Increased risk of the onset and exacerbation of COPD in metabolic syndrome patients: A 2-sample Mendelian randomization study Authors: Ling Xu, Jiantong Sun, Lan Yang, Yaxin Chen, Haoyu Wang, Lei Li Published: 2026 ABSTRACT This study aimed to assess the genetic associations of metabolic syndrome (MetS) and its components with the onset and exacerbation of chronic obstructive pulmonary disease (COPD), including respiratory failure and pneumonia. Summary statistics were sourced from publicly available genome-wide association studies. Genetic instruments were selected at genome-wide significance and linkage disequilibrium clumped via PLINK software. Instrument strength was confirmed using F-statistics. Mendelian randomization (MR) analyses were performed using inverse-variance weighted, maximum likelihood, MR-Egger, and median-based approaches. Multivariable MR was performed to determine the independent effects of MetS and waist circumference (WC). Sensitivity analysis was performed to validate the robustness of the results. Statistical power was evaluated a priori using the mRnd. We controlled for multiple comparisons using the Benjamini–Hochberg false discovery rate procedure and reported the q-values. Using the inverse variance weighted method, MetS was associated with an increased risk of COPD onset (odds ratio [OR] = 1.27, 95% confidence interval [CI]: 1.14–1.40), respiratory insufficiency (OR = 1.66, 95% CI: 1.34–2.06), and pneumonia (OR = 1.13, 95% CI: 1.08–1.19). Among components, WC exhibited the strongest associations with COPD (OR = 1.69, 95% CI: 1.54–1.84), respiratory insufficiency (OR = 2.27, 95% CI: 1.89–2.72), and pneumonia (OR = 1.20, 95% CI: 1.16–1.25). Hypertension showed a nominal association with pneumonia (OR = 1.10, 95% CI: 1.00–1.20, P = .040, q = 0.103), which did not remain significant after false discovery rate correction. The other components showed no causal associations with COPD. Our study provides genetic evidence consistent with the causal contribution of MetS to COPD onset and exacerbation, with central adiposity (WC) emerging as the dominant driver. This finding supports closer respiratory risk assessment and prevention efforts among patients with MetS. Patients with comorbid COPD and MetS require personalized multidisciplinary strategies to prevent COPD exacerbation and slow progression. 1. Introduction Chronic obstructive pulmonary disease (COPD), a progressive respiratory disorder characterized by persistent airflow limitation and chronic airway inflammation, is the leading cause of death attributable to chronic respiratory diseases, accounting for more than 3.23 million annual deaths worldwide.[1,2] Notably, the global prevalence of COPD is projected to reach approximately 600 million cases by 2050.[3] Approximately 20.1% of patients experience severe acute exacerbations of COPD annually, whereas nearly 48.4% encounter at least 1 severe exacerbation throughout a 5-year period.[4] However, the etiology and pathogenesis of COPD are only partially understood. Metabolic syndrome (MetS) is a cluster of metabolic abnormalities characterized by the concomitant presence of central obesity (measured by waist circumference [WC]), hyperglycemia, hyperlipidemia (including hypertriglyceridemia and low high-density lipoprotein [HDL] cholesterol levels), and hypertension.[5] Notably, the prevalence of MetS is markedly higher among COPD patients (21–62%) than in the general population (12.5–31.4%),[6,7] and observational studies have linked MetS to accelerated lung function decline and increased COPD risk.[8–10] However, these observational studies may have been hampered by unobserved confounding variables, limited follow-up periods, and potential reverse causation. Therefore, further work is needed to fully elucidate the causal involvement of MetS and its components in the genesis and worsening of COPD. Mendelian randomization (MR) is a rigorous analytical method that employs genetic variants linked to exposures to evaluate causal links by utilizing the inherent randomization of genetic diversity at conception to reduce confounding biases.[11] In this study, we performed MR analyses using the most extensive and recent genome-wide association study (GWAS) data to assess the causal effect of MetS on COPD. 2. Materials and methods 2.1. Study design This 2-sample MR study aimed to investigate the causal effects of MetS and its components (WC, hypertension, fasting blood glucose [FBG], triglycerides, and HDL cholesterol) on the onset of COPD, as well as on the exacerbation events in COPD patients, including COPD/asthma/interstitial lung disease (ILD)-related pneumonia and COPD-related respiratory insufficiency. The analysis conformed to 3 fundamental MR assumptions[12]: genetic instrumental variables (IVs) must demonstrate a robust correlation with exposure (association hypothesis); IVs must be independent of confounders that influence exposure-outcome relationships (independence hypothesis); and IVs must affect outcomes exclusively through exposure, without influencing them via alternative pathways (exclusivity hypothesis) (Fig. 1). This report adheres to the strengthening the reporting of observational studies in epidemiology using Mendelian randomization guidelines. Overview of the MR study design (created with BioRender; Yang, L. [2025] https://BioRender.com/tfs3l70). HDL = high-density lipoprotein, ILD = interstitial lung disease, IVW = inverse variance weighted, MR = Mendelian randomization, MR-PRESSO = Mendelian randomization pleiotropy residual sum and outlier, SNP = single nucleotide polymorphism. 2.2. Data sources The datasets included in this study were sourced from the most recent and largest available GWAS sources. Summary statistics for MetS were obtained from the Complex Trait Genetics Laboratory database, which includes 461,902 individuals of European descent. The MetS phenotype in this cohort was defined based on the National Cholesterol Education Program Adult Treatment Panel III criteria, which required the presence of at least 3 of the 5 standard components: abdominal obesity, elevated triglycerides, reduced HDL cholesterol, hypertension, and hyperglycemia. Data for WC, FBG, triglycerides, and HDL cholesterol were obtained from the Medical Research Council Integrative Epidemiology Unit United Kingdom Biobank GWAS Pipeline. Data on hypertension, comprising 129,909 cases and 354,689 controls, were obtained from the GWAS Catalog database (detailed information of the included datasets is shown in Table 1). Table 1GWAS summary datasets used in this study.Exposure/outcomeGWAS IDYearSample size(Case/ Control)PopulationWeb sourcesMetSCF_MetS_results2022461,920European https://cncr.nl/research/summary_statistics/ WCukb-b-94052018462,166European https://gwas.mrcieu.ac.uk/datasets/ukb-b-9405/ HypertensionGCST900386042021129,909/ 354,689European https://www.ebi.ac.uk/gwas/studies/GCST90038604 FBGebi-a-GCST005186201258,074European https://gwas.mrcieu.ac.uk/datasets/ebi-a-GCST005186/ Triglyceridesieu-b-1112020441,016European https://gwas.mrcieu.ac.uk/datasets/ieu-b-111/ HDL cholesterolieu-b-1092020403,943European https://gwas.mrcieu.ac.uk/datasets/ieu-b-109/ COPDfinngen_R11_J10_COPD202321,617/ 372,627European https://storage.googleapis.com/finngen-public-data-r11/summary_stats/finngen_R11_J10_COPD.gz COPD-related respiratory insufficiencyfinngen_R11_COPD_INSUFFICIENCY20233996/ 433,110European https://storage.googleapis.com/finngen-public-data-r11/summary_stats/finngen_R11_COPD_INSUFFICIENCY.gz COPD/asthma/ILD related pneumoniafinngen_R11_PULM_INFECTIONS2023144,808/ 308,925European https://storage.googleapis.com/finngen-public-data-r11/summary_stats/finngen_R11_PULM_INFECTIONS.gz COPD = chronic obstructive pulmonary disease, FBG = fasting blood glucose, GWAS = genome-wide association study, HDL = high-density lipoprotein, ILD = interstitial lung disease, MetS = Metabolic syndrome, WC = Waist circumference. GWAS summary datasets used in this study. COPD = chronic obstructive pulmonary disease, FBG = fasting blood glucose, GWAS = genome-wide association study, HDL = high-density lipoprotein, ILD = interstitial lung disease, MetS = Metabolic syndrome, WC = Waist circumference. COPD-related phenotypic data were sourced from the FinnGen Consortium (https://www.finngen.fi/fi, Release R11, accessed March 2, 2025). The summary data for COPD included 21,617 cases and 372,627 controls. Disease exacerbation outcomes were assessed using 2 distinct endpoints: COPD-related respiratory insufficiency (3996 cases/433,110 controls) and pneumonia associated with COPD, asthma, or ILD (144,808 cases/308,925 controls). No sample overlap was observed between the exposure and outcome cohorts. No information that could identify individual participants during or after data collection was accessed. 2.3. Instrumental variable selection The genetic instruments were subjected to a rigorous multi-step quality-control process. Initially, we identified genome-wide significant variants (P < 5 × 10-8) associated with exposure. To ensure genetic independence, variants were clumped using the PLINK v1.9 algorithm to remove linkage disequilibrium (R2 < 0.001, window size = 10,000 kb). Palindromic single nucleotide polymorphisms (SNPs) with intermediate allele frequencies were excluded to prevent strand-alignment errors. The statistical strength of the IVs was assessed using F-statistics, with all retained instruments exceeding 10 to minimize instrument bias. For each exposure, the variance explained (R2) was derived from allele frequency, effect size, and standard error using standard formulae; per-exposure instrument strength was summarized using F = (R2/(1 − R2)) × ((N − K − 1)/K) (N: sample size; K: instrument count). Furthermore, a harmonization process was implemented to standardize SNP orientations and to remove palindromic SNPs exhibiting ambiguous allele frequency matching from the dataset. In addition, variants that did not reach genome-wide significance or exhibited strong linkage disequilibrium were excluded to preserve instrument validity and independence under the MR assumptions. 2.4. MR analysis and sensitivity analysis We used the inverse variance weighted (IVW) method as the primary method to calculate causal effects and optimize efficiency under the balanced pleiotropy assumption. We employed a multimethod framework to enhance analytical robustness and confirm pleiotropy assumptions. The maximum likelihood (ML) approach enhances the precision of causal inference by simultaneously modeling the genetic instrument effects and limiting pleiotropic variance, thus providing reliable impact estimates when the conditions for valid IVs are met. The MR-Egger regression yielded unbiased causal estimates in the presence of directional pleiotropy by integrating an intercept test. Additionally, the simple median technique offered robust causal inference, maintaining reliability even when up to 50% of the instruments were faulty, thereby demonstrating increased resilience to outliers. To address the potential genetic overlap between MetS and abdominal obesity, we performed Multivariable Mendelian Randomization (MVMR). Genetic instruments for MetS and WC were pooled to ensure independence. We extracted summary statistics for both exposures and outcomes to construct a multivariate model. We calculated Cochran Q test statistic to test for possible heterogeneity and identified pleiotropic outliers using MR-PRESSO global testing and RadialMR radial regression. If outliers were identified, they were removed and the MR analysis was repeated until no outliers were present. Leave-one-out analyses were performed to evaluate the individual SNP influence by repeating the IVW analysis by dropping each SNP associated with exposure in turn. To account for multiple comparisons across the 18 primary IVW exposure-outcome tests, we controlled the false discovery rate using the Benjamini-Hochberg procedure. All analyses were conducted in R version 4.3.3 using the packages TwoSampleMR and RadialMR. 2.5. Power calculation Statistical power for the MR analyses was estimated using the mRnd online tool (https://shiny.cnsgenomics.com/mRnd/).[13] Power calculations were conducted with a type I error rate of 0.05, encompassing essential parameters such as sample size, proportion of cases derived from the original GWAS study, R2 of IVs, and odds ratio (OR) generated from the IVW method. Results beyond 0.8 were deemed strong. 3. Ethical approval The present analysis exclusively utilized anonymized and published data that had already been approved by their respective institutional review boards. Consequently, no additional ethical approval or consent to participate was obtained. 4. Results A detailed summary of the data sources, including GWAS ID, sample characteristics, and research populations, is presented in Table 1. The final count of IVs per phenotype following thorough screening varied between 22 and 320. The details of the selected IVs are presented in Tables S1–S18, which include SNP identifiers, chromosome positions, effect/other alleles, beta coefficients, standard errors, P values, F-statistics, and R2. Statistical power is listed in Table S19. Visualization of scatter plots, funnel plots, and leave-one-out analyses is shown in Figure S1–S54. 4.1. Causal effects of metabolic syndrome on COPD onset and exacerbation The IVW analysis demonstrated a significant causal association between MetS and increased COPD risk (OR = 1.27, 95% confidence interval [CI]: 1.14–1.40, P = 7.07 × 10 − 6, q = 2.12 × 10 − 5), with consistent findings from ML method (OR = 1.27, 95% CI: 1.15–1.41, P = 5.88 × 10 − 6) and simple median approaches (OR = 1.24, 95% CI: 1.05–1.47, P = .01) (Fig. 2). The MR-Egger regression showed a nonsignificant positive trend (OR = 1.12, 95% CI: 0.84–1.48, P = .441, Figure 2). Statistical power was 100% (Table S19). Sensitivity analyses confirmed the robustness of the causal estimates. Heterogeneity tests using Cochran Q revealed no significant heterogeneity (Cochran Q, P = .58; Table S20). The absence of directional pleiotropy was supported by nonsignificant MR-Egger intercepts (MR-Egger intercept, P = .35; Table S20). MR-PRESSO global testing yielded a P value of .59, indicating no evidence of influential outliers (Table S20). Funnel plots demonstrated symmetry, ruling out directional bias (Figure S2), and leave-one-out sensitivity analyses confirmed that no single SNP had a substantial influence on the final causal estimate (Figure S3). Mendelian randomization results of causal effects of metabolic syndrome on COPD and related outcomes. CI = confidence interval, COPD = chronic obstructive pulmonary disease, IVW = inverse variance weighted, OR = odds ratio. The causal relationship between MetS and the risk of COPD exacerbation was further investigated. Notably, using IVW estimation, MetS showed the strongest associations with COPD-related respiratory insufficiency (OR = 1.66, 95% CI: 1.34–2.06; P = 3.27 × 10-6, q = 1.18 × 10 − 5). Robustness was supported by consistent estimates from ML (OR = 1.68, P = 2.72 × 10 − 6), MR-Egger (OR = 1.95, P = .03), and simple median analysis (OR = 1.53, P = .006) (Fig. 2). In addition, there was a positive causal relationship between MetS and COPD/asthma/ILD-associated pneumonia (IVW OR = 1.13, 95% CI: 1.08–1.19; P = 1.20 × 10 − 7, q = 5.40 × 10 − 7) (Fig. 2). The statistical power of these results reached 100% (Table S19). No significant heterogeneity or horizontal pleiotropy was observed (Table S20). Funnel plot symmetry further supported the absence of bias (Figure S5 and S8), and leave-one-out validation indicated that the causal effects were not influenced by any single SNP (Figure S6 and S9). 4.2. Causal effects of metabolic syndrome components on COPD onset and exacerbation Using the IVW method, central obesity (WC increase) exhibited a strong causal association with the risks for COPD (OR = 1.69, 95% CI: 1.54–1.84, P = 6.11 × 10-31, q = 1.10 × 10 − 29) and pneumonia (OR = 1.20, 95% CI: 1.16–1.25, P = 7.23 × 10-21, q = 6.51 × 10 − 20), with particularly pronounced effects on respiratory insufficiency (OR = 2.27, 95% CI: 1.89–2.72, P = 1.21 × 10 − 18, q = 7.26 × 10 − 18, Tables S21). The ML, MR-Egger, and simple median methods yielded consistent results, supporting the causal effect of central obesity on the onset and exacerbation of COPD (Table S21), with the statistical power reaching 100% (Table S19). Hypertension showed a nominal association with pneumonia that did not remain significant after false discovery rate (FDR) correction (OR = 1.10, 95% CI: 1.00–1.20, P = .040, q = 0.103) (Tables S21). No causal relationship was found between FBG, triglyceride, or HDL cholesterol levels and the onset or exacerbation of COPD (Table S21). Sensitivity analyses revealed the absence of heterogeneity and pleiotropy in our findings (Table S20). 4.3. Causal effects of both metabolic syndrome components and WC on COPD onset and exacerbation Given the strong correlation between MetS and its components, particularly abdominal obesity, we conducted MVMR analysis. In the multivariable model adjusted for WC, the independent effect of MetS on COPD was attenuated and no longer statistically significant (OR = 0.80, 95% CI: 0.38–1.71, P = .572; Table S22). Similarly, the effect of WC was attenuated after adjusting for MetS (OR = 0.53, 95% CI: 0.25–1.17, P = .115; Table S22). 5. Discussion To our knowledge, this study represents the first investigation of genetically predicted associations between MetS and its components and the onset and worsening of COPD. Our results suggest that MetS is associated with higher odds of COPD onset, COPD-related respiratory insufficiency, and pneumonia. Further investigation of MetS components revealed a strong association between WC and COPD onset and exacerbation. A nominal association was observed between hypertension and the COPD-related pneumonia endpoint; however, it did not remain significant after FDR correction. MetS has become an increasingly important problem.[5] Consistent with the results of our study, a recent prospective cohort study utilizing data from the United Kingdom Biobank revealed that, regardless of the genetic predisposition to COPD, MetS was associated with an increased risk of COPD.[14] A nationwide cohort study in Korea, with an average follow-up of 8.35 years, demonstrated that MetS was associated with a significantly elevated risk of COPD in adults aged 20 to 39 years (hazard ratio = 1.18, 95% CI: 1.11–1.24).[10] In addition, a prospective cohort analysis revealed that MetS was significantly associated with the frequency and duration of hospitalization for acute exacerbations in COPD patients.[15] In contrast to previous studies, our approach minimized the interference of confounding factors, thereby ensuring the reliability of the conclusion. The mechanisms connecting MetS and COPD are mainly related to a complex interaction between systemic inflammation, endothelial dysfunction, and oxidative stress. First, chronic low-grade inflammation in conjunction with oxidative stress is central to the pathogenesis of MetS.[5] Systemic inflammation is primarily attributed to the imbalance between the enhanced secretion of pro-inflammatory cytokines, including interleukin-6,leptin, and C-reactive protein, and the reduction in anti-inflammatory components, such as adiponectin levels.[16] The increase in circulating inflammatory factors, ectopic lipid accumulation, and development of insulin resistance would promote macrophage recruitment to adipose tissue, thereby intensifying the inflammatory response and even leading to the spread of systemic inflammation.[17,18] Leptin is essential for sustaining lung homeostasis and dramatically affects lung growth, breathing, and immunological regulation.[19] Elevated blood leptin concentrations are correlated with heightened airway and systemic inflammation.[20] A 7-year longitudinal cohort study revealed that high levels of airway and systemic inflammatory markers are correlated with a more rapid decline in lung function.[21] Second, under the pathological conditions of obesity and chronic inflammation, the overproduction of reactive oxygen species and antioxidant depletion results in oxidative stress. The irreversible accumulation of oxidation products subsequently facilitates endothelial dysfunction, which leads to MetS.[22] Elevated oxidative stress contributes to the pathogenesis of COPD through multiple mechanisms, including activation of the pro-inflammatory transcription factor nuclear factor-κB, p38 mitogen-activated protein kinase pathway, generation of autoantibodies to carbonylated proteins, decreased expression of sirtuin-1 and histone deacetylase -2, deoxyribonucleic acid damage, diminished activity of antiproteases, increased release of transforming growth factor-β, and heightened expression of mucin genes.[23] This study further investigated the causal relationship between these components and COPD using MR analysis. Previous studies have reported that, except for hyperglycemia, the remaining 4 components of MetS (abdominal obesity, hypertension, hypertriglyceridemia, and diminished HDL cholesterol) are collectively and significantly linked to the onset of COPD in individuals aged 20 to 39 years, with abdominal obesity demonstrating the strongest correlation.[10] WC is a common anthropometric parameter for assessing abdominal obesity, defined as excessive accumulation of adipose tissue in the abdominal area.[24] A longitudinal study revealed that abdominal obesity accelerates declines in peak expiratory flow and forced expiratory volume in COPD patients.[25] Increasing WC was also associated with worsened overall quality of COPD patients.[26] We provide evidence that WC may increase the risk of COPD onset and exacerbation, which is consistent with the conclusions of several previous studies. This may be ascribed to lipolysis-activating cluster of differentiation 36, which facilitates Z-DNA binding protein 1-mediated necroptosis, thereby hindering lung regeneration in COPD.[27] In addition, abdominal adiposity and being underweight were confirmed to correlate with an increased risk of hospitalization or mortality in COPD patients, according to a 10-year follow-up population-based cohort study.[28] In recent years, emerging evidence indicates that systemic adipocytic inflammation may lead to the disruption of homeostatic bone-muscle-adipose cross talk and subsequently induce osteosarcopenic obesity, which is characterized by osteopenia, sarcopenia, and fat expansion and redistribution[29] (such as accumulation in the visceral and abdominal area[30]). It is noteworthy that sarcopenia was found in 21.6% of COPD patients,[31] further suggesting important interactions between metabolic abnormalities and musculoskeletal health in COPD. Therefore, increased attention to the distribution of abdominal fat is required in clinical practice. Compared to body mass index, routine WC measurements should be integrated into COPD risk stratification. Our MVMR analysis revealed 1 notable finding. When adjusting for WC, the causal association between MetS and COPD not only lost statistical significance but also showed a reversal in the direction of the effect estimate (OR < 1). Statistically, the mutual attenuation and sign flipping observed in MVMR are characteristic signatures of high multicollinearity between the exposures.[32] This indicates that the genetic architecture of MetS and abdominal obesity are so inextricably intertwined that they cannot be statistically disentangled as separate risk entities for COPD. Biologically, this suggests that MetS does not exert a synergistic risk beyond its components; rather, its effect on COPD is mostly driven by central adiposity. This finding challenges the traditional view that MetS is an independent respiratory risk factor. Instead, it highlights central obesity as the core pathogenic driver. Mechanistically, abdominal adipose tissue restricts diaphragmatic excursion and reduces lung compliance, thereby directly compromising pulmonary function.[33] Visceral adipose tissue operates as an active endocrine organ, releasing pro-inflammatory cytokines (e.g., interleukin-6 and tumor necrosis factor-α) into systemic circulation, thereby driving airway inflammation.[34] Therefore, the clinical management of patients with COPD and MetS should prioritize central weight reduction over solely targeting metabolic indices. Our univariate analysis of hypertension showed a nominal association with COPD-associated pneumonia, although it did not remain significant after FDR correction. A cross-sectional study utilizing data from the United States National Health and Nutrition Examination Survey (1999–2018) revealed that the prevalence of COPD was markedly higher in hypertensive individuals (8.6%) than in those without hypertension (5.2%).[35] This discrepancy might be attributed to the cross-sectional design of the study; therefore, no causal inference could be drawn. A previous MR analysis found that hypertension was significantly associated with an increased risk of incident pneumonia, chronic lower respiratory disease, and other incident respiratory diseases,[36] which may support a potential link, although our findings were only nominal and should be interpreted cautiously. This may arise from 2 interconnected processes: hypertensive stimuli cause dysregulation of the adaptive immune response and concurrently induce endothelial dysfunction and facilitate pneumonia progression.[36] The major strength of our study is that the systematic evaluation of the putative causal effects (under MR assumptions) of MetS in both COPD onset and exacerbation based on MR analysis was not affected by reverse causation or traditional confounding factors. Sensitivity analyses confirmed the robustness and consistency of causal estimates. For all statistically significant associations, F-statistics exceeding 10 and statistical power above 99% confirmed the reliability of these findings, effectively minimizing Type II error risk. These steps support the robustness and interpretability of causal estimates. However, our study had some limitations. First, the GWAS data involved only participants of European ancestry, which restricts the generalizability of our findings to other racial populations owing to potential differences in linkage disequilibrium patterns. Second, publicly available summary statistics from the FinnGen consortium only provided a combined endpoint for pneumonia associated with COPD, asthma, or ILD. Owing to this database limitation, we were unable to extract data specifically for COPD-related pneumonia to perform a stratified analysis. Therefore, the results regarding pneumonia should be interpreted with caution as they may reflect a broader susceptibility to respiratory infections in patients with chronic lung diseases. Third, additional specific data, including the degree of irreversible reduction in COPD airflow and scores from questionnaires, were not available in the GWAS database. Further investigations into the relationship between MetS and COPD severity are required. 6. Conclusion This MR study provides genetic evidence of the association between MetS and both COPD onset and exacerbation, with WC showing the strongest association among MetS components. In contrast, the nominal association between genetically predicted hypertension and the combined pulmonary infection endpoint did not remain significant after FDR correction and should be cautiously interpreted. In clinical practice, the early assessment and diagnosis of COPD should be enhanced in patients with MetS. For those with comorbid COPD and MetS, multidimensional integrated management strategies and individualized strategies may help prevent acute exacerbation and slow down disease progression. Acknowledgments We sincerely acknowledge the invaluable contributions of the CTG, MRC-IEU UK Biobank GWAS Pipeline, GWAS Catalog, and FinnGen Consortium for their selfless sharing of GWAS summary data with the public. Author contributions Conceptualization: Ling Xu, Jiantong Sun, Lei Li. Data curation: Ling Xu, Jiantong Sun, Yaxin Chen. Formal analysis: Jiantong Sun. Methodology: Jiantong Sun, Lan Yang. Project administration: Yaxin Chen, Haoyu Wang. Resources: Ling Xu, Jiantong Sun, Lan Yang, Yaxin Chen. Software: Lan Yang, Haoyu Wang, Lei Li. Validation: Yaxin Chen, Haoyu Wang, Lei Li. Visualization: Ling Xu, Yaxin Chen, Lei Li. Writing – original draft: Ling Xu, Jiantong Sun, Lei Li. Writing – review & editing: Ling Xu, Jiantong Sun, Lan Yang, Lei Li.