The association between two common levels of apolipoprotein and chronic obstructive pulmonary disease: A Mendelian randomized study Authors: Chengqian Chen, Zhaobin Sun, Wentao Zhong, Hao Zheng, Zheng Zuo, Xin Zhang, Yushi Wang, Wei Zhao, Botao Shen Published: 2026 ABSTRACT The association between apolipoprotein and chronic obstructive pulmonary disease (COPD) has been reported in observational studies. However, the causality of apolipoprotein on COPD remains unknown. A 2-sample Mendelian randomization (MR) analysis was performed to investigate the causal association of apolipoprotein with COPD using summary-level data. Genome-wide association studies of two apolipoproteins, apolipoprotein A-1 (ApoA-I) and apolipoprotein B, were selected as instrumental variables. COPD was included as the outcome. A significant association between ApoA-I and COPD risk was found using the inverse-variance weighted (IVW) method (odds ratio [OR]: 0.993, 95% confidence interval [CI]: 0.990–0.997, P = .0003). However, contrasting results were observed using the weighted median (OR: 0.995, 95% CI: 0.989–1.001, P = .119) and MR-Egger (OR: 0.992, 95% CI: 0.986–0.998, P = .015) methods. Since both pleiotropy and heterogeneity tests were negative (intercept = 6.032e-06, P = .948; Q = 98.996, Q-value = 0.696), and after MR-PRESSO correction, the IVW estimates can be considered acceptable. After adjusting for confounding factors, the genetic predisposition to higher levels of ApoA-I (OR = 0.993, 95% CI: 0.990–0.997, IVW P value = 0.0003) is causally associated with a reduced risk of COPD. No causal effect was found between apolipoprotein B and COPD. 1. Introduction Chronic obstructive pulmonary disease (COPD) is a common chronic progressive airway and lung disease characterized by persistent respiratory symptoms and restricted airflow, and it has become the third leading cause of death worldwide.[1,2] Although COPD can be prevented and treated, it often progresses slowly and remains incurable. It is projected that from 2020 to 2050, the number of COPD cases among the global population aged 25 and above will increase by 23%. By 2050, the global number of COPD patients is expected to approach 600 million, imposing a significant disease burden worldwide.[3] Smoking is the primary cause of COPD, while air pollution, occupational exposure, infections, and genetic factors are also significant risk factors. These factors may contribute to the disease through mechanisms such as oxidative stress and inflammatory cell activity.[4–6] The roles of high-density lipoprotein and low-density lipoprotein in the pathogenesis of atherosclerosis have been extensively studied, and their roles in inflammation are also recognized. Apolipoprotein A-1 (ApoA-I) and Apolipoprotein B (ApoB) are 2 well-studied components of these lipoproteins. Recent national cross-sectional studies have shown that high high-density lipoprotein levels are associated with improvements in forced vital capacity and forced expiratory volume in 1 secondFEV1.Additionally, some studies have shown a positive correlation between ApoA-I and forced expiratory volume in 1 second in atopic asthma, while ApoB is associated with more severe airflow obstruction, though the underlying mechanisms remain unclear.[7,8] The pathological and physiological basis of COPD is chronic airway inflammation, accompanied by a progressive decline in lung function. Similar mechanisms may involve ApoA-I and ApoB in COPD. Existing studies examining the relationships between ApoA-I, ApoB, and COPD are predominantly observational and therefore unable to establish causal inference; moreover, no Mendelian randomization (MR) studies to date have evaluated the potential causal effects of these lipoproteins on COPD. MR is a genetic epidemiological method that uses independent single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) to investigate the causal relationships between exposure and outcomes.[9] It can reduce common sources of bias in traditional observational studies, such as confounding and reverse causation, thereby allowing more reliable and robust causal inference and avoiding potential biases inherent in observational designs. The aim of this study is to evaluate the potential causal effects of ApoA-I and ApoB levels on COPD risk from an MR perspective, specifically by applying a 2-sample MR approach. 2. Methods 2.1. Two-sample MR Study Design Based on summary-level data, a 2-sample MR analysis was performed to investigate the causal association between apolipoprotein and COPD. Two apolipoproteins, ApoA-I and ApoB, were selected for our MR study. Table 1 provides detailed information on all the summary data used in the study. The exposure and outcome genome-wide association study (GWAS) were derived from different study cohorts with no known sample overlap; therefore, bias is unlikely to have occurred. Table 1Characteristics of GWAS summary data for exposures (ApoA-I, ApoB) and outcome COPD used in the MR study.PhenotypeGWAS IDAncestrySample sizeCases/ControlsnSNPPMIDApoA-Iieu-b-107European393,193-12,321,87532,203,549ApoBieu-b-108European439,214-12,321,87532,203,549COPDukb-b-20464European-1658/110,9259,851,867-ApoA-I = apolipoprotein A-1, ApoB = apolipoprotein B, COPD = chronic obstructive pulmonary disease, GWAS = genome-wide association study, MR = Mendelian randomization, nSNP = number of single nucleotide polymorphism, PMID = PubMed unique identifier. Characteristics of GWAS summary data for exposures (ApoA-I, ApoB) and outcome COPD used in the MR study. ApoA-I = apolipoprotein A-1, ApoB = apolipoprotein B, COPD = chronic obstructive pulmonary disease, GWAS = genome-wide association study, MR = Mendelian randomization, nSNP = number of single nucleotide polymorphism, PMID = PubMed unique identifier. MR studies must meet 3 main assumptions: genetic variables should be closely related to exposure; genetic variables should not be associated with any confounders that may affect exposure and outcome; genetic variables can only influence outcomes through exposure[10] (Fig. 1). Schematic diagram of the 3 core assumptions of MR analysis in this study. ApoA-I and ApoB are the exposure factors; COPD is the outcome; SNPs serve as IVs; “Relevance assumption” indicates that SNPs are closely associated with the exposure factors (ApoA-I/ApoB); “Independence assumption” means SNPs are not associated with confounders (e.g., BMI, obesity) that affect both exposure and outcome; “Exclusion restriction” denotes that SNPs can only influence the outcome (COPD) through the exposure factors (ApoA-I/ApoB) without other direct pathways. ApoA-I = apolipoprotein A-1, ApoB = apolipoprotein B, BMI = body mass index, COPD = chronic obstructive pulmonary disease, IV = instrumental variable, MR = Mendelian randomization, SNP = single nucleotide polymorphism. MR analysis workflow: Firstly, SNP Selection: We extracted SNPs that were significantly associated with the exposure from large-scale GWAS datasets. The selection criteria were as follows: a genome-wide significance threshold of P < 5 × 10−8; removal of correlated variants through linkage disequilibrium clumping (R2 < 0.001, clumping window 10,000 kb); and inclusion of data from individuals of European ancestry only, in order to minimize population stratification bias. Secondly, Instrument Strength: We evaluated the statistical efficacy of IVs by calculating the F statistic using the formula: F = R2*(N − 2)/(1 − R2), where N represents the sample size and R2 is the percentage of exposure variability accounted for by IVs.[11] The calculation of R2 follows the method described in reference.[12] We calculated a general F statistic for all SNPs. Thirdly, Harmonization: To ensure consistent effect allele orientation between the exposure and outcome datasets, harmonization was performed for all selected SNPs. This procedure included identifying and processing palindromic SNPs, checking allele compatibility, excluding variants with ambiguous strand alignment, and confirming that the IVs were not directly associated with the outcome. We subsequently applied the Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) method to detect and remove potential horizontal pleiotropic outliers. Ultimately, only SNPs that met the harmonization criteria and showed no evidence of horizontal pleiotropy were retained for MR analyses. The workflow for SNP selection related to the exposure is illustrated in Figure 2. Workflow of IV selection for ApoA-I and ApoB in the MR study. The workflow includes extracting GWAS data of ApoA-I (ieu-b-107) and ApoB (ieu-b-108), screening SNPs meeting the genome-wide significance threshold (P < 5 × 10−8), removing correlated variants via LD clumping (R2 < 0.001, clumping window 10,000 kb), retaining SNPs of European ancestry, evaluating instrument strength using F statistic (F > 10), performing data harmonization (processing palindromic SNPs, verifying allele compatibility, excluding ambiguous strand variants), detecting and removing horizontal pleiotropic outliers via MR-PRESSO method, and finally confirming SNPs that meet MR assumptions for subsequent causal association analysis with COPD. ApoA-I = apolipoprotein A-1, ApoB = apolipoprotein B, COPD = chronic obstructive pulmonary disease, GWAS = genome-wide association study, IV = instrumental variable, LD = linkage disequilibrium, MR = Mendelian randomization, MR-PRESSO = Mendelian randomization pleiotropy residual sum and outlier, SNP = single nucleotide polymorphism. 2.2. Data source for apolipoproteins and COPD 2.2.1. ApoA-I and ApoB We extracted genetic variants for ApoA-I and ApoB from the open GWAS (https://api.opengwas.io/, ApoA-I: ieu-b-107; ApoB: ieu-b-108). The GWAS for ApoA-I comprised 393,193 individuals and 12,321,875 SNPs, while the ApoB GWAS included 439,214 individuals with the same number of SNPs.[13] IVs were selected independently for each exposure using identical criteria. SNPs were required to be significantly associated with the exposure at the genome-wide significance level (P value < 5 *10–8) and pruned for linkage disequilibrium using a clumping threshold of R2 < 0.01 within a 1000-kb window. Using these criteria, 299 SNPs were initially identified for ApoA-I and 198 SNPs for ApoB. The strength of the candidate IVs was evaluated using the F statistic, calculated as F = R2 (N − 2)/(1 − R2), where N denotes the sample size and R2 represents the proportion of variance in the exposure explained by the IVs. Among the ApoA-I–associated SNPs, 198 variants exhibited strong instrument strength with F statistics >10 and were retained for subsequent analyses. Similarly, 146 of the 198 ApoB-associated SNPs had F statistics exceeding 10 and were considered sufficiently strong instruments.[14] These exposure-associated SNPs were then harmonized with the outcome GWAS dataset. After harmonization, 111 SNPs for ApoA-I and 73 SNPs for ApoB were available for MR analyses. During this process, several palindromic SNPs with moderate allele frequencies were identified (ApoA-I: rs12977604, rs133015, rs3760230; ApoB: rs4782568, rs4935356), including 1 SNP with incompatible alleles (rs4935356). Although these SNPs were flagged as “MR_keep = TRUE” during harmonization, they were excluded from the final MR analyses to avoid potential ambiguity in allele alignment. To satisfy the third core assumption of MR, all retained SNPs were examined for direct associations with the outcome, and no SNPs showing evidence of such associations were identified. Furthermore, horizontal pleiotropy was assessed using the MR-PRESSO method, and no outlier SNPs were detected for either ApoA-I or ApoB.[15] 2.2.2. COPD We used the COPD GWAS summary data from the MRC Integrative Epidemiology Unit Consortium (ID:ukb-b-20464), which included 112,583 individuals of European ancestry (1658 cases and 110,925 controls). A total of 9,851,867 SNPs were included in this study. The summary statistics are freely downloadable from the site. All data are de-identified, openly downloadable, and can be utilized without restrictions.[16] 2.3. MR analysis and data visualization To estimate the combined effect of all SNPs, the inverse-variance-weighted (IVW) method was performed.[17] Additionally, as complementary MR methods, weighted median,[18] MR-Egger regression,[19] simple mode, and weighted mode were employed. Due to differences in analysis platforms, experimental conditions, inclusion populations, and SNPs, 2-sample MR analysis may exhibit heterogeneity. Therefore, the Cochrane Q test was performed to assess heterogeneity, if the P value > .05, it is considered that there is no heterogeneity in the IVs, and the influence of heterogeneity on the estimation of causal effects can be ignored.[20] Forest plots were used to visualize the heterogeneity.[21] MR-Egger regression analysis was used to calculate the intercept and P value of the sensitive SNPs for horizontal pleiotropy.[19]Statistic analysis and data visualization were performed using the TwoSampleMR package (version 0.5.7; MRC Integrative Epidemiology Unit, University of Bristol, Bristol, UK) and MR-PRESSO[15] package (version1.0) in R(version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria) (https://www.r-project.org/). 3. Result According to our criteria, a total of 111 SNPs associated with ApoA-I were included in this study. Among these SNPs, the minimum value of the F statistic was 10.08, indicating that the SNPs can adequately predict ApoA-I concentrations without weak instrumental bias. Utilizing these 111 ApoA-I‐related SNPs, we found evidence of a potential causal effect of ApoA-I on the risk of COPD. Specifically, the association for an effect of ApoA-I on COPD risk was found using the IVW method (odds ratio: 0.993, 95% confidence interval: 0.990–0.997, P = .0003). The results from other methods, such as weighted mode and the MR-Egger method, showed a consistent direction of influence and a significant association between ApoA-I and COPD. Additionally, heterogeneity and horizontal pleiotropy tests were performed. No heterogeneity was detected (IVW, Q = 98.996, Q-value = 0.696 and MR-Egger, Q = 98.992, Q-value = 0.672),nor was horizontal pleiotropy found (intercept = 6.032e-06, P = .948). Furthermore, the MR-PRESSO method did not identify any outlier SNPs. We also generated scatter plots (Fig. S1), sensitivity plots(Fig. S2), forest plots (Fig. S3) and funnel plots (Fig. S4). Similarly, based on our standards, this study included a total of 73 SNPs associated with ApoB. Among these SNPs, the minimum value of the F statistic was 10.11, indicating that the SNPs can adequately predict ApoB concentrations without weak instrumental bias. Unfortunately, using these 73 ApoB-related SNPs, we could not establish a potential causal impact of ApoB on the risk of COPD. In the IVW analysis, ApoB was not associated with COPD (odds ratio: 0.997, 95% confidence interval: 0.993–1.002, P = .330).Similarly, results from other methods, such as weighted median and MR-Egger, also showed no significant association between ApoB and COPD in Mendelian randomized analyses. All MR analysis results are shown in Table 2 and Table 3. Table 2MR analysis results for the causal association between ApoA-1 and COPD.MethodSEOR[95% LCI, 95% UCI]P value Q Q-valueInterceptIVW0.0017743690.9936889[0.990, 0.997].0003---Simple mode0.0057232260.9935880[0.982, 1.004].263---Weighted mode0.0029406760.9939845[0.988, 0.999].042---Weighted median0.0030779720.9953175[0.989, 1.001].127---MR Egger0.0031456500.9935203[0.987, 0.999].041---Heterogeneity(IVW)----98.9960.696-Heterogeneity(MR-Egger)----98.9920.672-Pleiotropy---.9486.032e-06ApoA-I = apolipoprotein A-1, CI = confidence interval, COPD = chronic obstructive pulmonary disease, IVW = inverse-variance weighted, LCI = lower confidence interval, MR = Mendelian randomization, nSNP = number of single nucleotide polymorphisms, OR = odds ratio, Q = Cochran’s Q statistic, Q_df = degrees of freedom for Q, Q_pval = P value for heterogeneity test, SE = standard error, UCI = upper confidence interval. MR analysis results for the causal association between ApoA-1 and COPD. ApoA-I = apolipoprotein A-1, CI = confidence interval, COPD = chronic obstructive pulmonary disease, IVW = inverse-variance weighted, LCI = lower confidence interval, MR = Mendelian randomization, nSNP = number of single nucleotide polymorphisms, OR = odds ratio, Q = Cochran’s Q statistic, Q_df = degrees of freedom for Q, Q_pval = P value for heterogeneity test, SE = standard error, UCI = upper confidence interval. Table 3Mendelian randomization analysis results for the causal association between apolipoprotein B and chronic obstructive pulmonary disease (COPD).MethodSEOR[95% LCI, 95% UCI]P valueIVW0.0022780270.9977855[0.993, 1.002].330Simple mode0.0060905770.9967537[0.984, 1.008].595Weighted mode0.0035141960.9992701[0.992, 1.006].836Weighted median0.0034318800.9990867[0.992, 1.005].790MR Egger0.0047712781.0023625[0.993, 1.011].622ApoB = apoprotein B, CI = confidence interval, COPD = chronic obstructive pulmonary disease, IVW = inverse-variance weighted, LCI = lower confidence interval, nSNP = number of single nucleotide polymorphisms, OR = odds ratio, Q = Cochran’s Q statistic, Q_df = degrees of freedom for Q, Q_pval = P value for heterogeneity test, SE = standard error, UCI = upper confidence interval. Mendelian randomization analysis results for the causal association between apolipoprotein B and chronic obstructive pulmonary disease (COPD). ApoB = apoprotein B, CI = confidence interval, COPD = chronic obstructive pulmonary disease, IVW = inverse-variance weighted, LCI = lower confidence interval, nSNP = number of single nucleotide polymorphisms, OR = odds ratio, Q = Cochran’s Q statistic, Q_df = degrees of freedom for Q, Q_pval = P value for heterogeneity test, SE = standard error, UCI = upper confidence interval. 4. Discussion This is the first study to use MR to explore the causal relationship between ApoA-I and ApoB and COPD. Our results suggest that ApoA-I may have a potential protective association with COPD, whereas ApoB does not show a causal relationship at the genetic level. Inflammation plays a crucial role in the occurrence and development of COPD. Several potential mechanisms could explain the protective role of ApoA-I in COPD. Neutrophils, under the action of chemokines, can aggregate at specific sites and cause inflammation, with CXC chemokine receptor 2 being a key component of the chemokine receptor.[22] In vivo experiments, ApoA-I showed the ability to inhibit CXC chemokine receptor 2 dependent neutrophil migration.[23] ApoA-I may reduce the chemotaxis of macrophages and monocytes towards C-C chemokine motif ligand 5 by regulating the phosphoinositide 3-kinase/protein kinase B signaling pathway.[24] Smoking is the primary cause of COPD, and significant amounts of lipopolysaccharide (LPS) have been found in tobacco. LPS can induce emphysema and has been used in the construction of COPD models.[25–27] Lipid rafts play a critical role in promoting pro-inflammatory signaling through Toll-like receptor-4 (TLR4), and cluster of differentiation (CD)14 is involved in the recognition of LPS, activating TLR4 signaling.[28,29] ApoA-I has been shown to exert anti-inflammatory effects by modulating these pathways. Specifically, ApoA-I can reduce the expression of key inflammatory molecules, including human leukocyte antigen-DR, CD86, CD11b, CD11c, CD14 and TLR4, thereby inhibiting the secretion of pro-inflammatory cytokines like tumor necrosis factor-alpha, interleukin 6, and monocyte chemoattractant protein-1, which are critical in the inflammatory response in COPD.[30–32] Additionally, ApoA-I can disrupt lipid rafts by increasing the expression of ATP-binding cassette subfamily A member 1 and cholesterol efflux, which may contribute to the attenuation of inflammatory signaling. ApoA-I’s ability to regulate cholesterol efflux and inhibit macrophage pro-inflammatory signaling through ATP-binding cassette subfamily A member 1-mediated mechanisms is also relevant to COPD pathology. These effects provide a plausible explanation for the protective role of ApoA-I in reducing inflammation and lung damage associated with COPD.[33] Data also showed that overexpression of ApoA-I alleviated cigarette smoke-induced lung inflammation and emphysema in mice.[34] ApoB is considered a pro-inflammatory factor. For example, the ApoB/A-1 ratio is a criterion for evaluating severe acute pancreatitis, but in this study, it was not associated with COPD.[35] This can be considered from 2 perspectives. First, from a biological standpoint, ApoB may have a relatively weak association with the pathogenesis of COPD, as its relationship with airway inflammation or pulmonary remodeling is far less clear than that of ApoA-I; thus, ApoB may indeed lack a significant biological effect. Second, this may be related to several limitations of the present study. First, the number of cases in the COPD GWAS was relatively small (1658 cases), which may limit the statistical power to detect genetic variants with modest effects and could result in false-negative findings. Second, due to the absence of individual-level data, we were unable to perform stratified analyses based on smoking status, body mass index, or other lifestyle and clinical characteristics. Third, both the exposure and outcome GWAS were conducted in populations of European ancestry, which limits the generalizability of our findings. Finally, due to data constraints, we were unable to perform nonlinear MR analyses or subgroup analyses based on variables such as age, sex, or smoking intensity. Because the primary aim of this study was to investigate the individual genetic causal effects of each biomarker, and the available GWAS datasets for the exposures and outcomes were not fully compatible, multivariable MR analysis was not appropriate in the present study. Therefore, further research involving diverse populations is necessary to draw more accurate conclusions. 5. Conclusion This MR study indicates that genetically predicted higher ApoA-I levels are associated with a lower risk of COPD, suggesting a potential protective effect, whereas no significant association was observed between ApoB and COPD. Given several limitations of the study, these findings warrant further validation in larger samples, diverse populations, and studies with more detailed phenotypic data. Acknowledgments We sincerely thank all the researchers who have contributed to the public GWAS database. Author contributions Conceptualization: Chengqian Chen. Data curation: Chengqian Chen. Formal analysis: Chengqian Chen. Investigation: Xin Zhang. Methodology: Zhaobin Sun, Wentao Zhong. Resources: Chengqian Chen, Wentao Zhong, Hao Zheng, Zheng Zuo, Xin Zhang. Software: Chengqian Chen, Wentao Zhong, Hao Zheng, Xin Zhang. Supervision: Zheng Zuo, Yushi Wang, Wei Zhao, Botao Shen. Validation: Botao Shen. Visualization: Zheng Zuo, Wei Zhao. Writing – original draft: Chengqian Chen, Zhaobin Sun, Wentao Zhong, Hao Zheng. Writing – review & editing: Yushi Wang, Wei Zhao, Botao Shen.