Lipid profile shift in high-altitude migrants as a key promoting factor for chronic myocardial injury revealed by lipidomics

Abstract

Background

Lipid metabolic dysregulation contributes to cardiovascular disease, yet its role in high-altitude heart disease (HAHD) remains unclear.

Methods

We enrolled 525 high-altitude immigrants and 98 plain controls. Cardiac function and serum lipids were clinically assessed. Untargeted lipidomics was performed to profile circulating lipids in a matched discovery subgroup (N = 15 per group): plain controls, high-altitude residents with normal cardiac function, and high-altitude residents with cardiac dysfunction. Correlation analyses identified lipids linked to myocardial injury biomarkers. ROC analysis evaluated diagnostic potential.

Results

Cardiac dysfunction prevalence was 46.4% in high-altitude subjects and correlated with hyperlipidemia. Lipidomics revealed HAHD-specific remodeling characterized by increased glycerolipids (e.g., triglycerides, diglycerides) and free fatty acids, alongside decreased sphingolipids (ceramides, hexosylceramides) and phospholipids (phosphatidylserine, phosphatidic acid). Twenty-five lipids were associated with myocardial injury biomarkers, of which eight, notably PS(20:4_20:4) and DG(16:0_20:4), showed good discriminatory performance (AUC > 0.8 in the matched group). These lipids correlated with immune traits, suggesting lipid-immune crosstalk.

Conclusion

This study identifies a distinct lipid signature in HAHD and implicates lipid-immune interactions in disease progression. These exploratory findings provide mechanistic insights and warrant further validation as potential biomarkers for early detection and targeted intervention of high-altitude cardiac dysfunction in future independent cohorts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-02918-9.

Introduction

Given the increasing frequency of human sojourns in high-altitude regions, an environment of hypobaric hypoxia with low ambient partial pressure of oxygen, altitude-related heart disease has raised significant public health concerns. High-altitude heart disease (HAHD) is an important type of chronic mountain sickness resulting from prolonged hypobaric hypoxia, which may progress into pathological cardiac hypertrophy and, in severe cases, lead to heart failure [1]. In the progression of HAHD, myocardial injury represents a critical and frequently encountered pathological event [2, 3]. Recent reports revealed that myocardial injury at high altitude is driven by oxidative stress disturbances [4], ferroptosis [5, 6], and imbalances in energy metabolism, specifically lipid metabolism and carbohydrate metabolism [7, 8]. Among these, lipid metabolism holds particular significance, as it directly regulates fatty acid oxidation, the core pathway for mitochondrial energy production in cardiomyocytes [9]. Dysregulated lipid metabolism not only disrupts this primary energy source but also indirectly impairs other metabolic pathways by damaging mitochondrial function [10], making it the pivotal hub where oxidative stress, ferroptosis, and global energy disturbance converge in high-altitude cardiac injury.

In multiple clinical cohort studies, elevated plasma lipid levels, such as total cholesterol, low-density lipoprotein cholesterol, and triglycerides (TGs), are closely associated with impaired cardiac structure and function, encompassing myocardial injury, ventricular hypertrophy, and diastolic dysfunction [11–13]. Furthermore, extensive studies have demonstrated a direct positive correlation between hypoxia and circulating TG levels [14]. In vivo evidence from murine models has established that cardiac lipid accumulation directly impairs cardiac function [15]. Mechanistically, prolonged lipid overload has been shown to trigger Mfn2 degradation, thereby inducing maladaptive cardiac lipotoxicity [16]. Additionally, palmitic acid has been demonstrated to drive endothelial damage and cardiovascular dysfunction via PKM2-C31 palmitoylation in mice, a mechanism that may underlie the elevated risk of major adverse cardiovascular events observed in clinical cohorts [17]. These findings suggest that dyslipidemia may play a key role in the pathogenesis of HAHD by driving myocardial injury, and the lipids may hold promise as biomarkers for the early detection and prognosis of HAHD. Nevertheless, current evidence remains largely correlative, and the causal mechanisms linking lipid metabolism to hypoxic cardiac damage remain poorly defined, creating an urgent need to elucidate the specific lipid species and regulatory circuits involved.

Lipidomics enables comprehensive profiling and quantification of lipid species in biological systems, allowing identification of disease-specific lipid molecules and providing mechanistic insights into their functional roles in complex disorders [18]. In cardiovascular pathophysiology, lipidomic analysis has uncovered clinically relevant lipid biomarkers. For example, elevated circulating levels of specific ceramides (Cers), notably Cer(d18:1/16:0) and Cer(d18:1/24:0), are significantly associated with the development of heart failure and increased cardiovascular risk [19]. Similarly, higher plasma concentrations of sphingomyelin (SM), such as SM(d18:1/16:0), are linked to an elevated incidence of sudden cardiac death [20]. Beyond biomarker discovery, lipid-mediated mechanisms have also been delineated. Duan et al. [21] demonstrated that sphingosine-1-phosphate (S1P) mediates cardioprotective effects via endothelial S1P receptor 1 signaling yet promotes adverse cardiac outcomes through S1P receptor 2. Despite these advances in understanding lipid dynamics and function in cardiovascular diseases, studies applying lipidomics in high-altitude cardiac injury remain limited. Previous metabolomic and lipidomic studies in high-altitude settings have primarily focused on acute hypoxia exposure (e.g., hours to days) using animal models or small human cohorts, with an emphasis on systemic metabolic changes such as enhanced glutamine and fatty acid metabolism [22–25]. Recently, several studies have begun to explore the interplay between glucose and lipid metabolism under chronic hypoxia using animal models [26] and have characterized immune and metabolic profiles during high-altitude mountaineering [27]. However, these studies either focused on preclinical models or examined relatively short-term exposure, leaving a gap in our understanding of the lipidomic alterations specifically associated with chronic myocardial injury in long-term high-altitude migrants.

In this study, we first conducted a comprehensive analysis on a large sample group comprising 525 high-altitude immigrants and 98 plain controls to investigate the correlation between myocardial injury and hyperlipidemia within the high-altitude population. Then, we employed plasma untargeted lipidomics in 45 well-matched individuals to map the lipid metabolic profile of long-term high-altitude migrants, focusing on lipid species closely associated with myocardial injury. By integrating lipidomic profiles with clinical cardiac function data, we aim to elucidate the pathophysiological role of lipid disturbances in high-altitude-induced myocardial injury, identify potential biomarkers, and provide novel mechanistic insights into the development of HAHD.

Methods

Clinical data sources and sample collection

This study enrolled 98 male adults residing in Chengdu, Sichuan Province, and 525 male adults who had migrated from the lowlands to Tibet (≥ 3,500 m). A semi-structured questionnaire was administered to evaluate baseline characteristics, including age, altitude of residence, duration of residence in Tibet, body mass index (BMI), dietary habits, smoking and alcohol consumption histories, antibiotic use, and medical history. Enrolled young adults, aged 19 to 45 years, shared comparable socioeconomic backgrounds and dietary environments and were free from colds, fever, chronic conditions such as diabetes and hypertension, or any medication use within the two weeks prior to enrollment. Additionally, blood routine examinations, blood biochemistry tests, echocardiography (ECHO), and electrocardiograms (ECG) were performed to collect clinical baseline data. Blood samples were obtained from all participants and centrifuged for 15 min at 2823 × g. The supernatant was then isolated and stored at − 80 °C. This study was conducted in accordance with the Declaration of Helsinki, and informed consent was obtained from all participants. Ethical approval was granted by the Beijing Institute of Radiation Medicine Ethics Committee (No. AF/SC-08/02.153).

To assess the impact of high altitude on cardiac injury, participants were categorized into three groups based on inclusion and exclusion criteria defined by creatine kinase isoenzyme (CKMB) levels, ECG, and ECHO results. CKMB was selected as the primary marker of myocardial injury in this study, reflecting our focus on chronic myocardial injury during long-term high-altitude adaptation rather than acute cardiac events, a choice further supported by previous studies that validated CKMB as a reliable marker in high-altitude populations, where it showed significant correlation with cardiac troponin I and good predictive performance for myocardial injury (AUC = 0.749) [3]. Using SPSS case–control matching for altitude, age, BMI, and duration of residence, 15 participants per group were selected for lipidomic analysis. The groups comprised a plain group with normal cardiac function (CKMB ≤ 24 U/L, normal ECG and ECHO); a high-altitude group with normal cardiac function (CKMB ≤ 24 U/L, normal ECG and ECHO); and a high-altitude group with abnormal cardiac function (CKMB > 24 U/L, abnormal ECG and/or ECHO).

Plasma lipidomic analysis

Frozen plasma samples of the three groups were thawed on ice, vortexed for 10 s, and centrifuged at 3,000 × g for 5 min at 4 °C. A 50 μL aliquot of each sample was homogenized with 1 mL of extraction solvent (methanol (Merck, Darmstadt, Germany)/tert-butyl methyl ether (Merck, Darmstadt, Germany)/internal standard mixture (Avanti Polar Lipids, Alabaster, AL), v/v/v) by vortexing for 15 min. After adding 200 μL of water and vortexing for 1 min, samples were centrifuged at 12,000 × g for 10 min at 4 °C. The supernatant (500 μL) was collected, concentrated, and reconstituted in 200 μL of solvent before LC–MS/MS analysis.

Untargeted lipidomics was performed using an LC–MS/MS system consisting of an ExionLC AD UPLC coupled to a QTRAP® 6500 + mass spectrometer (SCIEX). Chromatographic separation was achieved on a Thermo Accucore™ C30 column (2.6 μm, 2.1 × 100 mm) maintained at 45 °C with a flow rate of 0.35 mL/min. The mobile phase consisted of (A) acetonitrile (Merck, Darmstadt, Germany)/water (60:40, v/v) with 0.1% formic acid (Sigma-Aldrich, St.Louis, MO.USA) and 10 mmol/L ammonium formate (Sigma-Aldrich, St.Louis, MO.USA), and (B) acetonitrile/isopropanol (Merck, Darmstadt, Germany) (10:90, v/v) with 0.1% formic acid and 10 mmol/L ammonium formate. The gradient program was as follows: 80% A (0 min), 70% A (2.0 min), 40% A (4 min), 15% A (9 min), 10% A (14 min), 5% A (15.5–17.3 min), returning to 80% A at 20 min.

The mass spectrometer was operated in both positive and negative ionization modes with the following parameters: ion spray voltage ± 5500 V (positive/negative), source temperature 500 °C, curtain gas 35 psi, and ion source gases 1 and 2 at 45 and 55 psi, respectively. Instrument calibration was performed using 10 and 100 μmol/L polypropylene glycol solutions for triple quadrupole and LIT modes. Multiple reaction monitoring experiments were conducted with nitrogen collision gas set to 5 psi, with individual multiple reaction monitoring transitions optimized for declustering potential and collision energy. Lipid identification and quantification were performed using the MetWare platform (http://www.metware.cn/) based on the QTRAP 6500 LC–MS/MS system. Specific multiple reaction monitoring transitions were monitored for each metabolite according to its retention time.

Statistical analysis

Baseline clinical data were analyzed using GraphPad Prism 9.0 and R version 4.3.1 software. Intergroup comparisons were performed as follows: for normally distributed continuous variables, unpaired two-tailed Student’s t-tests were used for comparisons between two groups when variances were equal, while Welch's t-tests were applied when variances were unequal. For non-normally distributed data, the Wilcoxon rank-sum test was employed. Categorical variables were analyzed using chi-square tests (χ²), with a significance level set at P < 0.05. Experimental data are presented as mean ± SD. The effects of traditional lipids on cardiac injury markers were evaluated using robust regression analyses controlling for BMI, age, altitude, and duration of residence. Similarly, associations between immune traits and plasma CKMB or plasma TG levels were accessed by robust regressions controlling for BMI, age, and duration of residence; statistical significance was defined as P < 0.05, with adjustment for false discovery rate (FDR).

Untargeted lipidomic profiling was conducted using R version 4.3.1 software. Lipid abundances were normalized via Z-score transformation. Partial least squares-discriminant analysis (PLS-DA) was performed using the MetaboAnalyst 5.0 platform. Model complexity was optimized via fivefold cross-validation. Model reliability was assessed by permutation testing (1,000 permutations). Differential lipids were identified based on the following criteria: P value < 0.05 from the Wilcoxon rank-sum test, absolute log₂(FC) > 1, and variable importance in projection (VIP) score > 1 from PLS-DA. K-means clustering was employed to group lipids according to their abundance profiles. Pathway enrichment analysis was performed using the Small Molecule Pathway Database through the MetaboAnalyst 5.0 platform. The correlation between myocardial injury indicators and lipid species was assessed using the Mantel test as implemented in the LinkET version 0.0.7.4 package, and then validated through multiple linear regression and logistic regression, with adjustment for glucose, age, and BMI. Receiver operating characteristic (ROC) curve analysis was conducted to evaluate the predictive performance of candidate lipids, with FDR-adjusted P < 0.05 considered statistically significant. To assess model stability and avoid overfitting, we performed leave-one-out cross-validation (LOOCV) and bootstrap resampling (1,000 iterations). Model calibration was assessed using calibration curves with 10-group binning and the Hosmer–Lemeshow goodness-of-fit test (P > 0.05 indicating adequate calibration), along with the Brier Score. Spearman correlation analyses between lipids and immune traits were also conducted, with associations considered statistically significant at FDR-adjusted P < 0.05.

Results

High risk of impaired cardiac function in high-altitude migrants

Using identical criteria for ECG and ECHO interpretation, we identified abnormal cardiac function in 46.4% of high-altitude participants (N = 525) but in none of the plain participants (N = 98) (Fig. 1A). This observation aligns with the low baseline prevalence of cardiac dysfunction in healthy young adults at low altitude and highlights the markedly higher prevalence of cardiac abnormalities associated with long-term high-altitude residence [28, 29]. Notably, the high-altitude population had significantly higher age and BMI than the plain population (both P < 0.001; Supplementary Table S1). Although age and BMI are established risk factors for cardiovascular disease in general populations, we sought to determine their specific role in our unique cohort exposed to chronic high-altitude hypoxia. Logistic regression analysis conducted within the high-altitude population showed that, in this specific group, neither age nor BMI was significantly associated with the risk of abnormal cardiac function (age: OR = 1.02, P = 0.337; BMI: OR = 0.94, P = 0.118; Supplementary Fig. S1). Although age and BMI differed significantly between plain and high-altitude groups (Supplementary Table S1), intergroup logistic regression was not feasible due to the absence of cardiac abnormalities in the plain controls. Importantly, the lack of association within the high-altitude group argues against these demographic factors being the primary drivers of the striking disparity in dysfunction prevalence, supporting a predominant role of altitude-related factors.

Fig. 1.

High risk of abnormal cardiac function induced by high altitude. A The schematic plot of the region and the risk of abnormal cardiac function of the enrolled population. B The ratio of electrocardiogram (ECG) and echocardiograph (ECHO) abnormalities of the enrolled population based on altitudes. C The frequency of ECHO abnormalities in high-altitude migrants. D The frequency of ECG abnormalities in high-altitude migrants

Among high-altitude participants with abnormal cardiac function, the prevalence of ECG abnormalities was higher in lower-altitude regions, whereas ECHO abnormalities increased with elevation (Fig. 1B). No ECG or ECHO abnormalities were observed in the plain participants. Among high-altitude migrants, common ECHO findings included mild tricuspid regurgitation, mild pulmonary valve regurgitation, mild mitral regurgitation, and ventricular septal thickening (Fig. 1C). ECG abnormalities predominantly manifested as sinus arrhythmia, sinus bradycardia, incomplete right bundle branch block (RBBB), and right axis deviation (Fig. 1D). Mild regurgitation of the mitral, pulmonary, and tricuspid valves was primarily physiological or mildly pathological, often associated with valvular structural anomalies, mild right ventricular dilation, or elevated pulmonary artery pressure. Ventricular septal thickening may reflect myocardial hypertrophy. Sinus arrhythmia with right axis deviation and incomplete RBBB could indicate structural cardiac alterations or conduction system abnormalities. These findings suggest that high-altitude exposure increases cardiac load and elevates the risk of myocardial injury in migrant populations.

The effects of dyslipidemia on high-altitude myocardial injury

To investigate whether lipid metabolism is associated with myocardial injury under prolonged high-altitude conditions, we performed correlation analysis between traditional lipid traits (TG, total cholesterol (CHO), low-density lipoprotein (LDL), high-density lipoprotein (HDL), apoprotein B (APOB), apoprotein A-1 (APOA1)) and biochemical markers of myocardial injury (CKMB, lactate dehydrogenase (LDH), α-hydroxybutyrate dehydrogenase (α-HBDH), and aspartate aminotransferase (AST)) in all plain and high-altitude populations. As shown in Fig. 2A, C–D, significant positive correlations were observed between six traditional traits and LDH, α-HBDH, AST (FDR-adjusted P < 0.05). TG (β = 1.66, 95% CI: 0.33 to 2.98, FDR-adjusted P < 0.05), LDL (β = 1.21, 95% CI: 0.35 to 2.06, FDR-adjusted P < 0.05), and CHO (β = 1.08, 95% CI: 0.35 to 1.80, FDR-adjusted P < 0.05) had significant positive effects, while APOA1 (β = −0.02, 95% CI: −0.04 to 0, FDR-adjusted P < 0.05) had a slight negative effect on CKMB (Fig. 2B). These results provide compelling evidence suggesting that lipid metabolism is closely linked to myocardial injury in high-altitude, highlighting the potential significance of lipid-related factors in the pathophysiology of HAHD.

Fig. 2.

Associations of traditional lipids with cardiac injury risk. A Associations between AST and traditional lipids. B Associations between CKMB and traditional lipids. C Associations between α-HBDH and traditional lipids. D Associations between LDH and traditional lipids. Robust regression analysis was performed, adjusting for age, altitude, BMI, and residence duration. The β-coefficients with error bars representing 95% CI were shown separately for significance (red) or not (grey). N = 623; *, P < 0.05; **, P < 0.01; ***, P < 0.001

Descriptive data of matched groups

To characterize the lipid metabolic profile of high-altitude migrants and identify distinctive features in myocardial vulnerability, we established three matched groups. These groups were matched based on altitude, age, BMI, and duration of residence, with each group comprising 15 participants: a plain control with normal myocardial function (PLAIN), a high-altitude group with normal myocardial function (HCN), and a high-altitude group with myocardial abnormalities (HCA) (Fig. 3). As shown in Supplementary Table S2, no significant differences were observed between the PLAIN group (age: 23.87 ± 1.51 years; BMI: 21.19 ± 0.73 kg/m²) and high-altitude groups (HC, age: 24.47 ± 2.33 years; BMI: 22.27 ± 2.45 kg/m²) in baseline demographics. The HCN and HCA groups were further matched for altitude (HCN: 4370 m, HCA: 4340.67 ± 101.73 m), duration of residence (HCN: 6.13 ± 2.33 years, HCA: 6.00 ± 2.51 years), age (HCN: 24.33 ± 2.29 years, HCA: 24.60 ± 2.44 years), and BMI (HCN: 21.97 ± 1.59 kg/m², HCA: 22.58 ± 3.12 kg/m²).

Fig. 3.

Flowchart illustrating the inclusion and exclusion process for matching the population in lipidomic analysis. PLAIN: the plain group with normal myocardial function (CKMB ≤ 24 U/L, normal ECG, and ECHO). HCN: the high-altitude group with normal myocardial function (CKMB ≤ 24 U/L, normal ECG, and ECHO). HCA: the high-altitude group with abnormal myocardial function (CKMB > 24 U/L, abnormal ECG and/or ECHO)

Biochemical analyses revealed significant differences between the PLAIN and the HC groups in glucose metabolism (P < 0.01), myocardial injury markers (P < 0.01), and classic lipid profiles (P < 0.05) (Supplementary Table S2), indicating that chronic high-altitude exposure substantially impacts systemic metabolism and elevates myocardial injury risk. As expected, the HCA group exhibited markedly elevated myocardial injury markers (CKMB, α-HBDH, LDH, and AST, with all P values < 0.05) compared to the HCN group (Supplementary Figs. S2A–D). ECHO results further demonstrated significantly reduced ejection fraction (EF, P < 0.05) and fractional shortening (FS, P < 0.05)) in the HCA group, confirming impaired cardiac function in this population (Supplementary Figs. S2E–G).

Differentiated lipids in high-altitude migrants

Through untargeted lipidomic analysis on the matched groups (N = 15 per group, total N = 45), 1,274 lipids were identified for subsequent analysis. PLS-DA demonstrated clear separation based on blood lipid profiles among the three groups (Fig. 4A). The optimal two-component model (R²X = 0.191, R²Y = 0.404, Q² = 0.37) was selected based on accuracy and Q², validated by permutation testing (1,000 permutations, P < 0.001) (Fig. S3A-B). To investigate lipid alterations in high-altitude adaptation, we compared the combined high-altitude groups (N = 30) with plain controls (N = 15), revealing 438 differentially expressed lipids (VIP ≥ 1, P < 0.05). Among these, 232 lipids were upregulated, including TG(16:0_18:1_24:1), TG(18:1_20:1_20:1), LPE(22:1), etc., while 206 were downregulated, including SM(d18:2/24:2), CerP(d18:2/24:0), etc. (Fig. 4B).

Fig. 4.

Differential lipid analysis of the general high-altitude population (HC vs. PLAIN). A PLS-DA score plot of lipid species between three groups. B Volcano plot of the differential lipids. C Lipid subclass contents and counts of the differential lipids. D Bubble plot of fatty acyl chain length and unsaturation in down-regulated lipids. E Bubble plot of fatty acyl chain length and unsaturation in up-regulated lipids

Detailed subclass analysis demonstrated, as shown in Fig. 4C, significant increases in monoacylglycerols (MG), diacylglycerols (DG), and TG in the HC groups (P < 0.05), with 87 TG species upregulated versus only 7 downregulated. Free fatty acids (FFAs) showed balanced regulation in species number but significant content increase (P < 0.05). Sphingolipid analysis revealed decreased ceramide phosphate (CerP) and sphingomyelin (SM) levels (P < 0.05), while Cers exhibited a 2:1 down/up regulation ratio without reaching statistical significance. Glycerophospholipid subclasses displayed complex regulation patterns: lysophosphatidylglycerol (LPG), lysophosphatidic acid (LPA), phosphatidylserine (PS), and phosphatidic acid (PA) were significantly decreased (P < 0.05), whereas lysophosphatidylethanolamine (LPE), lysophosphatidylcholine (LPC), phosphatidylinositol (PI), phosphatidylcholine (PC), and phosphatidylethanolamine (PE) showed increased levels (P < 0.05). At the species level, LPG, LPA, and PA subclasses were completely downregulated, while other subclasses exhibited balanced up/down regulation. Fatty acyl composition analysis revealed that both up- and downregulated lipids were predominantly composed of long-chain (C13–18) and very long-chain (> C18) fatty acyls. Of note, downregulated lipids were enriched in di-unsaturated long chains with evenly distributed unsaturation (0–6) in very-long-chain species, while upregulated lipids showed a predominance of saturated/mono-unsaturated long chains and mono-unsaturated very-long chains (Fig. 4D–E). The counts and relative content analysis of fatty acid chains revealed high enrichment of C18:2 in GP and C18:1 in GL (Figs. S4A–B). These comprehensive lipidomic alterations, observed in a limited sample size, suggest significant metabolic remodeling in high-altitude migrants, potentially reflecting adaptive responses to chronic hypobaric hypoxic conditions, warranting validation in larger independent cohorts.

Differentiated lipids in high-altitude migrants with impaired cardiac function

To identify candidate lipids associated with myocardial injury in high-altitude populations, all 1,274 detected lipids across the three groups were classified into six clusters (Clusters 1–6) using K-means clustering (Fig. 5A). Importantly, the lipid profiles in Clusters 5 and 6 exhibited distinct patterns in the HCA group that were independent of both the PLAIN and HCN groups, suggesting their potential relevance to high-altitude-associated myocardial injury. Functional annotation revealed that Cluster 5 was predominantly composed of glycerophospholipids and sphingolipids, which are implicated in steroid biosynthesis (Fig. 5A). Cluster 6, however, was enriched in GLs and GPs, with a notable increase in free fatty acids compared to Cluster 5, suggesting involvement in fatty acid and steroid biosynthesis pathways (Fig. 5A). These exploratory findings provide a focused subset of lipids for further investigation into the metabolic mechanisms underlying cardiac vulnerability in high-altitude environments.

Fig. 5.

Specific differential lipid analysis of the high-altitude migrants with myocardial injury. A K-means clustering analysis of all plasma lipids detected from the matched population (left). Metabolite set enrichment of lipid clusters from K-means clustering analysis (right). B Counts of down-regulated lipid subclasses in Cluster 5. C Counts of up-regulated lipid subclasses in Cluster 6. D Contents of differential lipid subclass in Cluster 5 and 6. E Bubble plot of fatty acyl chain length and unsaturation in down-regulated lipids in Cluster 5. F Bubble plot of fatty acyl chain length and unsaturation in up-regulated lipids in Cluster 6. G Number of fatty acid chains identified in the differential lipids by lipid class (bar color). H Log₂ fold changes of significantly altered fatty acid chains in HCA vs. HCN by lipid class (dot color). Each dot represents 1 fatty acid chain. I Diagram of lipid metabolism changes between the HCA group and the HCN group. HCA vs. HCN: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001

Comparative analysis of lipids between the HCN and HCA groups in Clusters 5 and 6 revealed distinct alterations. In Cluster 5, we identified 54 significantly downregulated lipids (Fig. 5B), predominantly sphingolipids, including 16 hexosylceramides (HexCer), 11 Cers, 2 SMs, and 1 CerP. Cluster 6 exhibited 21 upregulated lipids (Fig. 5C), primarily GLs, consisting of 7 TGs, 2 DGs, and 1 MGs, along with fatty acyls (2 acylcarnitines (CARs) and 3 FFAs). Subclass analysis of these lipids showed significant changes in the HCA group compared to HCN (Fig. 5D). The most pronounced decreases were observed in Cer (P < 0.01) and HexCer (P < 0.01), while TG (P < 0.01) and FFA (P < 0.05) showed the most marked increases. Downregulated lipids were enriched in C18:0, C18:1, and C18:2 fatty acyl chains (Fig. 5E), mainly contributed to GP (Fig. 5G–H). In contrast, upregulated lipids were predominantly composed of C16:0 and C18:1 fatty acyl (Fig. 5F), mainly contributed to GL (Fig. 5G–H). These findings suggest selective modulation of specific lipid species in myocardial injury at high altitude, with sphingolipid depletion and glycerolipid/fatty acyl accumulation representing key metabolic signatures.

Correlation analysis between specific differential lipids and myocardial injury

Figures 6A and B display the top 20 differential lipids ranked by VIP scores in Clusters 5 and 6, respectively. Compared with the HCN group, the most significantly downregulated lipids in the HCA group were PS(20:4_20:4), PE(18:2_20:2), Cer(t18:1/15:1), etc., while the most markedly upregulated lipids included DG(16:0_20:4), MG(18:1), CAR C17:0, etc. The clustered correlation network heatmap results (Fig. 6C) revealed strong correlations in expression levels among differential lipids within the same subcategories, such as Cer, HexCer, FFA, CAR, TG, DG, and lysophosphatidylinositols (LPI). At the first-class level, SP and GL exhibited strong intra-class correlations, while a pronounced negative correlation was observed between SP and GL lipid levels. Among these differential lipids, twelve downregulated lipids (e.g., HexCer(t18:2/20:0(2OH)), HexCer(t22:1/16:1(2OH)), and Cer(t14:1/24:1(2OH))) and thirteen upregulated lipids (e.g., TG(14:0_16:0_20:4), TG(15:0_16:0_18:1), and TG(16:0_16:1_17:1)) showed significant correlations with myocardial injury markers. After adjusting for glucose, age, and BMI, multiple linear regression revealed that PS(20:4_20:4) was significantly associated with lower CKMB levels (β = −0.993, 95% CI: −1.609 to −0.378, FDR-adjusted P < 0.05), while DG(16:0_20:4) was associated with higher CKMB levels (β = 1.011, 95% CI: 0.118 to 1.905, P < 0.05, FDR-adjusted P = 0.058) (Fig. 6D, Supplementary Table S3). Logistic regression confirmed these findings, showing that PS(20:4_20:4) was associated with reduced risk of myocardial injury (OR = 0.000, 95% CI: 0 to 0.017, FDR-adjusted P < 0.05), whereas DG(16:0_20:4) was associated with increased risk (OR = 6054.273, 95% CI: 11.708 to 3.5 × 10⁷, FDR-adjusted P < 0.05) (Supplementary Table S3).

Fig. 6.

Analysis of specific differential lipids for assessing myocardial susceptibility at high altitude. A Top 20 specific lipids differentially expressed in Cluster 5. B Top 20 specific lipids differentially expressed in Cluster 6. C Correlation network heatmap between cardiac injury biomarkers and specific differentially expressed lipids based on the Mantel test. Lipids highlighted in red indicate a significant correlation with cardiac injury biomarkers. D Forest plot of linear regression coefficients (β) and 95% CIs for associations between specific differentially expressed lipids and log-transformed CKMB after adjusting for glucose, age, and BMI. E ROC curves for up-regulated lipids with significant correlations. F ROC curves for down-regulated lipids with significant correlations. G–H Bootstrap validation of ROC analysis for DG(16:0_20:4) and PS(20:4_20:4), respectively. I–J Relative content analysis of anti-inflammatory lipid mediators, lipoxin A₄ and 15-oxoETE, across groups. *, P < 0.05; **, P < 0.01; ***, P < 0.001 versus the PLAIN group

ROC curve analysis revealed that PS(20:4_20:4) (AUC = 0.831, bootstrap 95% CI: 0.67 to 0.95, FDR-adjusted P < 0.05) and DG(16:0_20:4) (AUC = 0.822, bootstrap 95% CI: 0.67 to 0.96, FDR-adjusted P < 0.05) exhibited relatively favorable predictive performance for high-altitude myocardial injury among downregulated and upregulated lipids, respectively (Fig. 6E–H, Supplementary Table S4). Cross-validation and calibration analyses supported the robustness of these findings, with DG(16:0_20:4) demonstrating good stability (selected in 66.7% of LOOCV folds) and adequate calibration (Hosmer–Lemeshow P = 0.235, Supplementary Table S4). Since DG(16:0_20:4) is a classic pro-inflammatory messenger, we analyzed the levels of anti-inflammatory lipids, 15-oxoETE and lipoxin A₄. As shown in Fig. 6I and J, both were significantly downregulated in the HCN group and also exhibited a decreasing trend in the HCA group, suggesting a potential shift toward a pro-inflammatory lipid milieu.

Moreover, DG(16:0_20:4) had a significantly positive relationship with circulating blood lymphocyte counts, while PS(20:4_20:4) had a significantly negative relationship with white blood cells in the circulating blood (Fig. 7A), indicating potential associations between these lipids and immune cell populations. To verify the potential relationship between lipids and the immune system, robust regression was used to test the correlations between immune traits and CKMB and TG in 483 individuals. Both CKMB and TG showed a positive relationship with lymphocyte and white blood cells (Fig. 7B and C). Collectively, these correlational findings suggest that decreased levels of polyunsaturated fatty acid (PUFA)-containing PS and other membrane lipids, along with a shift in lipid mediators toward a pro-inflammatory phenotype, may contribute to the lipid-immune crosstalk associated with myocardial injury among long-term high-altitude migrants. Further mechanistic studies are warranted to establish causality.

Fig. 7.

Associations among inflammatory and immune traits, specific differential lipids, and myocardial injury traits. A Associations between immune traits and the specific differential lipids tested by Spearman analysis (N = 45). B Associations between immune traits and CKMB tested by robust regression analysis. C Associations between immune traits and plasma TG tested by robust regression analysis. Robust regression analysis was performed, adjusting for age, BMI, and residence duration. The β-coefficients with error bars representing 95% CI were shown separately for significance (blue or yellow) or not (grey). WBC: white blood cell; Lymph: lymphocyte; Gran: granulocyte; NLR: neutrophil-to-lymphocyte ratio. N = 483. *, P < 0.05 FDR, **, P < 0.01 FDR

Discussion

In this study, through a comprehensive analysis on a large-scale sample group of 525 high-altitude immigrants and 98 plain controls, we identified a significant correlation between myocardial injury and hyperlipidemia within the high-altitude immigrant population. We further performed non-targeted lipidomics to investigate the plasma lipid profiles of the matching population and revealed characteristic lipid alterations in high-altitude migrants, particularly those with high-altitude–induced myocardial injury. We found that chronic high-altitude exposure led to significant changes in three major classes of lipid metabolites: an increase in glycerides, a decrease in sphingolipids, and complex alterations in glycerophospholipids. Among the lipids specifically dysregulated in the myocardial injury group, the same pattern was observed, accompanied by a reduction in beneficial glycerophospholipids and an increase in harmful ones. Furthermore, we identified PS(20:4_20:4) and DG(16:0_20:4) as promising candidate biomarkers that were significantly altered in the myocardial injury group and strongly correlated with myocardial injury indicators. However, given the exploratory nature of this study and the limited sample size, these findings should be considered preliminary. Further validation in larger independent cohorts is warranted before considering clinical application of these lipid biomarkers for identifying susceptibility to myocardial injury among high-altitude migrants.

Energy metabolism exhibits remarkable plasticity in high-altitude environments, with both altitude and exposure duration serving as critical modulators. Previous studies in humans and mice have reported enhanced glutamine and fatty acid metabolism alongside attenuated glycolysis following short-term high-altitude hypoxia exposure [23, 27]. In contrast, under chronic hypoxic conditions, metabolic reprogramming typically occurs, characterized by an energy metabolism shift from the efficient ‘fatty acid oxidation’ to ‘glucose oxidation’ and ‘glycolysis,’ the less efficient yet life-preserving strategy [30, 31]. Therefore, a high level of circulating TG was commonly observed in humans under chronic high-altitude exposure [32–34]. Consistently, our findings reveal a marked accumulation of storage lipids (glycerides) and fuel lipids (FFAs) in the plasma of high-altitude migrants, which may reflect a reduction in fatty acid oxidation. The buildup of glycerides and FFAs may represent an adaptive metabolic compromise in response to energy crisis, oxidative stress, and hypoxia-inducible factor signaling activation. It has been reported that chronic hypoxia prompted organ-specific fuel rewiring, and the heart stands alone as an increased glucose oxidizer, shuttling carbons from circulating glucose into the mitochondria to feed the TCA cycle [35]. The specific adjustment of energy utilization by the heart under hypoxic conditions may result in the higher levels of triglycerides and FFA that we observed in high-altitude migrants with myocardial injury. Additionally, in the HCA group, we observed elevated levels of intermediate products of fatty acids, such as CARs (C17:0 and C24:0), which reflected the CAR profile in cardiac tissues [36], raising the possibility that the heart was confronting a significant toxic crisis resulting from lipid intermediates. Excessive accumulation of glycerides, free fatty acids, and their intermediate metabolites can impair vascular endothelial function [37], promote inflammatory responses [38], and elevate the risk of cardiovascular diseases [39]. Additionally, ectopic lipid deposition in non-adipose tissues (e.g., the heart) may induce cellular dysfunction, inflammation, and insulin resistance, ultimately creating a vicious cycle [40]. Collectively, these mechanisms may constitute a key pathological basis for high-altitude myocardial injury, though direct evidence remains to be established.

In addition to storage and fuel lipids, other lipid species, such as sphingolipids and phospholipids, also underwent significant alterations under high-altitude conditions. These lipids not only serve as fundamental structural constituents of cellular membranes but also act as signaling molecules. Sphingolipids, in particular, are widely recognized as essential components of lipid rafts in cell membranes [41]. However, the understanding of cardiac sphingolipid metabolism under high-altitude conditions remains limited. Our findings indicate that levels of CerP and SM are significantly reduced in high-altitude migrants. CerP, a key regulator of cell growth, survival, migration, and immune response, has been shown to mitigate mitochondrial damage in mouse models of high-altitude pulmonary edema through exogenous supplementation, which thereby helps maintain mitochondrial homeostasis and alleviate oxidative stress [42]. SM, the most abundant sphingolipid in humans [43], has been implicated in a recent clinical study where its metabolism was strongly associated with dilated right heart pulmonary hypertension [44]. In cardiac endothelial cells, the presence of neutral sphingomyelinase, an enzyme involved in SM-metabolism, leads to the production of metabolites that stimulate endogenous nitric oxide (NO) synthesis, thereby promoting peripheral vasodilation [45]. The observed reduction in CerP and SM under high-altitude conditions raises the hypothesis that hypobaric hypoxia may suppress the synthesis of these cytoprotective sphingolipids.

Glycerophospholipids constitute the most abundant subclass of lipids [46]. In this study, we observed that under high-altitude conditions, glycerophospholipids undergo complex bidirectional alterations, while LPA and PA consistently exhibit a decreasing trend. PA, primarily generated via the hydrolysis of phosphatidylcholine, serves as a key precursor for the biosynthesis of various phospholipids and acts as a crucial signaling molecule within mitochondrial and other cellular membranes [47]. In the heart, PA contributes to increased intracellular free calcium concentration in adult cardiomyocytes, thereby enhancing cardiac contractility [48, 49]. Furthermore, PA promotes protein synthesis in cardiomyocytes through the activation of phospholipase C and protein kinase C [50, 51]. Consequently, a reduction in PA may compromise membrane stability and functionality, potentially elevating the risk of impaired cardiac diastolic function. LPA exhibits a dual role in ischemic and hypoxic cardiovascular diseases. On one hand, LPA and its receptors have been shown to improve cardiac function, attenuate fibrosis and ventricular remodeling after myocardial infarction [52, 53]. On the other hand, LPA serves as a potent pro-inflammatory and pro-proliferative signaling molecule, directly stimulating vascular smooth muscle cell proliferation and endothelial inflammatory activation [54], and promoting pathological cardiac hypertrophy via activation of cardiac endothelial cells and fibroblasts [55]. Notably, studies using chronically hypobaric hypoxia-exposed rat models have revealed LPA accumulation in the intestine and lungs, where it strongly correlates with oxidative stress and inflammatory responses [56]. Additionally, in acute hypobaric hypoxia mouse models, LPA has been shown to promote cardiac inflammation and ferroptosis via the activation of the RhoA/ROCK pathway [57]. In contrast, our findings indicate that high-altitude populations exhibit low LPA expression, which may represent an adaptive metabolic mechanism to mitigate inflammatory responses and avoid potential pathological effects of LPA under chronic hypoxic conditions.

Although overall lipid class changes were consistent between individuals with high-altitude myocardial injury and general high-altitude migrants, notable differences were observed at the subtype level, particularly among sphingolipids and phospholipids. Compared with the plain group, only CerP and SM were significantly decreased in the HC group. In contrast, relative to both the plain and HCN groups, the HCA group exhibited a broad reduction across sphingolipid subclasses, with the most pronounced decreases observed in HexCer and Cer. In lowland populations, Cers are recognized as key lipotoxic mediators in metabolic dysregulation [58]. Clinically, serum Cer levels serve as prognostic indicators for major adverse cardiovascular events due to their pleiotropic effects, including inhibition of nitric oxide synthase, reduction of insulin sensitivity, disruption of mitochondrial bioenergetics, and promotion of apoptosis and fibrosis [59]. HexCer, formed by the glycosylation of Cer, is an essential component of cell membranes, with elevated blood levels strongly associated with cardiovascular risk [60]. Moreover, specific HexCer and Cer species containing C16:0, C18:0, and C24:1 acyl chain have demonstrated significant predictive value for major cardiovascular events [61, 62]. Importantly, these associations in lowland populations differ from our observations under the high-altitude condition, where sphingolipids were markedly reduced rather than elevated. This distinction underscores the unique metabolic adaptation to chronic hypoxia compared with classic cardiometabolic disease. Under high-altitude hypoxic conditions, cardiomyocytes experience severe ATP deficiency. The marked reduction in sphingolipids, including Cer and HexCer, may reflect substrate depletion or a significant inhibition of the enzymatic activities involved in their synthesis. Notably, correlation analysis revealed a significant negative association between glycerolipid subtypes (TG, DG, MG) and sphingolipid subtypes (Cer, HexCer). Since palmitate derived from glycerolipid hydrolysis serves as a key substrate for Cer synthesis, we speculate that inhibition of glycerolipid catabolic enzymes could lead to substrate scarcity for sphingolipid synthesis. Disruption of the sphingolipid metabolic network may directly exacerbate myocardial injury through multiple mechanisms, including heightened energy exhaustion, dysregulation of apoptotic balance, and compromised electrophysiological stability.

In individuals with high-altitude myocardial injury, we observed a distinctive pattern of phospholipid alterations: PS, PA, PI, phosphatidylglycerol (PG), and LPE were significantly downregulated, whereas LPI and LPC were markedly upregulated. Numerous in vitro and in vivo studies have demonstrated the cardioprotective and anti-inflammatory effects of PS [63, 64]. PI is involved in ischemic preconditioning signaling and activates most cardioprotective pathways [65]. Moreover, LPE, typically reduced in the serum of patients with chronic heart failure and acute coronary syndrome, ameliorates aging-related diastolic dysfunction in mice by improving mitochondrial structure and function when supplemented [66–68]. PG exhibits anti-inflammatory activity [69] and serves as a downstream metabolite of PA and a precursor of cardiolipin; its diminished levels may undermine mitochondrial structural integrity [47, 70]. In contrast, LPI may impair cardiac contractility by promoting calcium release from the sarcoplasmic reticulum and aggravate ischemia–reperfusion injury via the GPR55/ROCK/p38 MAPK signaling pathway [71, 72]. Similarly, LPC, a recognized biomarker for cardiovascular diseases such as myocardial infarction and atherosclerosis [73], may exacerbate pulmonary edema under high-altitude conditions through pro-inflammatory mechanisms [74]. In summary, this study reveals an imbalanced lipid profile in high-altitude myocardial injury characterized by reduced levels of beneficial phospholipids and elevated levels of detrimental species. This imbalance may synergistically contribute to the development and progression of myocardial injury, though the precise mechanisms underlying these changes warrant further investigation.

Notably, PS(20:4_20:4) and DG(16:0_20:4), the most significantly downregulated and upregulated specific lipids in the HCA group, respectively, share a common fatty acyl chain, arachidonic acid (AA). PS containing AA plays a crucial role in maintaining membrane stability and fluidity [75]. The loss of PS, along with lipid raft components such as HexCer, substantially impairs membrane integrity and signal transduction [76]. Under these conditions, phospholipase C (PLC) signaling on the plasma membrane may become hyperactivated in response to stimuli, resulting in the abundant production of sn-2 AA-containing DG hydrolyzed from phosphatidylinositol 4,5-bisphosphate (PIP2) [77, 78]. The accumulation of DG(16:0_20:4) can directly activate protein kinase C (PKC), thereby disrupting cell proliferation, differentiation, and survival, while also promoting inflammatory responses and oxidative stress [79–81]. The impact of hypoxia on the inflammatory response has been extensively documented in high-altitude studies [82]. Our findings further indicate a significant reduction in pro-resolving lipid mediators within high-altitude environments, implying the involvement of maladaptive inflammatory pathways. As expected, DG(16:0_20:4) exhibited a positive correlation with immune traits, whereas PS(20:4_20:4) demonstrated a negative correlation. Notably, plasma levels of CKMB and TG were significantly and positively correlated with blood lymphocyte counts. Previous research has positioned peripheral blood lymphocytes as biosensors for whole-body responses to high-altitude exposure [83]. Based on these correlational findings, we propose the hypothesis that elevated levels of specific pro-inflammatory lipids may be involved in myocardial injury through potential interactions with immune systems under prolonged high-altitude conditions. However, the precise causal relationships among plasma glycerides, lymphocytes, and high-altitude myocardial damage warrant further clarification.

Overall, our study offers novel insights into the association between susceptibility to high-altitude myocardial injury and alterations in lipid metabolism, with an emphasis on population-specific lipid variations. However, several limitations should be acknowledged. First, all participants were male, which substantially limits the generalizability of our findings to female populations. Future studies including both sexes are warranted. Second, the lipidomics discovery cohort was small (N = 15 per group); these findings are exploratory and require validation in larger independent cohorts. Third, lipidomic profiling was performed on plasma samples, which may not fully reflect cardiac tissue metabolism; tissue-level validation is needed. Fourth, this study employed a cross-sectional design, which precludes causal inference; longitudinal studies are required to establish temporal relationships. Fifth, despite adjustment for key covariates (age, BMI, and glucose), residual confounding from unmeasured variables (e.g., physical activity and genetic factors) cannot be excluded. Additionally, detailed long-term medication history was not collected, though participants were young and free from chronic diseases. Finally, while we performed cross-validation to assess model stability, the absence of an independent external validation cohort limits generalizability. These findings should therefore be interpreted as exploratory, and further validation is required before clinical application.

Conclusion

This study identified potential risk factors of HAHD that differ from those of cardiovascular disease in plain and revealed distinct lipid remodeling in long-term high-altitude migrants characterized by the accumulation of storage and fuel lipids alongside reductions in specific sphingolipids and phospholipids. Notably, individuals with cardiac dysfunction exhibited broader sphingolipid downregulation, accompanied by a decrease in cardioprotective phospholipids and an increase in pro-inflammatory mediators. PS (20:4_20:4) and DG (16:0_20:4) were identified as candidate biomarkers associated with high-altitude cardiac dysfunction. Overall, this study provides insights into lipid metabolic adaptations to prolonged high-altitude exposure and highlights potential avenues for early risk assessment of HAHD, offering valuable leads for further research in large population cohorts.

Abbreviations

AA

Arachidonic acid

APOA1

Apoprotein A-1

APOB

Apoprotein B

AST

Aspartate aminotransferase

BMI

Body mass index

CAR

Acylcarnitine

Cer

Ceramide

CerP

Ceramide phosphate

CHO

Total cholesterol

CKMB

Creatine kinase isoenzyme

DG

Diacylglycerol

ECG

Electrocardiogram

ECHO

Echocardiography

EF

Ejection fraction

FA

Fatty acid

FDR

False discovery rate

FFA

Free fatty acid

FS

Fractional shortening

GL

Glycerolipid

GP

Glycerophospholipid

Gran

Granulocyte

HAHD

High-altitude heart disease

HDL

High-density lipoprotein

HexCer

Hexosylceramide

LDH

Lactate dehydrogenase

LDL

Low-density lipoprotein

LPA

Lysophosphatidic acid

LPC

Lysophosphatidylcholine

LPE

Lysophosphatidylethanolamine

LPG

Lysophosphatidylglycerol

LOOCV

Leave-one-out cross-validation

Lymph

Lymphocyte

MG

Monoacylglycerol

NLR

Neutrophil-to-lymphocyte ratio

PA

Phosphatidic acid

PC

Phosphatidylcholine

PE

Phosphatidylethanolamine

PG

Phosphatidylglycerol

PI

Phosphatidylinositol

PLS-DA

Partial least squares-discriminant analysis

PS

Phosphatidylserine

PUFA

Polyunsaturated fatty acid

RBBB

Right bundle branch block

ROC

Receiver operating characteristic

SM

Sphingomyelin

SP

Sphingolipid

ST

Sterol

TG

Triglyceride

VIP

Variable importance in projection

α-HBDH

α-Hydroxybutyrate dehydrogenase

WBC

White blood cell

Authors’ contributions

H.D.: Conceptualization, Methodology, Data curation, Formal analysis, Writing—original draft. G.L.: Methodology, Data curation, Formal analysis. P.S.: Conceptualization, Methodology, Data curation, Formal analysis. Z.Y.: Writing—original draft & revising. L.Z., Z.B., P.Z., and Y.Z.: Material preparation, data collection. N.W.: Project administration. Y.G.: Resources, Project administration. W.Z.: Conceptualization, Writing—review & editing, Funding acquisition. All authors read and approved the submitted version.

Funding

This study was supported by the National Key R&D Program of China (No. 2025YFC3507500, W.Z.).

Data availability

The datasets generated in this study are included within the article and supplementary files. Other data supporting the findings of this study are available from the corresponding authors upon request.

Declarations

Competing interests

The authors declare no competing interests.