [24]. Previous studies have shown that L-citrulline supplementation increased the bioavailability of L-arginine and promoted NO synthesis [24]. Taurine is a semiessential amino acid in mammals and has been proven to have multiple bene cial effects, including attenuating in ammation-and endoplasmic reticulum stress-induced organ injuries [25][26]. Taurine treatment inhibited ethanolmediated cell apoptosis in the cerebellum [27]. Therefore, decreased L-citrulline and taurine levels might be related to the pathogenesis of young patients with ischemic stroke.
As the key components of bile, bile acids are essential for regulating the digestion and absorption of dietary fat through the intestine. Recently, increasing evidence has shown that, beyond the gastrointestinal tract, circulating bile acids in the bloodstream also act as important signaling molecules for many pathophysiological processes [28][29]. The present study showed that the serum TBA levels were similar between the stroke and control groups, while the component of bile acid was signi cantly different. The present study showed that young patients with ischemic stroke had signi cantly increased GCDCA levels. GCDCA is a glycine-conjugated bile acid and has been demonstrated to be one of the most abundant bile acids in human serum [30][31]. GCDCA causes increased oxidative stress and promotes apoptosis by inducing JNK activation in rat hepatocytes [32]. Previous studies have shown that increased GCDCA levels are associated with liver injury induced by alcoholism or cholestasis [30][31]. Therefore, increased GCDCA levels might be related to the pathogenesis of young patients with ischemic stroke.
However, most previous studies were performed in older people, and their results also were controversial. The present study showed that lipid-related metabolites are signi cantly changed in young patients with ischemic stroke. PC, SM, and PE are all major constituents of cell membranes and play an important role in membrane-mediated cell signaling [35]. Previous studies have found that PC has many bene cial effects, including attenuating liver steatosis, slowing down aging-related processes, and improving brain function [36][37]. Consistent with previous research in older people, serum PCs were decreased in patients with ischemic stroke [33][34]. Moreover, the present study also found that young patients with ischemic stroke had increased PE ( [38][39]. Until now, the function of SM and PE remained unclear. Consistent with the present study, a recent study performed in three independent, follow-up, population-based cohorts also found a possible protective role for SM in stroke development [40]. The serum SM (32:1) level was demonstrated to inversely relate to the onset of ischemic stroke [40]. Further studies are needed to investigate whether changed levels of PC and PE are involved in the pathogenesis of young patients with ischemic stroke.
Besides the changed metabolites pattern, young patients with ischemic stroke had higher TG and FBG and lower HDL-C levels, when compared with age-, gender-and BMI-matched healthy controls. Consistent with the previous studies, relatively higher FBG within the normal range and increased TG and decreased HDL-C levels were observed in young patients with ischemic stroke but without a history of diabetes or hypercholesterolemia [9]. Therefore, more attention should be focused on young patients who have relatively higher FBG within the normal range and increased TG and decreased HDL-C levels.
The present study has some advantages and limitations. This was a case-control study, and the sample size was relatively small, which might limit the generalizability of the results. Our ndings still warrant further studies to con rm our results. Ischemic stroke has an enormous in uence on a working-age adult.
Although several risk factors have been discovered, over a third of ischemic strokes in young adults remain cryptogenic. This study investigated metabolic differences and attempted to explore the possible mechanisms of ischemic stroke in young patients without common risk factors. Taken together, changes in metabolites might be involved in the pathogenesis of young patients with ischemic stroke. However, further independent validations, including human, animal, and cell experiments, are required before translating the results into clinical practice.

Conclusions
Serum metabolomic patterns were signi cantly different between young patients with ischemic stroke and healthy controls. Our study is bene cial in providing a further view into the pathophysiology of young patients with ischemic stroke. However, further independent validations, including human, animal, and cell experiments, are required to con rm these results and get better insight into the underlying mechanisms. Availability of data and materials

Abbreviations
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests