Untargeted metabolomics identifies altered metabolome contributing to hypertension in late adolescents

Importance

Childhood hypertension is a growing health problem, yet its specific biomarkers are not yet fully elucidated.

Objective

The aim of this study was to investigate the functional metabolic alteration associated with hypertension in late adolescents.

Design

This study employed a cluster random sampling method based on the Health Promotion Program for Children and Adolescents. In the first stage in 2020-2021, three schools were selected for hypertension screening, followed by four schools in the second stage in 2022-2023. Hypertensive students in their late adolescent were matched 1:1 with normotensive controls for an untargeted metabolomics study to identify differential metabolites. In vitro cellular experiments were further performed to explore the functional roles of the interested metabolite.

Setting

Two separate case-control studies and cellular experiments.

Participants

In the first stage, a total of 51 late adolescents were identified with hypertension, and then were matched with 51 normotensive controls of similar sex and age from the same dormitory to collect their fasting urine samples. In the second stage, 91 hypertensive adolescents were identified and 91 matched normotensive adolescents were selected from the same dormitory to collect their fasting serum samples.

Main outcomes and measures

Hypertension was diagnosed by blood pressure measurements taken on three separate occasions.

Results

Detailed metabolomic evaluation revealed four distinct metabolites differentially expressed in hypertensive adolescents and control individuals in both urine and serum samples. As compared with the control group, higher levels of 2-hydroxycinnamic acid and xanthine, but lower levels of hypoxanthine and N-acetylornithine were observed in hypertensive adolescents. These metabolites also slightly enhance the discriminatory ability for hypertension based on body mass index Z score, as revealed by increased area under receiver operating characteristic curve. Notably, 2-hydroxycinnamic acid could inhibit cell proliferation, induce oxidative stress and inflammatory responses, and then disrupt the cellular function of human umbilical vein endothelial cells, inferring it as a potential detrimental metabolite for hypertension in adolescents.

Conclusions and relevance

Our result revealed distinct metabolic alterations in urine and serum samples of hypertensive adolescents. Combined with metabonomic and in vitro experiments, 2-hydroxycinnamic acid was identified as a potential detrimental metabolite for hypertension in adolescents.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07914-8.

Highlights

• Four differentially expressed metabolites (2-hydroxycinnamic acid, xanthine, hypoxanthine and N-acetylornithine) were revealed in hypertensive adolescents in both urine and serum samples.

• In vitro cellular experiments showed that 2-hydroxycinnamic acid could inhibit cell proliferation and induce oxidative stress and inflammatory responses, and then disrupt the cellular function of human umbilical vein endothelial cells.

• 2-hydroxycinnamic acid may be a potential detrimental metabolite for hypertension in late adolescents.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07914-8.

Introduction

Childhood hypertension is a growing health concern globally [1]. It not only affects the immediate well-being of children, but also induce long-term impact [2], contributing to a higher risk of cardiovascular disease (CVD) and consequently premature death in adulthood [3, 4]. It is reported that hypertension develops at an earlier age tends to be longer-lasting and harder to manage, and also presents poorer prognosis [5]. In recent decades, the prevalence of hypertension and the levels of blood pressure (BP) levels have increasing rapidly among children and adolescents in China [6]. It is estimated that the overall prevalence of childhood hypertension among Chinese children aged 6–18 years was 3.11% in 2020, equivalent to 6.80 million children [7].

Identification of childhood hypertension poses great challenges due to BP variations along with the age, sex and height. If BP elevated on a single occasion, it is defined as elevated BP; whereas if BP is consistently increasing on three or more occasions, it is classified as hypertension [8, 9]. The prevalence of elevated BP in children gradually decreased as the number of separate BP visits increased [10]. And thus, the pediatric hypertension should be diagnosed based on three or more separated occasions of elevated BP. In addition, adolescent hypertension is usually asymptomatic, which will make the prevention and treatment to be more complicated. Therefore, early detection, diagnosis and intervention of adolescent hypertension, along with a shift from treatment to prevention, play crucial roles in health promotion in their later life.

Since the concentrations of metabolites can directly reflect the underlying biochemical activities and the status of cells or tissues more accurately than transcriptomics and proteomics, metabolomic is recently reported to provide valuable insights into the etiology and identification of biomarkers in a variety of diseases [11]. Hypertension is a metabolite disorder accompanied by lots of metabolism alternations [12]. The metabolomics analysis in adult hypertension has now been extensively studied [13]. Whereas, the metabolic processes vary with age, and thus findings from adult population may be not applicable to children and adolescents [14, 15]. Up to now, limited metabolomic studies have been conducted to reveal the pathogenesis of either pediatric elevated BP [16–18] or hypertension [16, 19, 20]. Meanwhile, current studies in pediatric hypertension often involve small sample sizes and lack rigorous external or experimental validation. Therefore, a systemic metabolomic study is still needed to be conducted to clarify the possible pathogenesis of adolescent hypertension.

Therefore, the aim of this study was to explore specific metabolites as the potential biomarkers associated with hypertension in late adolescents using untargeted metabolomics. Untargeted metabolomic profiling was conducted in two groups. The discovery study included urine samples from 102 subjects, consisting of 51 control adolescents and 51 adolescents with hypertension. An independent validation study comprised serum samples from 182 subjects, with 91 control adolescents and 91 hypertensive adolescents. This analysis was performed using ultra-performance

liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) and ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS), respectively. Additionally, to identify a specific metabolite as the potential biomarker of hypertension in late adolescents, the role of the distinct metabolite in adolescent hypertension was validated in human umbilical vein endothelial cells (HUVECs) by using the in vitro cellular experiments. The flow chat of this study is displayed in Fig. 1.

Fig. 1.

The flowchart of this study

Materials and methods

Untargeted metabolomic profiling

Human subjects

The metabolomic study involved late adolescents participating in the Health Promotion Program for Children and Adolescents (HPPCA) in Suzhou, China [21]. Our study participants were from two stages of the HPPCA program

During the first stage, we conducted a study to examine the prevalence of hypertension in 2020-2021 academic year [22]. Briefly, cluster random sampling was used to select the students of three schools. Within these schools, a total of 51 adolescents were identified with hypertension and were exactly matched with normotensive controls of the same gender and similar age from the same dormitory in a 1:1 ratio. Thereafter, their morning fasting urine samples were collected.

During the second stage, four schools were selected for hypertension screening by using cluster random sampling method based on the HPPCA program conducted in 2022-2023 academic year. Consequently, 91 hypertensive adolescents were identified and 91 matched normotensive adolescents were selected from the same class to collect their fasting serum samples.

Basic demographic characteristics including gender, age, ethnicity, school, grade, place of residence, exercise hour and parent education levels, were collected by questionnaires. The inclusion criteria for the metabolomic study were as follows: (1) participants must be aged 15–19 years, within the range of late adolescent according to some articles [23, 24]. (2) possess local household registration in Suzhou City, and (3) have complete records of blood pressure, height, and weight taken on three separate days. Participants were excluded if they (1) had recently suffered from severe acute infections, (2) exhibited severe anemia, (3) suffered from hepatic or renal insufficiency, or other major organ diseases, (4) had tumors or disabilities or deformities, (5) had autoimmune diseases, (6) suffered from dyslipidemia or high jaundice, (7) had diabetes.

Measurement of BP and definition of adolescent hypertension

The measurement of BP by trained staff of HPPCA project are detailed in our previously study [22]. Briefly, a clinically validated Electronic BP Monitor (i.e., Omron HBP1300, HBP1320) of appropriate size was used to measure BP in adolescents. Participants were required to sit and rest in a quiet environment for at least 15 minutes before the measurement. The BP device was positioned at the same level as the participant’s heart, using a cuff on the right arm. Two consecutive BP values were measured at two-minute intervals each time, and the average of the two closest BP readings were recorded. If the difference between these two consecutive BP values was more than 5 mmHg (1 mmHg = 0.133 kPa), a third reading was taken, and the average of the two closest values was recorded were for this occasion.

The participants with elevated BP were required to re-evaluate their BP at least two weeks later. If elevated BP was again confirmed during the second visit, the third measurement would be performed following the same protocol [22]. The BP measurements during the second and third measurements were conducted by familiar school nurses in the respective student’s campus.

According to the Chinese standard “Reference of screening for elevated blood pressure among children and adolescents aged 7–18 years” (WS/T 610–2018) [8, 9], elevated BP was defined as systolic BP (SBP) and/or diastolic BP (DBP) equal to or above the age-, sex- and height-specific 95th percentile (P₉₅) (eTable 1). The subject with three separate visits of elevated BP was defined as the hypertensive participant [8]. For the matched controls, their BP were also measured at three separate occasions to ensure them have normal BP.

Sample collection

The fasting morning urine were collected in the 102 subjects (51 control and 51 hypertensive adolescents) who met the inclusion criteria. They were advised to refrain from eating after 8 PM at the night before collection. The following morning, approximately 10 mL of midstream second morning urine was collected from each participant and centrifuged for 10 minutes at 10, 000 rpm within two hours. The extracted supernatant was then divided into 2 mL centrifuge tubes, frozen in liquid nitrogen for 15 minutes, and subsequently stored at − 80 °C for further analysis.

To collect the serum in the 182 subjects (91 control and 91 hypertensive adolescents), participants were instructed to fast for 12 hours and avoid strenuous exercise during the day before blood collection. Fasting venous blood was drawn into coagulation-promoting tubes and centrifuged for 10 minutes at 3000 rpm within two hours. The serum was then transferred into 2 mL centrifuge tubes and refrigerated at − 80 °C.

Untargeted metabolomics

The untargeted metabolomics of urine samples in the first stage was performed using UPLC-MS/MS. The workflow for analysis included peak detection, alignment, extraction, integration, missing value filling, feature filtering, and metabolite annotation. Firstly, the raw data were converted to mzXML format using MSConvert from the ProteoWizard software package (version 3.0.8789). Then, the data were processed with XCMS for feature detection, retention time correction, and alignment [25]. Metabolites were identified based on accurate mass measurements (within 30 ppm) and MS/MS data that matched entries in databases such as HMDB (http://www.hmdb.ca), MassBank (http://www.massbank.jp/), LipidMaps (http://www.lipidmaps.org), mzCloud (https://www.mzcloud.org), and KEGG (http://www.genome.jp/kegg/). To correct for systematic bias, robust LOESS signal correction (QC-RLSC) was applied for data normalization. After that, ion peaks with relative standard deviations (RSDs) of less than 30% in the quality control (QC) samples were retained to ensure proper metabolite identification.

Untargeted metabolomics in serum samples in the second stage was conducted using UHPLC-MS/MS. The raw data files generated from this process were processed by Compound Discoverer 3.3 (CD3.3, Thermo Fisher) for peak alignment, peak picking, and quantitation of each metabolite. Subsequently, peak intensities were normalized to the total spectral intensity. The normalized data were then used to predict molecular formulas based on additive ions, molecular ion peaks, and fragment ions. Peaks were matched with the mzCloud (https://www.mzcloud.org), mzVault, and MassList databases to obtain accurate qualitative and relative quantitative results. Key parameters included peak area correction with the first QC sample, an actual mass tolerance of 5 ppm, signal intensity tolerance of 30%, and a specified minimum intensity. Specific experimental procedures were provided in the supplementary material.

Cellular experiments

Chemicals and reagents

The Dulbecco’s modified Eagle medium (DMEM) and fetal bovine serum (FBS) were obtained from Thermo Fisher Scientific (Waltham, MA, USA). Cell counting kit-8 (CCK-8) and cell malonaldehyde (MDA) kit were purchased from Beyotime Biotechnology (Shanghai, China). The hiscript III RT supermix for qPCR and the chamQ universal SYBR qPCR master mix were obtained from Vazyme Co. Ltd (Nanjing, China). The mouse β-actin antibody was purchased from Sigma-Aldrich (St. Louis, MO, USA). The antibodies for nuclear factor erythroid 2-related factor 2 (Nrf2) and heme oxygenase-1 (HO-1) were obtained from Cell Signaling Technology (Danvers, MA, USA). The ECL reagent for western blotting was purchased from Millipore Corporation (Billerica, MA, USA).

Cell culture

Cells were incubated in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin at 37 °C in a 5% CO₂ incubator. For 2-hydroxycinnamic acid (2-HCA) treatment, cells were plated in 96-well plates for CCK-8 assay, 6-well plates for western blot and PCR assays or 10 cm culture plates for MDA assay. After a 24-h incubation, cells were treated with 0, 100, 200 and 400 μM of 2-HCA for another 24 h. The dilution concentration was based on cell proliferation assays. The cells treated with 0.01% DMSO were set to be the solvent control.

CCK-8 assay

Following the treatment of 2-HCA, HUVECs were incubated with CCK-8 solution at 37 °C for another 2 h in accordance with the manufacturer’s protocols. After that, the absorbance at 450 nm was measured by a SYNERGY 2 microplate reader (Bio-Tek, Vermont, USA). The results were expressed as the cell viability (100% of the solvent control).

MDA assay

After 2-HCA treatment, cells were harvested and lysed with lysis buffer. The MDA contents in protein samples were determined by using the cellular MDA assay kit according to the manufacturer’s instructions. The results were expressed as MDA concentration (nmol/mg protein).

Western blotting

After the 24-h treatment of 2-HCA, total proteins in HUVECs were extracted, denatured and separated by SDS-PAGE electrophoresis. After that, the proteins were transferred to a nitrocellulose membrane and blocked with 5% fat-free milk at room temperature for 1 h. And then, the membranes were incubated with the primary antibodies (dilution ratio of 1:1000) at 4 °C overnight and the secondary antibodies (dilution ratio of 1:3000) for 1 h at 37 °C. Finally, the protein bands were visualized with a chemiluminescence imaging system (Syngene, Cambridge, UK). The intensity of the bands was quantified by ImageJ software (NIH, Bethesda, MD, USA).

Quantitative reverse transcription PCR (RT-qPCR)

RT-qPCR was used to detect the expression levels of nucleotide-binding domain, leucine-rich repeat, and pyrin domain-containing protein 3 (NLRP3), interleukin-18 (IL-18), IL-6, IL-8 and vascular endothelial cadherin (VE-cadherin) genes. Total mRNA of HUVECs were extracted by the RNA-Quick Purification Kit and then were reversely transcribed into cDNA by the HiScript III RT SuperMix. The RT-qPCR was performed using a Model 7500 real-time fluorescence quantitative PCR instrument (Applied biosystems, MA, USA), with SYBR Green RT-qPCR master mix. The relative expression levels of target gene mRNA were calculated by the 2^-ΔΔCT method and expressed as fold of control. The primer sequences are listed in eTable 2.

Statistical analysis

Statistical analyses were performed using the statistical software R (R version R-4.4.0), Python (Python 2.7.6 version) and CentOS (CentOS release 6.6). Statistical significance was determined as p < 0.05.

Mean ± standard deviation (SD) or median (quartiles) were used for normal and skewed distribution variables, respectively, and n (%) was used for categorical variables. To compare the characteristics of adolescents with or without hypertension, the paired t-test was used for normal distributed data, Wilcoxon rank sum test for skewed distribution data, and paired chi-square tests for categorical variables. Experimental data were analyzed by one-way ANOVA, followed by the least significant difference or Dunnett’s test when necessary.

Regarding the analyses of metabolic data, the first step involved mean-centering and scaling the data. To investigate metabolites associated with adolescent hypertension, Partial Least Squares Discriminant Analysis (PLS-DA) and Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) were conducted by ropls package. To ensure the reliability of these analyses, we conducted permutation tests through Metaboprofile Cloud to check for overfitting. The descriptive performance of these models was assessed by R²X (cumulative) (with a perfect model having R²X (cum) = 1) and R²Y (cumulative) (with a perfect model having R²Y (cum) = 1) values. The prediction performance was measured by Q²(cumulative) (with a perfect model having Q² (cum) = 1) alongside permutation tests. For a permuted model, the R² and Q² values at the Y-axis intercept should be lower than those of the non-permuted model.

The OPLS-DA enabled the identification of discriminating metabolites using the Variable Importance in Projection (VIP) metric. Metabolites were considered statistically significant if P_{for ttest} < 0.05 and VIP > 1, in accordance with established literature standards [26, 27]. We then performed pathway analysis on the differential metabolites using MetaboAnalyst, which combines results from pathway enrichment analysis with pathway topology analysis. The identified metabolites from the metabolomics study were also mapped to the KEGG pathways for biological interpretation of higher-level systemic functions. Visualization of the metabolites and corresponding pathways was done using the KEGG Mapper tool.

Recognizing that age, gender, body mass index (BMI), exercise hours and parental education levels can influence various metabolic parameters [28, 29], we further conducted Analysis of Covariance (ANCOVA) to ensure the differential metabolites were not affected by these covariables. Receiver operating characteristic (ROC) analysis was performed to evaluate the diagnostic efficacy of a basic model incorporating only BMI z-score (zBMI) for hypertension. Metabolites were then added to construct a combined model, and its diagnostic efficacy was assessed by calculating area under curve (AUC) the corresponding 95% confidence interval (CI). Statistical power, which was analyzed by the method of Chow [30], was rigorously considered in our study to guarantee the reliability and scientific rigor of the results (detailed in supplemtentary methods).

Result

Deferentially expressed metabolites for adolescent hypertension

In the urine case-control study, the hypertensive participants had significantly higher weight, BMI, zBMI, SBP and DBP (p < 0.05), whereas no significant differences were observed among age, height, parental education level and exercise hours. About 39.22% of the subjects was boy. In the serum case-control study, the ratio of boys was up to 63.74% and the hypertensive participants showed higher height, weight, BMI, zBMI, SBP and DBP as compared with the control subjects (p < 0.05). Basic demographic characteristics are displayed in Table 1.

Table 1.

Basic characteristics of normotensive and hypertensive adolescents

Variables	Metabolomic study using urine samples			Metabolomic study using serum samples
	Without hypertension (n = 51)	Hypertension (n = 51)	P	Without hypertension (n = 91)	Hypertension (n = 91)	P
Boys, (n, %)	20 (39.22)	20 (39.22)	>0.99	58 (63.74)	58 (63.74)	>0.99
Age	16.95 ± 0.43	16.99 ± 0.43	0.65	17.86 ± 0.31	17.87 ± 0.32	0.89
Height, (cm)	164.40 (160.00,172.60)	167.40 (160.00,174.80)	0.52	172.00 (163.50, 177.50)	174.00 (165.50, 180.50)	0.03
Weight, (kg)	59.40 (51.30,65.80)	74.20(59.80,88.50)	<0.01	60.000 (54.50, 71.50)	73.00 (60.00, 87.00)	<0.01
BMIa, (kg/m2)	22.02 ± 3.41	27.00 ± 5.40	<0.01	21.77 ± 3.65	24.69 ± 6.09	<0.01
ZBMIb	0.18 ± 1.02	1.41 ± 1.20	<0.01	−0.08 ± 1.15	0.62 ± 02	<0.01
SBPc, (mmHg)	110.00 (95.00,125.00)	135.67 (124.00,147.34)	<0.01	116.26 (110.26, 123.76)	137.26 (132.01, 138.76)	<0.01
DBPd, (mmHg)	67.00 (62.50,69.50)	79.00 (71.50,85.50)	<0.01	68.26 (63.76, 75.01)	82.51 (77.63, 86.26)	<0.01
Father’s education level, (n, %)			0.61			0.24
Junior high school and below	11 (21.56)	11 (21.57)		29 (31.87)	36 (39.56)
Senior high school	27 (52.94)	31 (60.78)		55 (60.44)	44 (48.35)
University and above	13 (25.49)	9 (18.4)		7 (7.69)	11 (12.09)
Mother’s education level, (n, %)			0.34			0.03
Junior high school and below	15 (29.41)	21 (41.18)		48 (52.75)	36 (39.56)
Senior high school	30 (58.82)	27 (52.94)		40 (43.95)	43 (47.25)
University and above	6 (11.77)	3 (5.88)		3 (3.30)	12 (13.19)
Exercise (hours/day), (n, %)			0.80			0.63
<1	46 (90.20)	44 (86.28)		39 (42.86)	42 (46.15)
1 - 2	4 (7.84)	6 (11.76)		34 (37.36)	28 (30.77)
≥2	1 (1.96)	1 (1.96)		18 (19.78)	21 (23.08)

*a, body mass index; b body mass index z-score; c, systolic blood pressure; d, diastolic blood pressure

A total of 1013 metabolites were tested in urine samples, and 322 differential metabolites were revealed, including 279 metabolites identified using the positive ion model and 43 metabolites with the negative ion model (eTable 3). Additionally, 960 metabolites were tested for in serum samples and 74 differential metabolites were founded, with 39 metabolites identified in the positive ion model and 35 metabolites in the negative ion model, respectively (eTable 4).

All identified metabolites from serum and urine samples were separately analyzed using the Metaboanalyst. Distinct clusters of metabolites in urine and serum samples were demonstrated in both positive ion pattern (Fig. 2a, c) and negative ion pattern (Fig. 2b, d) models by PLS-DA score plots. A seven-cross-validation through permutation testing was also tested to validate model performance (eFig. 1 a-b, c-d). Fold changes (using volcano plots) were plotted in the levels of identified metabolites considering the statistically significant difference (p value) and VIP (Fig. 2e and f). Metabolic changes were compared between urine and serum samples by hierarchical cluster analysis to reveal metabolic phenotypes that are potentially involved in adolescent hypertension (eFig. 2 a-b).

Fig. 2.

Metabolic alterations found in urine and serum samples. (a-b) PLS-DA score plots for urine samples in positive ion pattern (a) and negative ion pattern (b). (c-d) PLS-DA score plots for serum samples in positive ion pattern (c) and negative ion pattern (d). (e-f) Volcano plots show the results of pairwise comparisons of metabolites of urine samples (e) and serum samples (f). The vertical dashed lines indicate the threshold for the twofold abundance difference. The horizontal dashed line indicates the p = 0.05 threshold. Between-group comparisons were performed using Student’s t test. metabolites with significant changes are presented in red (up-regulated) or blue (down-regulated). (g-h) Metabolic pathway analysis of urine samples (g) and serum samples (h). Plots depict the computed metabolic pathways as a function of −log₁₀(p) (y-axis) and the pathway impacts of the key metabolites (x-axis) that differed between the normal and hypertensive adolescent groups

To investigate the functional characteristics of differential metabolites in urine and serum samples, the metabolome view map was exhibited to display all matched pathways according to the p values from the pathway enrichment analysis and pathway impact values from the pathway topology analysis. Eight metabolic pathways were defined to be altered in the urine profiles, including biosynthesis of arginine and steroid hormone, the metabolism of tyrosine, the metabolism of alanine, aspartate and glutamate, the metabolism of arginine and proline, the metabolism of cysteine and methionine, the metabolism of caffeine and pyrimidine (p < 0.05，Pathway impact ≥ 0.1) (Fig. 2g). Two metabolic pathways were defined as altered in the serum profiles, including biosynthesis of arginine and steroid hormone (p < 0.05，Pathway impact ≥ 0.1) (Fig. 2h).

Among the differential metabolites identified, the concentrations of five metabolites including 2-HCA, xanthine, hypoxanthine, L-glutamic acid and N-acetylornithine were found to be significantly altered in both urine and serum samples (eTable 5). It is noteworthy that the association of L-glutamic acid with adolescent hypertension displayed different trend in urine and serum samples, then it was excluded for further study. Among the other four metabolites, the concentrations of 2-HCA and xanthine were higher in hypertensive subjects (Fig. 3a–b), while that of hypoxanthine and N-acetylornithine were lower (Fig. 3c–d).

Fig. 3.

The relative abundance of the selected differential metabolites. Scatter plots show the relative abundance of 2-HCA (a), Xanthine (b), Hypoxanthine (c) and N-acetylornithine (d) of hypertensive adolescents and healthy controls

Figure 4 illustrates the diagnostic efficacy of incorporating selected differential metabolites for hypertension in late adolescents into the basic zBMI model. The zBMI model alone achieved an AUC of 0.77 (95% CI: 0.69–0.85) in urine samples and 0.64 (95% CI: 0.57–0.71) in serum samples. Notably, the combination of 2-HCA and zBMI yielded an increased AUC of 0.81 (95% CI: 0.73–0.90) in urine and 0.65 (95% CI: 0.57–0.74) in serum. Next, xanthine and zBMI showed an AUC of 0.84 (95% CI: 0.76–0.91) in urine samples, while in serum samples, it had an AUC of 0.65 (95% CI: 0.57–0.73). For hypoxanthine and zBMI, the AUC was 0.96 (95% CI: 0.92–1.00) in urine samples and 0.67 (95% CI: 0.59–0.75) in serum samples. Lastly, N-acetylornithine and zBMI had an AUC of 0.82 (95% CI: 0.73–0.90) in urine samples and 0.68 (95% CI: 0.60–0.76) in serum samples.

Fig. 4.

ROC curves showing the diagnostic performance of the basic zBMI model with and without the addition of four differential metabolites in late adolescents in urine samples (a) and serum samples (b)

As shown in eTable 6, hypoxanthine, xanthine, and 2-HCA were significantly differently expressed in individuals with hypertension compared to those without it in both urine and serum samples, even after adjusting for age, gender, zBMI, exercise hours and parental education levels (p < 0.05). However, N-acetylornithine showed statistically significance in the urine samples (p < 0.05) but borderline significance in the serum samples (p = 0.05).

Xanthine and hypoxanthine were enriched in the purine metabolism pathway, while N-acetylornithine was enriched in the arginine biosynthesis pathway (eTable 7). Since there are already substantial studies involved in humans related with the role of purine metabolism pathway [31–33] and arginine biosynthesis pathway [34–36] in the progression of CVD, 2-HCA was finally chosen into the experimental stage.

Validation of 2-HCA as the potential detrimental metabolite in vitro

Next, we investigated the toxic effects of 2-HCA on HUVECs. Firstly, CCK-8 assay was used to reveal the effect of 2-HCA on the cell proliferation of HUVECs. As shown in Fig. 5a, after a 24-h treatment of 2-HCA, significant decreases in cell viability were observed in 200 µM- and 400 µM-treated groups (p < 0.05). Secondly, 2-HCA induced significantly oxidative damage in HUVECs as indicated by the observable increases in MDA concentration and the down-regulation of Nrf2 and HO-1 proteins, the typical regulators in cellular antioxidant response (Fig. 5b–d, p < 0.05). Thirdly, as shown in Fig. 5c–e, significant increases in expression of phosphorylated NF-κB-p65 protein and expression of NLRP3, IL-18, IL-6 and IL-8 genes were observed in HUVECs after the 2-HCA treatment, suggesting the activation of NLRP3 inflammasome and the following inflammatory response. Finally, as shown in Fig. 5f, expression of VE-cadherin gene was significantly down-regulated in the 200 µM and 400 µM 2-HCA treatment groups (p < 0.05), indicating the detrimental effects of 2-HCA on epithelial function.

Fig. 5.

Toxic effects of 2-HCA on HUVECs. HUVECs were treated with 0, 100, 200, 400 µM of 2-HCA for 24 h. (a) Cell viability detected by the CCK8-assay kit. (b) MDA content measured by cellular MDA assay. (c-d) Western blotting analysis of Nrf-2, HO-1 proteins, phosphorylated NF-κB p65 and NF-κB p65 proteins. (e) Expression of NLRP3, IL-18, IL-6 and IL-8 genes detected by RT-qPCR assay. (f) Expression of VE-Cadherin gene detected by RT-qPCR assay. Data is presented as mean ± SD of three individual experiments. * p < 0.05, ** p < 0.01, versus the solvent control

Discussion

In the present study, by using the metabolomics we revealed a distinct metabolic phenotype of hypertension among late adolescent. It was demonstrated that the concentrations of 2-HCA, xanthine, hypoxanthine and N-acetylornithine were significantly altered in the circulating flue of hypertensive participants, as compared with the control subjects. Hypoxanthine and N-acetylornithine were negatively associated with adolescent hypertension, while 2-HCA and xanthine showed positive associations.

Among the two identified metabolites in the purine metabolism pathway, xanthine was positively associated with adolescent hypertension, whereas hypoxanthine was negatively associated with adolescent hypertension. Supporting our findings, the recent Boston Birth Cohort study reported prospective associations of cord metabolites, including hypoxanthine, xanthine, and xanthosine, with elevated risk of high BP from ages 3 to 18 years [17]. Hypoxanthine can be initially oxidized to xanthine, which can then be further oxidized to uric acid [37]. Uric acid, the end-product of purine metabolism, plays an important role in the development of CVD [38]. Additionally, urate production process also generates reactive oxygen species that can lead to oxidative stress, endothelial dysfunction, and impairments in cardiac and vascular functions [39, 40]. High level of uric acid may affect BP by inducing inflammation, oxidative stress, and endothelial damage [37]. A metabolome-wide association study conducted in cord blood of children and adolescents showed that xanthine and hypoxanthine were strongly associated with BP [17].

In the current study, N-acetylornithine displayed negative associations with adolescent hypertension. A Mendelian randomization study reported that N-acetylornithine was negatively correlated with essential hypertension [41], which aligns with our results. N-acetylornithine is an important intermediate in arginine metabolism. Arginine is an essential amino acid participated in the production of nitric oxide, which is crucial for vascular health via promoting vasodilation and preventing platelet aggregation [42]. Studies indicate that levels of arginine are negatively associated with the occurrence of hypertension [43], and arginine supplementation could potentially help lower BP [44, 45]. Additionally, N-acetylornithine contributes to the synthesis of ornithine [46], which has been shown to be negatively correlated with lower BP [47]. Thus, in this study, we propose that N-acetylornithine may act as a protective metabolite against hypertension in late adolescents, potentially through the arginine or ornithine pathways.

2-HCA, known as o-coumaric acid, is a hydroxy derivative of cinnamic acid with three isomers: o-coumaric acid, m-coumaric acid and p-coumaric acid, including antioxidant, anti-inflammatory, bacteriostatic, and antitumor effects [48, 49]. Among them, p-coumaric acid, referred to 4-hydroxycinnamic acid, has been reported to decrease the production of pro-inflammatory cytokines and display significant anti-inflammatory effects in adjuvant-induced arthritic rats [50, 51]. However, the properties and the biological effects of the other isomer remain unclear. A metabolomic study on rats exposed to acrylamide, observed cardiovascular toxicity and a notable increase in 2-HCA in rat serum [52], which were generally in accordance with our findings. In our study, in vitro cellular assays revealed that 2-HCA could induce adverse effects in HUVECs, as evidenced by decreased cell viability, down-regulation of the cellular antioxidant response (specifically the Nrf2 signaling pathway), and activation of the NLRP3 inflammasome—a crucial component of the immune defense in organisms [53, 54]. These effects subsequently triggered lipid peroxidation and inflammation, indicated by increased concentration of MDA and the expression of inflammatory genes, ultimately causing endothelial dysfunction characterized by decreased expression of VE-cadherin [55]. Therefore, the positive association of 2-HCA and pediatric hypertension is biologically plausible, and we proposed 2-HCA as a potential detrimental metabolite for hypertension in late adolescents.

Taken together, the current study identified four metabolites that may play an important role in the development of late adolescent hypertension. Combined with metabolomics and in vitro experiments, we proposed 2-HCA as a novel and detrimental metabolite for hypertension in late adolescents. This study presents several innovative aspects. Firstly, to the best of our knowledge, our study, encompassing of urine and serum samples, involves the largest sample size of adolescents with hypertension in the field of metabolomics. Secondly, cellular experiments were conducted to demonstrate the potential role of the metabolite of interest in the pathogenesis of hypertension. However, there are also some limitations in this study. Firstly, the included participants were exclusively from Suzhou City, China, making it challenging to generalize the results. Secondly, the epidemiological design of our study is cross-sectional, which limits the ability to establish causality between identified metabolites and hypertension and requires cautious when drawing a causal conclusion, although we attempted to provide supporting evidence by conducting cellular experiments. To establish temporality and causality, well-designed prospective cohort or intervention studies should be undertaken. Thirdly, we did not specifically investigate the dietary patterns of the participants. Although almost all of the participants ate at the school cafeteria, suggesting that their dietary habits were relatively uniform, variations in socioeconomic status beyond parental education may still act as residual confounders affecting dietary status. Fourthly, human umbilical vein endothelial cells can not accurately reflect the endothelial condition during adolescence, which is also a limitation of our study. Primary endothelial cells from adolescent sources would be more relevant to the actual physiological context in future studies. Fifthly, despite our careful adjustment for known confounders, the possibility of residual confounding due to unmeasured factors or imperfect measurement of included covariables cannot be fully ruled out. Fifthly, blood samples showed a lower power (74.9%) to detect differences in 2-HCA between cases and controls, whereas urine samples demonstrated higher power (93.6%). Together, these results confirm that our datasets provide sufficient statistical power for this comparison. Finally, urine and serum samples differ in metabolite composition, and metabolomics results were analyzed across different metabolomics platforms, which may have led to omission of certain metabolites. However, the use of different platforms and varying sample sources ultimately underscores the robustness of our findings.

Conclusion

In this study, by using metabolomics analyses, we identified four differential metabolites including 2-HCA, xanthine, hypoxanthine and N-acetylornithine in hypertensive adolescents. Along with the toxic effects of 2-HCA on HUVECs, we therefore proposed 2-HCA as a functional biomarker for adolescent hypertension.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

BP

Blood pressure

HUVECs

Human umbilical vein endothelial cells

HPPCA

Health promotion program for children and adolescents

SBP

Systolic BP

DBP

Diastolic BP

QC

Quality control

CCK-8

Cell counting kit-8

MDA

Malonaldehyde

Nrf2

Nuclear factor erythroid 2-related factor 2

HO-1

Heme oxygenase-1

2-HCA

2-hydroxycinnamic acid

NLRP3

3-nucleotide-binding domain (NBD), leucine-rich repeat (LRR), and pyrin domain (PYD)-containing protein 3

IL

Interleukin

VE-cadherin

Vascular endothelial cadherin

VIP

Variable importance in projection

BMI

Body mass index

ROC

Receiver operating characteristic

AUC

Area under curve

CI

Confidence interval

zBMI

BMI z-score

Author contributions

All authors have participated in the concept and design of this manuscript. Ying Wang, Wenxin Ge, Fei Liang and Yue Xiao participated in the collecting of urine and blood samples. Ying Wang, Wenxin Ge and Jieyun Yin participated in the analysis and interpretation of data and drafting of the manuscript. Lili Xin, Weici Yan and Yue Su participated in the in vitro cell experiment; the rest of the authors took part in the revising of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (82204070 and 82273635), Gusu Health Talents Program Training Project in Suzhou, China (GSWS2020100), Suzhou Municipal Priority Medical Disciplines (SZXK202523) and by the Priority Academic Program Development of Jiangsu Higher Education Institutions. No potential conflict of interest was reported by the authors.

Data availability

The data that support the findings of this study are available from the corresponding author.

Declarations

Ethics approval

This study was approved by the Ethics Committee of Soochow University and the Center for Disease Control and Prevention of Suzhou, China. The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Soochow University and the Center for Disease Control and Prevention of Suzhou, China. Sign informed consent were provided by both the participants and their parents.

Consent for publication

Informed consent was obtained from all individual participants and their legal guardians included in the study.

Competing interests

No potential conflict of interest was reported by the authors.