https://orcid.org/0000-0003-1542-0828
Figures
Abstract
Lung cancer (LC) is the leading cause of cancer-related mortality worldwide, accounting for millions of deaths annually. Its major subtypes—lung squamous carcinoma (LUSC), lung adenocarcinoma, and small-cell LC—exhibit distinct risk factors and genetic susceptibilities, necessitating the use of subtype-specific biomarkers. Two-sample Mendelian randomization (MR) analyses were conducted using protein quantitative trait loci from the UK Biobank Pharma Proteomics Project and deCODE datasets. A robust analytical framework, including reverse MR, meta-analysis, summary-data-based MR tests, and colocalization, cisMR-cML, MR.CUE and phenotype scanning analyses were used to identify proteins associated with LC risk. We conducted a systematic review to contextualize our research findings. Follow-up analyses, including pathway enrichment, protein-protein interaction network analysis, and druggability evaluations, were used to explore the mechanisms and therapeutic potential of the identified proteins. Significant proteins were validated using population-level proteomic data from the UK Biobank (UKB). The results showed that twenty-five proteins were significantly associated with LC or its subtypes, including 15 novel findings. 60S ribosomal protein L14 (RPL14) and advanced glycosylation end-product-specific receptor (AGER) emerged as the strongest discovery, demonstrating consistent and significant associations across both MR and population-level analyses. RPL14 exhibited positive associations with overall LC risk (MR_meta: odds ratio [OR]: 2.012, 95% confidence interval [CI]: 1.297–3.119; UKB: OR: 1.509, 95% CI: 1.015–2.244). Similarly, AGER showed significant protective effects against LUSC risk (MR_meta: OR: 0.572, 95%CI: 0.368–0.889; UKB: OR: 0.366, 95% CI: 0.158–0.850). Pathway analysis revealed the involvement of these proteins in immune regulation and tumorigenesis. Among the 13 identified druggable targets, RPL14 and AGER showed therapeutic potential as approved or investigational drugs targeting these proteins. These findings offer new insights into the pathogenesis of LC and potential therapeutic targets.
Author summary
Lung cancer is responsible for millions of deaths worldwide each year, creating a substantial health burden. Proteins, as the end products of gene expression and essential functional components in biological processes, serve as valuable biomarkers and therapeutic targets for complex diseases. However, different subtypes exhibit varying genetic susceptibilities, and the connections between a wide range of proteins and lung cancer remain unclear. Therefore, investigating the relationships between proteins and lung cancer, along with its subtypes, is crucial for uncovering insights into risk prediction and treatment. In our study, we explored the associations between thousands of proteins and lung cancer, identifying causal links for twenty-five proteins related to lung cancer risk or its subtypes. New associations in our findings were identified using systematic reviews. Based on population evidence, we highlighted the significance of the RPL14 and AGER in lung cancer risk. We also proposed that RPL14 and AGER may play roles in lung carcinogenesis and serve as therapeutic targets, supported by follow-up analyses. These findings hold important implications for both predicting lung cancer risk and developing innovative treatment strategies.
Citation: Lyu Z, Si G, Xing M, Li W, Gao X, Wang M, et al. (2025) Circulating proteins associated with histological subtypes of lung cancer from genetic and population-based perspectives. PLoS Genet 21(8): e1011821. https://doi.org/10.1371/journal.pgen.1011821
Editor: Zihuai He, Stanford University, UNITED STATES OF AMERICA
Received: January 19, 2025; Accepted: July 22, 2025; Published: August 25, 2025
Copyright: © 2025 Lyu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data used here are publicly available. UKB-PPP plasma protein GWAS data obtained on the website Synapse (Project SynID: syn51364943, https://www.synapse.org/#!Synapse:syn51364943; doi: 10.1038/s41586-023-06592-6), deCODE plasma protein GWAS data (https://www.decode.com/summarydata/; doi: 10.1038/s41588-021-00978-w), GWAS data on lung cancer from the FinnGen database (https://www.finngen.fi/en;doi:10.1038/s41586-022-05473-8; lung cancer: https://storage.googleapis.com/finngen-public-data-r10/summary_stats/finngen_R10_C3_BRONCHUS_LUNG_EXALLC.gz; lung squamous carcinoma: https://storage.googleapis.com/finngen-public-data-r10/summary_stats/finngen_R10_C3_NSCLC_SQUAM_EXALLC.gz; lung adenocarcinoma: https://storage.googleapis.com/finngen-public-data-r10/summary_stats/finngen_R10_C3_NSCLC_ADENO_EXALLC.gz; small-cell lung cancer: https://storage.googleapis.com/finngen-public-data-r10/summary_stats/finngen_R10_C3_SCLC_EXALLC.gz), GWAS data on lung cancer from the TRICL can get from GWAS Catalog (http://www.ebi.ac.uk/; doi: 10.1038/ng.3892; lung cancer: http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST004001-GCST005000/GCST004748; lung squamous carcinoma: http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST004001-GCST005000/GCST004750; lung adenocarcinoma: http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST004001-GCST005000/GCST004744; small-cell lung cancer: http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST004001-GCST005000/GCST004746), SMR (https://yanglab.westlake.edu.cn/software/smr/; doi: 10.1038/ng.3538), the Human Protein Atlas (https://www.proteinatlas.org/; doi: 10.1111/j.1365-2796.2011.02427.x), phenoscanner (http://www.phenoscanner.medschl.cam.ac.uk; doi: 10.1093/bioinformatics/btz469), Metascape online tool (https://metascape.org/; doi: 10.1038/s41467-019-09234-6), STRING website (https://cn.string-db.org/; doi: 10.1093/nar/gkac1000), DrugBank (https://go.drugbank.com/; doi: 10.1093/nar/gkad976), Therapeutic Target Database (https://go.drugbank.com/; doi: 10.1093/nar/gkad751), ChEMBL (https://www.ebi.ac.uk/chembl/; doi: 10.1093/nar/gkad1004), DGIdb (https://www.dgidb.org/; doi: 10.1093/nar/gkaa1084), UK Biobank population-level data is available by application (https://www.ukbiobank.ac.uk/, doi: 10.1038/s41586-023-06592-6).
Funding: This work was supported by National Natural Science Foundation of China (82430107 to KXC; 82204121 to ZYL); Natural Science Foundation of Tianjin (23JCQNJC00800 to ZYL; 22KPHDRC00270 to MW); China Postdoctoral Science Foundation (2023M742617 to ZYL); Tianjin Health Talent Development Project (TJSJMYXYC-D2-038 to FJS); Tianjin Key Medical Discipline (Specialty) Construction Project (TJYXZDXK-009A). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors declare that they have no competing interests.
Abbreviations: LC, lung cancer; LUAD, lung adenocarcinoma; LUSC, lung squamous carcinoma; SCLC, small-cell lung cancer; SNPs, single nucleotide polymorphisms; MR, mendelian randomization; RCT, randomized clinical trial; IVs, instrumental variables; SMR, summary-data-based Mendelian randomization; UKB, UK Biobank; RPL14, 60s ribosomal protein l14; AGER, advanced glycosylation end-product-specific receptor; cis-pQTL, cis-protein quantitative trait locus; UKB-PPP, UK Biobank Pharma Proteomics Project; TRICL, Transdisciplinary Research In Cancer of the Lung; IVW, inverse variance weighted; PPH4, the posterior probability of hypothesis 4 in colocalization analysis; FDR, false discovery rate; LD, linkage disequilibrium; COPD, chronic obstructive pulmonary disease; PPI, protein-protein interaction; TTD, therapeutic target database; PSM, propensity score matching; TDI, Townsend deprivation index; SMD, standardized mean differences; BMI, body mass index; OR, odds ratio; CI, confidence interval; NSCLC, non-small-cell lung cancer; AT1, alveolar type 1; TPCPA, The Pan-Cancer Proteome Atlas; S100A4, protein S100-A4; MPI, mannose-6-phosphate; IL19, Interleukin 19; SERPING1, Plasma protease C1 inhibitor; USP28, Ubiquitin carboxyl-terminal hydrolase 28; TREM1, the triggering receptor expressed on myeloid cells 1
Introduction
Lung cancer (LC) remains the predominant cause of cancer-related mortality globally, accounting for approximately 1.8 million fatalities in 2022 [1]. The most prevalent histological subtype is lung adenocarcinoma (LUAD, 45%), followed by lung squamous carcinoma (LUSC, 20%), small-cell LC (SCLC, 11%), and large-cell lung carcinoma (7%) [2]. The 5-year survival rates for patients with LC are relatively low, ranging from 10% to 20% in most countries [3–5]. Despite advancements in the screening and treatment of LC, early detection and targeted anticancer therapies remain the most promising modalities for enhancing long-term survival [6]. Exploring noninvasive biomarkers holds significant potential for understanding the LC etiology and developing tailored treatment strategies [7,8].
Proposed biomarkers span a wide range, including single nucleotide polymorphisms (SNPs), circulating tumor DNA methylation, microRNAs, metabolites, and proteins [9–15]. Proteins, as the end products of gene expression and primary functional components of biological processes, represent promising intermediate phenotypes for discovering novel mechanisms underlying major chronic diseases [16]. Previous observational research leveraging high-throughput proteomic technologies has identified several circulating proteins associated with LC risk across various pathological types [17,18]. However, potential residual confounding bias (e.g., arising from unmeasured or inaccurately measured confounders) and reverse causality in observational research limit the ability to draw causal inferences from these studies.
Mendelian randomization (MR) study is a powerful causal inference tool in epidemiological research, known as Mother Nature’s randomized clinical trial (RCT) [19]. It utilizes genetic variants strongly associated with exposure (anthropometric characteristics, behavior, circulating proteins/metabolites/genes, etc.) as instrumental variables (IVs) to infer causal associations between exposure and outcome. Because these genetic variants are acquired naturally and randomly at fertilization (meiosis), the associations revealed by MR are not subject to reverse causation and are unlikely to be affected by confounding factors, effectively addressing the limitations of observational studies [19]. LC has a significant genetic basis (heritability ~ 18%) and familial history (first-degree relative) confers a 50% risk elevation [20–22]. Consequently, MR has been widely adopted to investigate its etiology. Given subtype-specific variations in genetic susceptibility loci and key risk factors (e.g., smoking, particulate matter exposure, automobile exhaust), distinct circulating protein biomarkers likely exist across subtypes [23–25]. However, a systematic review of existing MR studies on circulating proteins and LC risk (S1 and S2 Tables) reveals limitations in sample size [26], proteome coverage [27], subtype-specific analyses [23], and real-world population validation. Future studies should address these gaps to strengthen causal inference.
This study combined data from two extensive proteomic GWAS studies and two large LC GWAS studies to explore the associations between circulating proteins and the risk of LC and its three subtypes. Using MR methods and multiple analytical approaches (meta-analysis, summary-data-based MR [SMR] tests, colocalization, cisMR-cML, MR.CUE, and phenotype scanning), we identified 25 significant proteins linked to LC or its subtypes, 15 of which were novel. Follow-up assessments further revealed the functions of these proteins, identifying 13 proteins (e.g., RPL14, AGER, PDE5A) as having therapeutic potential. In addition, individual-level data obtained from the UK Biobank (UKB) validated the associations between 60S ribosomal protein L14 (RPL14) and LC risk, as well as advanced glycosylation end-product-specific receptor (AGER) and LUSC risk.
Methods
Ethics statement
The original studies of GWAS data used in this study have been approved by relevant ethical review institutions and informed consent of participants. This study only analyzed publicly available summary-level statistical data, and therefore, no new ethical review board approval was required. UK Biobank received ethics approval from the Research Ethics Committee (REC reference for UK Biobank is 21/NW/0157). The UK Biobank data used in this study were under the approved application number 76092. All participants provided written informed consent, and the North West Multi-centre Research Ethics Committee approved the study.
Fig 1 shows the comprehensive study design. Using cis-protein quantitative trait locus (cis-pQTL) as genetic instrumental variables, we conducted two-sample MR analyses to identify candidate causal proteins associated with LC risk. Meta-analyses, reverse MR analyses, SMR tests, HEIDI tests, colocalization analyses, cisMR-cML, MR.CUE and phenotype scanning analyses were conducted to identify significant causal proteins linked to LC risk. For the significant proteins, our review helped identify novel associations. Furthermore, follow-up analyses were performed to investigate the potential mechanisms underlying the proteins involved in LC carcinogenesis and their druggability. Additionally, we used plasma proteomic data from the UK Biobank to search for evidence of associations between the prediagnostic circulating levels of significant proteins and LC risk.
(LD: linkage disequilibrium; IVs: instrument variables; IVW: inverse variance weighted; MR: Mendelian randomization; LC: lung cancer; LUSC: lung squamous carcinoma; LUAD: lung adenocarcinoma; SCLC: small cell lung cancer; UKB-PPP: the UK Biobank Pharma Proteomics Project; SMR: summary-data-based MR test; PSM: propensity score matching).
Data sources
Summary-level GWAS data for circulating proteins were obtained from two large cohorts: the UK Biobank Pharma Proteomics Project (UKB-PPP) and the deCODE study. The UKB-PPP study examined plasma proteins in 54,219 participants with an average age of 57 years and publicly available GWAS data for 2,940 (2,922 unique) proteins [28], as detailed in S3 Table. The deCODE study involved 35,559 Icelanders with a mean age of 55 years and measured the 4,667 unique plasma protein levels of 4,907 aptamers [29], as detailed in S4 Table. The two studies covered a total of 1,822 overlapping proteins.
The GWAS data for LC and its three subtypes (LUSC, LUAD, and SCLC) were obtained from the FinnGen R10 release data, with a sample size of more than 412,181 Finnish individuals and a median age of 63 years [30]. The other summary-level GWAS data of LC were obtained from GWAS Catalog datasets and carried out under the auspices of the Transdisciplinary Research in Cancer of the Lung (TRICL) study linked to the International Lung Cancer Consortium. A total of 29,266 patients with LC and 56,450 healthy controls were included, of whom 88% were over 50 years of age [31]. Table 1 provides an overview of the websites for the GWAS data sources used in this study.
The population-level proteomic data were derived from the UK Biobank, which is a large prospective cohort study that collects information regarding demographics, body measurements, diet, lifestyle, health status, plasma markers, and genetic data from 500,000 participants aged 40–69 years recruited in 2006–2010 [32]. Proteomic data were provided as Normalized Protein eXpression (NPX) values on a log2 scale measured using an antibody-based Olink Explore 3072 PEA panel [28].
Two-sample MR analyses to identify candidate proteins
We used univariate MR to determine the association of circulating proteins with LC using cis-pQTL (± 1Mb) as genetic IVs (detailed in S1 Text). When only one genetic variant was available, analysis was performed using the Wald ratio method. When two or more genetic variants were present, the analysis was performed using the inverse variance weighted (IVW) method of the random-effects model. Meanwhile, MR-Egger, weighted median, weighted mode, and simple mode were used for supplementary analyses [33]. Heterogeneity and pleiotropy tests were conducted using Cochran’s Q statistic and the MR-Egger intercept, respectively. Steiger directionality tests were used to detect the direction of MR association, the result shows “TRUE” means it is from exposure to outcome. The power of MR analyses was measured using the mRnd online tool. The Benjamini-Hochberg method was used to adjust for multiple tests.
Candidate LC-associated proteins were identified if they fulfilled any of the following conditions: (1) significantly associated with LC from any source (false discovery rate [FDR] < 0.05) or (2) associated with LC from both the FinnGen and TRICL databases (p < 0.05).
MR_meta, reverse_MR, SMR, colocalization, cisMR-cML, MR.CUE and phenotype scanning analyses to identify significant proteins
We identified significant proteins in four stages.
First, we performed a meta-analysis of the MR results for the candidate proteins and LC in the TRICL and FinnGen studies. A random-effects model was used when heterogeneity was observed, while a fixed-effects model was applied otherwise. Heterogeneity was defined as a p < 0.05 in the Q-test.
Second, we performed reverse MR analyses to avoid the effect of reverse causality. When p < 0.05, it represents that the presence of LC can affect protein levels, indicating a possible reverse causality. The IV screening approach for LC is shown in S1 Text.
Third, SMR and colocalization analyses were conducted to validate the association between plasma proteins and LC (FinnGen) [34]. SMR analyses were performed using the default parameters (cis-wind = 2,000 kb, trans-wind = 1,000 kb, peqtl-trans = 5e-8, peqtl-other = 1e-5, thread-num = 10, and diff-freq-prop = 0.3). The HEIDI tests were used for sensitivity analyses to distinguish whether the causal relationships obtained were due to shared genetic variation rather than chain disequilibrium effects [34,35]. The significance of SMR analysis was defined as pSMR < 0.05 and pHEIDI > 0.01. Using the parameters of p1 = 1 × 10-4, p2 = 1 × 10-4, and p12 = 1 × 10-5, we conducted a colocalization analysis between plasma proteins and LC, and shared genetic variations were deemed to be present when the posterior probability of hypothesis 4 (PPH4) > 0.75 [23]. Proteins that satisfied the SMR or colocalization evidence would be processed for the fourth step of validation.
Fourth, to strengthen the robustness of causal inference and to correct for pleiotropic interference effects, we performed cisMR-cML and MR.CUE analyses of significant proteins and phenotype scanning analyses of IVs. In cisMR-cML analysis, we selected genetic variants based on cis-pQTL (±1Mb), using conditional and joint association analysis (called GCTA-COJO) to select SNPs [36,37]. Then, the cismr_cML_DP function is used to determine the association effects between proteins and LC. In MR.CUE analysis, using the independent reference panel data provided by Chen et al [38]. Preset pva_cutoff = 5E-8 to pick significant IVs, using the ReadSummaryStat function to match the three data sets (exposure, outcome, and panel data) and align effect sizes, where the shrinkage transformation parameter lambad = 0.85 for the linkage disequilibrium (LD) estimator is set [38]. The MRCUE function was used to estimate the association effect. Significance of any one of cisMR-cML, MR.CUE analyses were demonstrated by proving the robustness of the causal association. For phenotype scanning, we utilized SNP-phenotype association data from the GWAS Catalog (v1) to look at other relevant phenotypes for IV of the study proteins [39]. We excluded SNPs associated with known risk factors (e.g., smoking, asthma, chronic obstructive pulmonary disease [COPD]) from significant protein IVs (p ≤ 1E-5) and subsequently repeated the cis-MR analyses to confirm result stability. See more information in the S1 Text.
Follow-up analyses
To gain deeper insights into the mechanism of action of the significant proteins and their druggability, we performed single-cell expression analysis, pathway enrichment analysis, protein-interaction (PPI) network analysis, and druggability assessments. Single-cell expression analysis was used to investigate the cell types in which the genes encoding the significant proteins were predominantly expressed in lung tissues using the Human Protein Atlas database [40]. Pathway enrichment analysis was carried out on the Metascape website, with the protein-coding genes of UKB-PPP and deCODE as background gene sets, with the following screening criteria: minimum overlap = 3, p-value threshold = 0.05, and minimum enrichment = 1.5 to identify pathways and processes enriched for significant proteins [41]. To clarify the interactions between the significant proteins, as well as between the significant proteins and established LC therapy targets, we queried the drug targets related to LC in DrugBank (https://go.drugbank.com/) and then discovered the interactions using the STRING database. Cytoscape software was used to map PPI plots. Finally, we searched four drug target-related databases: DrugBank [42], Therapeutic Target Database (TTD, https://go.drugbank.com/) [43], ChEMBL (https://www.ebi.ac.uk/chembl/) [44], and DGIdb (https://www.dgidb.org/) [45], to further assess whether the significant proteins could be potential therapeutic targets. We documented the details of the drug names and possible indications related to the significant proteins.
Real-world study
S1 Fig and S1 Text present the study population screening flowchart and data processing. Using a nested case-control design, we performed propensity score matching (PSM) at a 1:5 ratio to match LC new cases (N = 322) with cancer-free controls (N = 1,610). The LC cases comprised LUSC (N = 47), LUAD (N = 103), and SCLC (N = 35). PSM was performed based on logistic regression-derived propensity scores. Matching covariates included age, sex, ethnicity, Townsend deprivation index (TDI), and family history of LC. Nearest neighbor matching was implemented with a caliper width of 0.1 standard deviations of the propensity score. Post-matching covariate balance was evaluated using standardized mean differences (SMD), with SMD < 0.1 indicating adequate balance. Multivariable conditional logistic regression models were used to assess associations between circulating proteins and lung cancer risk. Adjustment variables—determined based on scientific rationale and prior publications—included body mass index (BMI), smoking status, drinking status, physical activity, healthy diet, respiratory diseases, and blood glucose [46,47]. See S5 Table for variable definitions.
All participants provided written informed consent, and the North West Multi-centre Research Ethics Committee approved the study. The UK Biobank data used in this study were under the approved application number 76092.
Results
Identification of candidate LC-associated proteins
After screening and harmonizing with outcome data, the two studies covered a total of 879 overlapping proteins, with an additional 1,062 unique proteins provided by UKB-PPP and 792 unique proteins supplied by the deCODE study, resulting in a total of 2,733 unique predicted proteins analyzed. The F-statistics of the instrumental variables for these proteins were all higher than 10. The proportion of variance in exposure explained by a single genetic variant ranged from 0.0004 to 0.387, as outlined in S6 Table.
The complete results of the MR analyses are shown in S7–S22 Table. A total of 59 unique plasma proteins fulfilling the criteria for candidate protein screening were identified. Among these, 27 genetically predicted proteins were causally associated with LC risk, 14 with LUSC, 14 with LUAD, and eight with SCLC (Fig 2, S23 Table). Notably, four circulating proteins (APOBR, DAG1, MICB_MICA, and RNASET2) were causally linked to more than one LC subtype, while the remaining 55 proteins showed evidence of a primary causal association with only one LC subtype. Additionally, genetically predicted levels of AGER identified from both exposure databases were highlighted as candidate causal proteins, suggesting stronger evidence of association.
(LC: lung cancer; LUSC: lung squamous carcinoma; LUAD: lung adenocarcinoma; SCLC: small cell lung cancer; Candidate lung cancer-associated proteins were defined as proteins that were significantly associated with lung cancer (FDR < 0.05); or with lung cancer from both sources (FinnGen and TRICL) (p < 0.05); OR: odds ratio; FDR: false discovery rate; (1): the protein source is the UKB-PPP project; (2): the protein source is the deCODE study; Not applicable: no SNPs were available for analysis after harmonization).
No evidence of heterogeneity or pleiotropy was observed (p > 0.05). Steiger directionality tests confirmed that the direction of causality consistently pointed from proteins to LC risk. Most of these analyses achieved a statistical power of up to 80%, as detailed in S24 Table.
Identification of significant LC-associated proteins
The meta-analyses’ results revealed that 48 genetically predicted proteins were associated with the risk of LC or its subtypes. Significant associations were observed between 19 proteins and overall LC risk, 13 proteins with LUSC, 12 proteins with LUAD, and six proteins with SCLC (S25 Table). Notably, two of these proteins were causally linked to an increased risk of two LC subtypes. Specifically, genetic liability to elevated MICB_MICA protein levels increased the risk of LC and LUSC, while genetically predicted high RNASET2 levels were associated with an increased risk of LC and LUAD. Of the remaining proteins, 26 showed negative associations with a specific LC subtype, while 20 demonstrated positive associations.
The number of IVs for overall LC, LUSC, LUAD, and SCLC from the FinnGen database was 7, 2, 1, and 4, respectively, while the number of IVs from the TRICL database was 14, 7, 10, and 2. All IVs exhibited F-statistics greater than 10, and the proportion of variance explained by a single genetic variant ranged from 0.004 to 0.046. Reverse MR analyses with LC as the exposure and protein levels as the outcome, suggested that protein PLAU may be affected by LUAD (S26 Table). To avoid the potential reverse causality effect, this protein was excluded from subsequent analyses.
Furthermore, the results of the SMR analyses indicated that 26 proteins were significantly associated with LC or its subtypes (p < 0.05), with the direction of these associations aligning with the MR_meta evidence (S27 Table). Among them, 13 genetically predicted proteins were significantly causally associated with overall LC risk. Five proteins (MICB_MICA, IL19, UBE2L6, RPL14, and STX4) were associated with elevated risk, while eight (BTN3A1, CTF1, TNFAIP8L2, CEP43, S100A4, TSG101, MPI, and SERPING1) were linked to reduced risk. Six proteins were significantly causally associated with LUSC risk. Four proteins (TIE1, PDE5A, ADM2, and USP28) were causally linked to an increased risk, while two (BRD2 and AGER) were causally associated with reduced risk. In the case of LUAD, four proteins showed a significant causal link. Three (CTBS, RPN1, and ELOA) were linked to an increased risk of LUAD, whereas MCF2L was associated with a decreased risk. Finally, three proteins were significantly causally associated with SCLC risk: GLRX2 was linked to an increased risk, while TREM1 and DAG1 were identified as protective factors.
The HEIDI test indicated no heterogeneity in these associations (p > 0.01). Notably, the colocalization analysis provided strong support for two associations: AGER with LUSC (PPH4 = 0.821) and CTBS with LUAD (PPH4 = 0.785), indicating a high probability of shared causal variability between protein levels and LC risk (S28 Table).
CisMR-cML and MR.CUE analyses revealed causal associations between 25 proteins and specific LC subtypes, supported by at least one method. However, the association of SERPING1 with overall LC risk may be influenced by pleiotropy and was not validated through either approach (S29 and S30 Tables). Phenotype scanning identified associations between rs204993 (AGER, ID = oid20756) and asthma, as well as rs117960683 (SERPING1, ID = 4479_14_SERPING1_C1_Esterase_Inhibitor) and daily smoking (S31 Table). After removing these SNPs, AGER maintained a significant protective association with LUSC risk, whereas the association between SERPING1 and LC risk (FinnGen) was no longer robust (S32 Table).
The summary of 25 significant protein-LC associations is presented in S33 Table and Fig 3. Compared with 15 MR studies on circulating proteins and LC risk identified through our systematic review (S34 and S35 Tables), 10 protein-LC associations were further validated (LC: IL19, MICB_MICA, TSG101, and CTF1; LUSC: PDE5A and TIE1; LUAD: MCF2L and RPN1; and SCLC: GLRX2 and TREM1), while the remaining 15 causal associations represent novel findings (LC: RPL14, STX4, UBE2L6, S100A4, BTN3A1, CEP43, MPI, and TNFAIP8L2; LUSC: ADM2, USP28, AGER, and BRD2; LUAD: CTBS and ELOA; and SCLC: DAG1).
(LC: lung cancer; LUSC: lung squamous carcinoma; LUAD: lung adenocarcinoma; SCLC: small cell lung cancer; UKB-PPP: the UK Biobank Pharma Proteomics Project; TRICL: the Transdisciplinary Research In Cancer of the Lung Research Team; SMR: summary-data-based MR test; PPH4: the posterior probability of hypothesis 4 in colocalization analysis; NA: not applicable, no SNPs were available for analysis after harmonization).
Follow-up analysis
Single-cell expression.
These 25 significant protein-coding genes were primarily expressed in 11 distinct cell types in the lung, including alveolar cells (type 1/2), macrophages, endothelial cells, fibroblasts, T cells, granulocytes, club cells, B cells, ciliated cells, and smooth muscle cells (Fig 4A and S36 Table). Notably, BRD2, RPL14, and S100A4 were predominantly enriched in these cells, while AGER was particularly enriched in alveolar cells and fibroblasts.
(A: single-cell expression types of significant proteins; B: pathway enrichment analysis of significant proteins; C: Protein-Protein Interaction Networks of significant proteins; nTPM: Transcripts Per Million, all TPM values per sample were scaled to a sum of 1 million TPM).
Pathway enrichment.
Pathway enrichment analysis revealed that these proteins were primarily involved in pathways related to T cell activation, regulation of blood circulation, innate immune response, and other biological activities. These findings indicate that these proteins play crucial roles in biological processes such as immune system regulation and blood circulation control (Fig 4B and S37 Table).
PPI network.
Using the DrugBank database, we retrieved 88 LC-related drug targets (S38 Table). By combining these targets with the 25 significant proteins, we performed a PPI network analysis. Nine proteins (AGER, S100A4, TSG101, RPL14, DAG1, IL19, CTF1, GLRX2, and BRD2) interacted with known LC targets, such as TSG101 with EGFR. Additionally, AGER interacted with S100A4, suggesting involvement in inflammatory and immune processes (Fig 4C).
Druggability evaluation.
A search of four drug target databases identified 13 proteins with either investigational or approved drugs (S39 Table). Among these, PDE5A has been explored as a potential therapeutic target for LC. The drug Exisulind, targeting PDE5A, has potential applications in treating LC, colorectal adenomatous polyposis, and colon polyps. The remaining 12 proteins are also potential therapeutic targets and are currently associated with other diseases. Examples of approved drugs include AGER, targeted by Oxytocin (used for preeclampsia) and Vitamin B12 (used for anemia caused by B12 deficiency); BRD2, targeted by Triazolam (a sedative for insomnia) and Alprazolam (for anxiety disorders and epilepsy); CEP43 targeted by Brivanib, investigated for breast cancer and liver cancer; S100A4 targeted by Sp-Met-1, explored for melanoma, colorectal cancer, and breast cancer; RPL14 targeted by Artenimol, used for treating uncomplicated plasmodium falciparum infections.
Real-world study
The results of the balance assessment after matching and the baseline characteristics of the population are provided in S40 and S41 Tables, respectively. Multivariable logistic regression analysis identified 277 proteins significantly associated with overall LC risk (109 protective/168 risk), 212 with LUSC (171 protective/41 risk), 118 with LUAD (65 protective/53 risk), and 169 with SCLC (103 protective/66 risk). Detailed information is provided in S42 Tables. Notably, RPL14 significantly increased overall LC risk (odds ratio [OR] = 1.509, 95% confidence interval [CI] = 1.105–2.244), while AGER reduced LUSC risk (OR = 0.366, 95%CI = 0.158–0.850), consistent with MR results (Fig 5).
(The graph only shows the results of protein-lung cancer associations, which are detailed in S42 Table. LC: lung cancer; LUSC: lung squamous carcinoma; LUAD: lung adenocarcinoma; SCLC: small cell lung cancer; OR: odds ratio; CI: confidence interval).
Discussion
This study conducted comprehensive MR analyses to explore causal associations between a total of 2,733 unique genetically predictive proteins from the UKB-PPP and deCODE study with LC and its three subtypes. Through meta-analyses, SMR tests, and colocalization analyses, cisMR-cML, MR.CUE and phenotype scanning analyses, we identified 12, 6, 4, and 3 genetically predicted proteins causally associated with susceptibility to overall LC, LUSC, LUAD, and SCLC, respectively. Our systematic review of previous MR studies on proteins and the risk of LC suggested that 15 of these protein-LC associations were considered newly identified. Additionally, 13 proteins—including six novel proteins (AGER, RPL14, S100A4, CEP43, MPI, and BRD2)—were identified as potential therapeutic targets. Further validation using population-level proteomic data from the UK Biobank confirmed a significant association between RPL14 and an increased risk of LC, as well as protein AGER and a reduced risk of LUSC.
The genetic architecture of different LC subtypes exhibits significant heterogeneity, with a growing body of research identifying novel subtype-specific genetic variants [31]. This heterogeneity significantly influences treatment strategies for distinct LC subtypes, particularly in the realm of pharmacogenomics [48]. Proteins, as the most widely utilized biomarkers for disease diagnosis, prognosis, and drug development, make the investigation of subtype-specific associated proteins imperative. However, most previous studies have predominantly focused on overall LC rather than comprehensively addressing individual subtypes. Thus, among the 10 protein-LC associations from our study that aligned with previous findings, the majority were linked to overall LC risk, including IL19, MICB_MICA, TSG101, and CTF1.
Among the novel potential therapeutic targets for LC identified in our study, the association of protein RPL14 with overall LC and AGER with LUSC has been robustly demonstrated across genetic and real-world population studies. RPL14 is a member of the L14E ribosomal protein family. The protein consists of five introns and six exons and is located on chromosome 3p21.3 [49]. At the genetic level, the heterozygosity of the RPL14 locus is 68%, while it is only 25% in non-SCLC (NSCLC) cell lines. This suggests that RPL14 may play a potential role in LC development due to the loss of heterozygosity in the mechanism of LC [50]. Additionally, RPL14 plays a role in the occurrence and progression of other cancers, such as esophageal squamous carcinoma [51], hepatocellular carcinoma [52], and cervical cancer [53]. AGER is a transmembrane pattern recognition receptor in the immunoglobulin superfamily [54]. It is predominantly expressed in the lungs, where its normal expression level plays a critical role in maintaining alveolar type 1 (AT1) cell differentiation, adhesion, spreading, and apoptosis. Inhibition of AGER expression in alveolar cells or its downregulation during malignant transformation in LC cells enhances migratory, invasive, and proliferative properties, thereby promoting LC development [54,55]. In vitro studies demonstrated that co-culturing AGER-depleted human LC cells (H358) with lung fibroblasts (WI38) significantly enhanced cancer cell proliferation [56]. Clinically, reduced AGER protein expression in LUSC correlates with poor prognosis, while higher levels are associated with longer survival [57,58]. Additionally, AGER interacts with multiple endogenous ligands (e.g., HMGB1, S100 proteins, calreticulin) implicated in metastasis and disease progression within the LC microenvironment [59]. PPI analysis further revealed that AGER interacts with S100A4, and previous evidence suggests that S100A4 inhibits autophagy in an AGER-dependent manner, thereby influencing LC development. Knol et al. used data-independent acquisition mass spectrometry to report The Pan-Cancer Proteome Atlas (TPCPA) and highlighted AGER as a key LC biomarker [60]. We also observed a significant inverse association between AGER and overall LC risk in the UKB cohort (OR = 0.688). Given its mechanistic involvement and significant protective association with LC and LUSC, AGER is a promising target in LC research and therapeutic development.
Protein S100-A4 (S100A4) was associated with a reduced risk of overall LC in the MR analysis and a decreased risk of LUSC in the population analysis, aligning with Smith-Byrne et al.’s reported protective effect (OR = 0.814) [61]. While S100A4 is a well-established epithelial-mesenchymal transition marker implicated in metastasis across multiple malignancies [62,63], emerging evidence reveals its complex, context-dependent roles in cancer biology. On the one hand, S100A4 promotes tumor progression by up-regulating mitochondrial complex I subunit NADH dehydrogenase (ubiquinone) Fe-S protein 2 to enhance invasion and metabolic reprogramming [64]. Clinically, high S100A4 expression is associated with poor differentiation and poor prognosis of NSCLC [65–67]. On the other hand, S100A4 induction has also been shown to reduce motility and invasiveness, for example, in squamous cell carcinoma [68]. Upregulation of S100A4 by P53 mutant (H1299) contributes to the restoration of the tumor suppressor activity of plakoglobin, a tumor/metastasis suppressor [69]. Downregulation of S100A4 in astrocytes increased their migration capacity in vitro [68]. These seemingly contradictory observations suggest a biphasic role for S100A4 in lung carcinogenesis - acting as a tumor progression inhibitor during early disease stages while facilitating metastatic dissemination in advanced tumors [70]. Mannose-6-phosphate (MPI) was also found to be associated with a reduced risk of LC in our study. Multiple types of tumor cells exhibit enhanced glucose uptake characteristics, MPI could influence neoplastic growth by concurrently inhibiting critical metabolic pathways - glycolysis, Tricarboxylic acid cycle, and glycoconjugate biosynthesis [71]. Notably, MPI has been shown to potentiate therapeutic responses in acute myeloid leukemia, enhancing cellular sensitivity to both conventional cytarabine chemotherapy and targeted FLT3 inhibitors [71]. Furthermore, Luan et al. discovered that MPI is associated with a reduced risk of NSCLC, suggesting its potential tumor-suppressive role in pulmonary malignancies [72]. Further studies are needed to precisely delineate their context-specific mechanisms in LC, which may provide novel insights for targeted therapeutic strategies.
In addition, MR evidence from this study also supports the associations of Interleukin 19 (IL19) and Plasma protease C1 inhibitor (SERPING1) with LC. IL19, a member of the IL10 cytokine family, is primarily produced by monocytes and exerts its effects via IL-20R1/IL-20R2 chain complexes [73]. Our study demonstrated a significant association between elevated IL19 levels and an increased risk of LC, consistent with the findings reported by Zhang [74] and Fang [75]. Furthermore, in vitro assays indicated that elevated IL19 levels induce apoptosis in lung epithelial cells and promote neutrophil chemotaxis [76]. Other studies have also linked high IL19 expression to a poorer prognosis in interstitial lung disease among patients with SCLC [77]. Additionally, elevated IL19 levels have been implicated in bone metastasis from LC and the progression of other respiratory diseases, such as COPD and pulmonary fibrosis [78–80]. SERPING1, a member of the serine protease inhibitor superfamily, plays a key role in regulating inflammation. Its deficiency can result in angioedema [81]. In the current MR study, protein SERPING1 was consistently and significantly negatively associated with overall LC risk, aligning with findings from previous MR studies [82]. The analysis based on TCGA data indicated that SERPING1 expression was downregulated in LUAD [83]. Supporting this, one study reported that increased SERPING1 levels are associated with a poor prognosis in LUSC [84]. The role of SERPING1 in LC remains ambiguous, as there is a paucity of studies examining its relationship with specific subtypes. Additional research is necessary to elucidate the role of SERPING1 as either a protective or risk factor and to identify its potential mechanisms of action in LC pathogenesis.
Ubiquitin carboxyl-terminal hydrolase 28 (USP28), a catalytically active deubiquitinating enzyme, regulates key biological processes such as cell proliferation, DNA damage repair, apoptosis, and tumorigenesis [85]. Previous studies have shown that USP28 modulates the SREBP2 and mevalonate pathways, promoting tumor growth in LUSC [86]. USP28 has been implicated in the oncogenic transformation of respiratory cells, and its overexpression is associated with LUSC progression. Our findings are consistent with these studies, supporting the conclusion that USP28 contributes to increased LC risk, particularly in LUSC [87,88]. This study also identified several proteins that are significantly associated with LUAD and SCLC. Among these, the triggering receptor expressed on myeloid cells 1 (TREM1) was significantly associated with a decreased risk of SCLC. This finding was corroborated by Song et al., who also reported the protective role of TREM1 in SCLC [89]. TREM1 is a myeloid cell surface receptor involved in the tumor microenvironment, where it amplifies the inflammatory response and exhibits antitumor effects. Nie et al. demonstrated the potential therapeutic applications of TREM1 by developing a multichain DAP12/TREM1 CAR (DT CAR) targeting DLL3, which showed considerable antitumor effects in SCLC cells [90]. Despite these promising findings, the role of TREM1 in SCLC remains unclear. Further biological and mechanistic studies are required to elucidate its functions and potential as a therapeutic target for SCLC.
This study has the following strengths. First, we performed tens of thousands of causal association estimates based on two large-scale proteomics data sources and two LC data sources, selecting proteins with the highest level of evidence. Second, the newly published UKB-PPP project reported that 85% of its principal genetic associations were novel discoveries, thereby substantially expanding the potential for MR analyses to identify innovative protein biomarkers associated with LC pathogenesis. In addition, the large sample population protein data of UKB-PPP ensures temporal causality while providing detailed histopathological subtype information, significantly increasing the validity of our real-world evidence. Furthermore, this study incorporated multiple LC subtypes in the analysis and, based on an evidence grading system, highlighted the proteins with the strongest causal associations for each subtype. In addition, we conducted cisMR-cML and MR.CUE analyses effectively avoided the effect of pleiotropy. Finally, by combining genetic (MR, SMR, and colocalization) and population evidence (UK Biobank), we identified novel LC-associated proteins, enhancing the etiological network of LC.
This study has several limitations. First, European-focused GWAS data limited sex/age-stratified analyses and cross-ethnic generalizability. Second, given the analysis of thousands of plasma proteins, multiple testing correction may increase false negative rates [91]. To address this, we implemented a two-tiered screening strategy for primary protein selection: (1) FDR < 0.05 threshold, and (2) requiring statistically significant associations (p < 0.05) in two independent databases. This dual-validation approach maintains statistical rigor while effectively balancing false-positive and false-negative risks, and also identifies more potential proteins. Third, in screening primary candidate proteins, the direction of effect sizes is inconsistent between the two datasets, which may potentially be affected by factors such as the geographic background of the study population, the gender ratio, and differences in GWAS research methods. Fourth, the meta-analysis used the same exposures, which may cause some error in the results, but a series of subsequent analyses, such as SMR and colocalization, were carried out to minimize their effect on the results and ensure their reliability. Fifth, we identified subtype-specific proteins whose potential association with overall LC risk cannot be entirely excluded given shared genetic architecture. Their clinical interpretation thus requires larger studies or integrated evidence to clarify the underlying mechanisms. Sixth, although mean imputation in the UK Biobank cohort may lead to slight effect size attenuation through reduced variance, the impact is considered limited statistically, given the low average missing rate (~11%) in the protein data. Seventh, when enrichment analyses were performed, multiple corrections were not used because the small set of genes tested made it difficult to detect significant enrichment results. Finally, while this evidence strengthens our understanding of potential causal associations, these findings should be interpreted cautiously and further validated through experimental studies and large-scale population-based investigations.
Supporting information
S1 Fig. Flowchart of study population selection in UK Biobank.
(LC: lung cancer; LUSC: lung squamous carcinoma; LUAD: lung adenocarcinoma; SCLC: small cell lung cancer; BMI: body mass index; TDI: Thompson Deprivation Index).
https://doi.org/10.1371/journal.pgen.1011821.s004
(S1_Fig.DOCX)
S1 Table. Systematic review of Mendelian randomization studies on proteomics and lung cancer.
https://doi.org/10.1371/journal.pgen.1011821.s005
(S1_Table.XLSX)
S2 Table. Reasons for exclusion from the review literature on Mendelian randomization of circulating proteins and lung cancer (updated to August 28, 2024).
https://doi.org/10.1371/journal.pgen.1011821.s006
(S2_Table.XLSX)
S3 Table. Protein information for the UKB-PPP project.
https://doi.org/10.1371/journal.pgen.1011821.s007
(S3_Table.XLSX)
S4 Table. Protein information for the deCODE study.
https://doi.org/10.1371/journal.pgen.1011821.s008
(S4_Table.XLSX)
S5 Table. Definitions of adjustment factors and outcomes.
https://doi.org/10.1371/journal.pgen.1011821.s009
(S5_Table.XLSX)
S6 Table. Instrument variables of plasma protein in cis-MR analysis (after harmonization, F-statistics > 10).
https://doi.org/10.1371/journal.pgen.1011821.s010
(S6_Table.XLSX)
S7 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) on lung cancer (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s011
(S7_Table.XLSX)
S8 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and lung squamous carcinoma (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s012
(S8_Table.XLSX)
S9 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and lung adenocarcinoma (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s013
(S9_Table.XLSX)
S10 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and small-cell lung cancer (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s014
(S10_Table.XLSX)
S11 Table. The Mendelian randomization results for plasma proteins (deCODE) and lung cancer (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s015
(S11_Table.XLSX)
S12 Table. The Mendelian randomization results for plasma proteins (deCODE) and lung squamous carcinoma (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s016
(S12_Table.XLSX)
S13 Table. The Mendelian randomization results for plasma proteins (deCODE) and lung adenocarcinoma (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s017
(S13_Table.XLSX)
S14 Table. The Mendelian randomization results for plasma proteins (deCODE) and small-cell lung cancer (FinnGen).
https://doi.org/10.1371/journal.pgen.1011821.s018
(S14_Table.XLSX)
S15 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and lung cancer (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s019
(S15_Table.XLSX)
S16 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and lung squamous carcinoma (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s020
(S16_Table.XLSX)
S17 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and lung adenocarcinoma (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s021
(S17_Table.XLSX)
S18 Table. The Mendelian randomization results for plasma proteins (UKB-PPP) and small-cell lung cancer (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s022
(S18_Table.XLSX)
S19 Table. The Mendelian randomization results for plasma proteins (deCODE) and lung cancer (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s023
(S19_Table.XLSX)
S20 Table. The Mendelian randomization results for plasma proteins (deCODE) and lung squamous carcinoma (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s024
(S20_Table.XLSX)
S21 Table. The Mendelian randomization results for plasma proteins (deCODE) and lung adenocarcinoma (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s025
(S21_Table.XLSX)
S22 Table. The Mendelian randomization results for plasma proteins (deCODE) and small-cell lung cancer (TRICL).
https://doi.org/10.1371/journal.pgen.1011821.s026
(S22_Table.XLSX)
S23 Table. The summary results of Mendelian randomization analyses for candidate lung cancer-associated proteins.
https://doi.org/10.1371/journal.pgen.1011821.s027
(S23_Table.XLSX)
S24 Table. Statistical power, heterogeneity, pleiotropy, and Steiger directionality tests of Mendelian randomization analyses.
https://doi.org/10.1371/journal.pgen.1011821.s028
(S24_Table.XLSX)
S25 Table. The meta-analysis of the Mendelian randomization results.
https://doi.org/10.1371/journal.pgen.1011821.s029
(S25_Table.XLSX)
S26 Table. The Mendelian randomization results for lung cancer and plasma proteins.
https://doi.org/10.1371/journal.pgen.1011821.s030
(S26_Table.XLSX)
S27 Table. The results of SMR and HEIDI tests between plasma proteins and lung cancer.
https://doi.org/10.1371/journal.pgen.1011821.s031
(S27_Table.XLSX)
S28 Table. The colocalization analyses between plasma proteins and lung cancer.
https://doi.org/10.1371/journal.pgen.1011821.s032
(S28_Table.XLSX)
S29 Table. The results of cisMR-cML analysis between plasma proteins and lung cancer.
https://doi.org/10.1371/journal.pgen.1011821.s033
(XLSX)
S30 Table. The results of MR.CUE analysis between plasma proteins and lung cancer.
https://doi.org/10.1371/journal.pgen.1011821.s034
(S30_Table.XLSX)
S31 Table. The results of phenotype scanning.
https://doi.org/10.1371/journal.pgen.1011821.s035
(S31_Table.XLSX)
S32 Table. Cis-MR results of protein oid20756 (AGER, removal of rs204993) and 4479_14_SERPING1_C1_Esterase_Inhibitor (SERPING1, removal of rs117960683) with lung cancer risk.
https://doi.org/10.1371/journal.pgen.1011821.s036
(S32_Table.XLSX)
S33 Table. The summary results of 25 significant lung cancer-associated proteins.
https://doi.org/10.1371/journal.pgen.1011821.s037
(S33_Table.XLSX)
S34 Table. Systematic review of Mendelian randomization studies on proteomics and lung cancer.
https://doi.org/10.1371/journal.pgen.1011821.s038
(S34_Table.XLSX)
S35 Table. Summary information from previous Mendelian randomization studies.
https://doi.org/10.1371/journal.pgen.1011821.s039
(S35_Table.XLSX)
S36 Table. Single-cell type-specific expression analysis.
https://doi.org/10.1371/journal.pgen.1011821.s040
(S36_Table.XLSX)
S37 Table. The pathway of significant lung cancer-associated proteins enrichment.
https://doi.org/10.1371/journal.pgen.1011821.s041
(S37_Table.XLSX)
S38 Table. Drug targets in DrugBank for lung cancer treatment.
https://doi.org/10.1371/journal.pgen.1011821.s042
(S38_Table.XLSX)
S39 Table. Druggability evaluation of significant proteins.
https://doi.org/10.1371/journal.pgen.1011821.s043
(S39_Table.XLSX)
S40 Table. Propensity-matched equilibrium test for the UK biobank cohort.
https://doi.org/10.1371/journal.pgen.1011821.s044
(S40_Table.XLSX)
S41 Table. Baseline characteristics of lung cancer cases and matched controls after propensity score matching.
https://doi.org/10.1371/journal.pgen.1011821.s045
(S41_Table.XLSX)
S42 Table. Multivariate conditional logistic regression results of circulating proteins and lung cancer in the UK Biobank cohort.
https://doi.org/10.1371/journal.pgen.1011821.s046
(S42_Table.XLSX)
Acknowledgments
The authors would like to acknowledge all the participants and investigators in the GWAS cohort in this study and thank the researchers at UKB-PPP, deCODE, FinnGen, TRICL, and GWAS Catalog for sharing the GWAS data, as well as the developers of the R packages and online tools used in this study.
References
- 1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229–63. pmid:38572751
- 2. Zhang Y, Vaccarella S, Morgan E, Li M, Etxeberria J, Chokunonga E, et al. Global variations in lung cancer incidence by histological subtype in 2020: a population-based study. Lancet Oncol. 2023;24(11):1206–18. pmid:37837979
- 3. Ganti AK, Klein AB, Cotarla I, Seal B, Chou E. Update of Incidence, Prevalence, Survival, and Initial Treatment in Patients With Non-Small Cell Lung Cancer in the US. JAMA Oncol. 2021;7(12):1824–32. pmid:34673888
- 4. Zeng H, Chen W, Zheng R, Zhang S, Ji JS, Zou X, et al. Changing cancer survival in China during 2003-15: a pooled analysis of 17 population-based cancer registries. Lancet Glob Health. 2018;6(5):e555–67. pmid:29653628
- 5. Allemani C, Matsuda T, Di Carlo V, Harewood R, Matz M, Nikšić M, et al. Global surveillance of trends in cancer survival 2000-14 (CONCORD-3): analysis of individual records for 37 513 025 patients diagnosed with one of 18 cancers from 322 population-based registries in 71 countries. Lancet. 2018;391(10125):1023–75. pmid:29395269
- 6. Pasello G, Scattolin D, Bonanno L, Caumo F, Dell’Amore A, Scagliori E, et al. Secondary prevention and treatment innovation of early stage non-small cell lung cancer: Impact on diagnostic-therapeutic pathway from a multidisciplinary perspective. Cancer Treat Rev. 2023;116:102544. pmid:36940657
- 7. Wu JT-Y, Wakelee HA, Han SS. Optimizing Lung Cancer Screening With Risk Prediction: Current Challenges and the Emerging Role of Biomarkers. J Clin Oncol. 2023;41(27):4341–7. pmid:37540816
- 8. Mino-Kenudson M, Schalper K, Cooper W, Dacic S, Hirsch FR, Jain D, et al. Predictive Biomarkers for Immunotherapy in Lung Cancer: Perspective From the International Association for the Study of Lung Cancer Pathology Committee. J Thorac Oncol. 2022;17(12):1335–54. pmid:36184066
- 9. Shi J, Shiraishi K, Choi J, Matsuo K, Chen T-Y, Dai J, et al. Genome-wide association study of lung adenocarcinoma in East Asia and comparison with a European population. Nat Commun. 2023;14(1):3043. pmid:37236969
- 10. Mattox AK, Douville C, Wang Y, Popoli M, Ptak J, Silliman N, et al. The Origin of Highly Elevated Cell-Free DNA in Healthy Individuals and Patients with Pancreatic, Colorectal, Lung, or Ovarian Cancer. Cancer Discov. 2023;13(10):2166–79. pmid:37565753
- 11. Heeke S, Gay CM, Estecio MR, Tran H, Morris BB, Zhang B, et al. Tumor- and circulating-free DNA methylation identifies clinically relevant small cell lung cancer subtypes. Cancer Cell. 2024;42(2):225-237.e5. pmid:38278149
- 12. Wu P, Zhang C, Tang X, Li D, Zhang G, Zi X, et al. Pan-cancer characterization of cell-free immune-related miRNA identified as a robust biomarker for cancer diagnosis. Mol Cancer. 2024;23(1):31. pmid:38347558
- 13. Feng X, Wu WY-Y, Onwuka JU, Haider Z, Alcala K, Smith-Byrne K, et al. Lung cancer risk discrimination of prediagnostic proteomics measurements compared with existing prediction tools. J Natl Cancer Inst. 2023;115(9):1050–9. pmid:37260165
- 14. An Autoantibody Subset Can Be Used for SCLC Early Detection. Cancer Discov. 2023;13(3):526. pmid:36661372
- 15. Yao Y, Wang X, Guan J, Xie C, Zhang H, Yang J, et al. Metabolomic differentiation of benign vs malignant pulmonary nodules with high specificity via high-resolution mass spectrometry analysis of patient sera. Nat Commun. 2023;14(1):2339. pmid:37095081
- 16. Ding Z, Wang N, Ji N, Chen Z-S. Proteomics technologies for cancer liquid biopsies. Mol Cancer. 2022;21(1):53. pmid:35168611
- 17. Abbasian MH, Ardekani AM, Sobhani N, Roudi R. The Role of Genomics and Proteomics in Lung Cancer Early Detection and Treatment. Cancers (Basel). 2022;14(20):5144. pmid:36291929
- 18. Elshoeibi AM, Elsayed B, Kaleem MZ, Elhadary MR, Abu-Haweeleh MN, Haithm Y, et al. Proteomic Profiling of Small-Cell Lung Cancer: A Systematic Review. Cancers (Basel). 2023;15(20):5005. pmid:37894372
- 19. Arsenault BJ. From the garden to the clinic: how Mendelian randomization is shaping up atherosclerotic cardiovascular disease prevention strategies. Eur Heart J. 2022;43(42):4447–9. pmid:35869924
- 20. Benusiglio PR, Fallet V, Sanchis-Borja M, Coulet F, Cadranel J. Lung cancer is also a hereditary disease. Eur Respir Rev. 2021;30(162):210045. pmid:34670806
- 21. Mucci LA, Hjelmborg JB, Harris JR, Czene K, Havelick DJ, Scheike T, et al. Familial Risk and Heritability of Cancer Among Twins in Nordic Countries. JAMA. 2016;315(1):68–76. pmid:26746459
- 22. Coté ML, Liu M, Bonassi S, Neri M, Schwartz AG, Christiani DC, et al. Increased risk of lung cancer in individuals with a family history of the disease: a pooled analysis from the International Lung Cancer Consortium. Eur J Cancer. 2012;48(13):1957–68. pmid:22436981
- 23. Sun J, Luo J, Jiang F, Zhao J, Zhou S, Wang L, et al. Exploring the cross-cancer effect of circulating proteins and discovering potential intervention targets for 13 site-specific cancers. J Natl Cancer Inst. 2024;116(4):565–73. pmid:38039160
- 24. Wang X, Wang T, Hua J, Cai M, Qian Z, Wang C, et al. Histological types of lung cancer attributable to fine particulate, smoking, and genetic susceptibility. Sci Total Environ. 2023;858(Pt 2):159890. pmid:36334679
- 25. Ilar A, Plato N, Lewné M, Pershagen G, Gustavsson P. Occupational exposure to diesel motor exhaust and risk of lung cancer by histological subtype: a population-based case-control study in Swedish men. Eur J Epidemiol. 2017;32(8):711–9. pmid:28585123
- 26. Ji T, Lv Y, Liu M, Han Y, Yuan B, Gu J. Causal relationships between mitochondrial proteins and different pathological types of lung cancer: a bidirectional mendelian randomization study. Front Genet. 2024;15:1335223. pmid:38596213
- 27. Ren F, Jin Q, Liu T, Ren X, Zhan Y. Proteome-wide mendelian randomization study implicates therapeutic targets in common cancers. J Transl Med. 2023;21(1):646. pmid:37735436
- 28. Sun BB, Chiou J, Traylor M, Benner C, Hsu Y-H, Richardson TG, et al. Plasma proteomic associations with genetics and health in the UK Biobank. Nature. 2023;622(7982):329–38. pmid:37794186
- 29. Ferkingstad E, Sulem P, Atlason BA, Sveinbjornsson G, Magnusson MI, Styrmisdottir EL, et al. Large-scale integration of the plasma proteome with genetics and disease. Nat Genet. 2021;53(12):1712–21. pmid:34857953
- 30. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613(7944):508–18. pmid:36653562
- 31. McKay JD, Hung RJ, Han Y, Zong X, Carreras-Torres R, Christiani DC, et al. Large-scale association analysis identifies new lung cancer susceptibility loci and heterogeneity in genetic susceptibility across histological subtypes. Nat Genet. 2017;49(7):1126–32. pmid:28604730
- 32. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12(3):e1001779. pmid:25826379
- 33. Hartwig FP, Davey Smith G, Bowden J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol. 2017;46(6):1985–98. pmid:29040600
- 34. Zhu Z, Zhang F, Hu H, Bakshi A, Robinson MR, Powell JE, et al. Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat Genet. 2016;48(5):481–7. pmid:27019110
- 35. Wu Y, Zeng J, Zhang F, Zhu Z, Qi T, Zheng Z, et al. Integrative analysis of omics summary data reveals putative mechanisms underlying complex traits. Nat Commun. 2018;9(1):918. pmid:29500431
- 36. Lin Z, Pan W. A robust cis-Mendelian randomization method with application to drug target discovery. Nat Commun. 2024;15(1):6072. pmid:39025905
- 37. Yang J, Ferreira T, Morris AP, Medland SE, Genetic Investigation of ANthropometric Traits (GIANT) Consortium, DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium, et al. Conditional and joint multiple-SNP analysis of GWAS summary statistics identifies additional variants influencing complex traits. Nat Genet. 2012;44(4):369–75, S1-3. pmid:22426310
- 38. Cheng Q, Zhang X, Chen LS, Liu J. Mendelian randomization accounting for complex correlated horizontal pleiotropy while elucidating shared genetic etiology. Nat Commun. 2022;13(1):6490. pmid:36310177
- 39. Sollis E, Mosaku A, Abid A, Buniello A, Cerezo M, Gil L, et al. The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Res. 2023;51(D1):D977–85. pmid:36350656
- 40. Pontén F, Schwenk JM, Asplund A, Edqvist P-HD. The Human Protein Atlas as a proteomic resource for biomarker discovery. J Intern Med. 2011;270(5):428–46. pmid:21752111
- 41. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10(1):1523. pmid:30944313
- 42. Knox C, Wilson M, Klinger CM, Franklin M, Oler E, Wilson A, et al. DrugBank 6.0: the DrugBank Knowledgebase for 2024. Nucleic Acids Res. 2024;52(D1):D1265–75. pmid:37953279
- 43. Zhou Y, Zhang Y, Zhao D, Yu X, Shen X, Zhou Y, et al. TTD: Therapeutic Target Database describing target druggability information. Nucleic Acids Res. 2024;52(D1):D1465–77. pmid:37713619
- 44. Zdrazil B, Felix E, Hunter F, Manners EJ, Blackshaw J, Corbett S, et al. The ChEMBL Database in 2023: a drug discovery platform spanning multiple bioactivity data types and time periods. Nucleic Acids Res. 2024;52(D1):D1180–92. pmid:37933841
- 45. Freshour SL, Kiwala S, Cotto KC, Coffman AC, McMichael JF, Song JJ, et al. Integration of the Drug-Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Res. 2021;49(D1):D1144–51. pmid:33237278
- 46. GBD 2019 Respiratory Tract Cancers Collaborators. Global, regional, and national burden of respiratory tract cancers and associated risk factors from 1990 to 2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Respir Med. 2021;9(9):1030–49. pmid:34411511
- 47. Zhu M, Wang T, Huang Y, Zhao X, Ding Y, Zhu M, et al. Genetic Risk for Overall Cancer and the Benefit of Adherence to a Healthy Lifestyle. Cancer Res. 2021;81(17):4618–27. pmid:34321244
- 48. Long E, Patel H, Golden A, Antony M, Yin J, Funderburk K, et al. High-throughput characterization of functional variants highlights heterogeneity and polygenicity underlying lung cancer susceptibility. Am J Hum Genet. 2024;111(7):1405–19. pmid:38906146
- 49. Zhang Z, Zhang Y, Qiu Y, Mo W, Yang Z. Human/eukaryotic ribosomal protein L14 (RPL14/eL14) overexpression represses proliferation, migration, invasion and EMT process in nasopharyngeal carcinoma. Bioengineered. 2021;12(1):2175–86. pmid:34057029
- 50. Shriver SP, Shriver MD, Tirpak DL, Bloch LM, Hunt JD, Ferrell RE, et al. Trinucleotide repeat length variation in the human ribosomal protein L14 gene (RPL14): localization to 3p21.3 and loss of heterozygosity in lung and oral cancers. Mutat Res. 1998;406(1):9–23. pmid:9920051
- 51. Huang X-P, Zhao C-X, Li Q-J, Cai Y, Liu F-X, Hu H, et al. Alteration of RPL14 in squamous cell carcinomas and preneoplastic lesions of the esophagus. Gene. 2006;366(1):161–8. pmid:16316724
- 52. Liu Y, Yu X, Wang Y, Wu J, Feng B, Li M. The role of differentially expressed genes and immune cell infiltration in the progression of nonalcoholic steatohepatitis (NASH) to hepatocellular carcinoma (HCC): a new exploration based on bioinformatics analysis. Nucleosides Nucleotides Nucleic Acids. 2024;43(12):1415–30. pmid:38319987
- 53. Feng Y, Ma J, Fan H, Liu M, Zhu Y, Li Y, et al. TNF-α-induced lncRNA LOC105374902 promotes the malignant behavior of cervical cancer cells by acting as a sponge of miR-1285-3p. Biochem Biophys Res Commun. 2019;513(1):56–63. pmid:30935691
- 54. Marinakis E, Bagkos G, Piperi C, Roussou P, Diamanti-Kandarakis E. Critical role of RAGE in lung physiology and tumorigenesis: a potential target of therapeutic intervention? Clin Chem Lab Med. 2014;52(2):189–200. pmid:24108211
- 55. Bartling B, Hofmann H-S, Weigle B, Silber R-E, Simm A. Down-regulation of the receptor for advanced glycation end-products (RAGE) supports non-small cell lung carcinoma. Carcinogenesis. 2005;26(2):293–301. pmid:15539404
- 56. Bartling B, Demling N, Silber R-E, Simm A. Proliferative stimulus of lung fibroblasts on lung cancer cells is impaired by the receptor for advanced glycation end-products. Am J Respir Cell Mol Biol. 2006;34(1):83–91. pmid:16166741
- 57. Yang J, Lin M, Zhang M, Wang Z, Lin H, Yu Y, et al. Advanced Glycation End Products’ Receptor DNA Methylation Associated with Immune Infiltration and Prognosis of Lung Adenocarcinoma and Lung Squamous Cell Carcinoma. Genet Res (Camb). 2023;2023:7129325. pmid:37497166
- 58. Yu YX, Pan WC, Cheng YF. Silencing of advanced glycosylation and glycosylation and product-specific receptor (RAGE) inhibits the metastasis and growth of non-small cell lung cancer. Am J Transl Res. 2017;9(6):2760–74. pmid:28670367
- 59. Santilli F, Vazzana N, Bucciarelli LG, Davì G. Soluble forms of RAGE in human diseases: clinical and therapeutical implications. Curr Med Chem. 2009;16(8):940–52. pmid:19275604
- 60. Knol JC, Lyu M, Böttger F, Nunes Monteiro M, Pham TV, Rolfs F, et al. The pan-cancer proteome atlas, a mass spectrometry-based landscape for discovering tumor biology, biomarkers, and therapeutic targets. Cancer Cell. 2025;43(7):1328-1346.e8. pmid:40446800
- 61. Smith-Byrne K, Hedman Å, Dimitriou M, Desai T, Sokolov AV, Schioth HB, et al. Identifying therapeutic targets for cancer among 2074 circulating proteins and risk of nine cancers. Nat Commun. 2024;15(1):3621. pmid:38684708
- 62. Ning Q, Li F, Wang L, Li H, Yao Y, Hu T, et al. S100A4 amplifies TGF-β-induced epithelial-mesenchymal transition in a pleural mesothelial cell line. J Investig Med. 2018;66(2):334–9. pmid:29141874
- 63. Hua T, Liu S, Xin X, Cai L, Shi R, Chi S, et al. S100A4 promotes endometrial cancer progress through epithelial-mesenchymal transition regulation. Oncol Rep. 2016;35(6):3419–26. pmid:27109209
- 64. Liu L, Qi L, Knifley T, Piecoro DW, Rychahou P, Liu J, et al. S100A4 alters metabolism and promotes invasion of lung cancer cells by up-regulating mitochondrial complex I protein NDUFS2. J Biol Chem. 2019;294(18):7516–27. pmid:30885944
- 65. Wu X, Zhang H, Jiang G, Peng M, Li C, Lu J, et al. Exosome-transmitted S100A4 induces immunosuppression and non-small cell lung cancer development by activating STAT3. Clin Exp Immunol. 2022;210(3):309–20. pmid:36370151
- 66. Tsuna M, Kageyama S-I, Fukuoka J, Kitano H, Doki Y, Tezuka H, et al. Significance of S100A4 as a prognostic marker of lung squamous cell carcinoma. Anticancer Res. 2009;29(7):2547–54. pmid:19596927
- 67. Zhang J, Gu Y, Liu X, Rao X, Huang G, Ouyang Y. Clinicopathological and prognostic value of S100A4 expression in non-small cell lung cancer: a meta-analysis. Biosci Rep. 2020;40(7):BSR20201710. pmid:32696952
- 68. Fang Z, Duthoit N, Wicher G, Källskog O, Ambartsumian N, Lukanidin E, et al. Intracellular calcium-binding protein S100A4 influences injury-induced migration of white matter astrocytes. Acta Neuropathol. 2006;111(3):213–9. pmid:16463066
- 69. Alaee M, Nool K, Pasdar M. Plakoglobin restores tumor suppressor activity of p53R175H mutant by sequestering the oncogenic potential of β-catenin. Cancer Sci. 2018;109(6):1876–88. pmid:29660231
- 70. Kongsgaard A, Lund-Iversen M, Berge G, Brustugun OT, Solberg SK, Mælandsmo GM, et al. Expression of S100A4, ephrin-A1 and osteopontin in non-small cell lung cancer. BMC Cancer. 2012;12:333. pmid:22853000
- 71. Gonzalez PS, O’Prey J, Cardaci S, Barthet VJA, Sakamaki J-I, Beaumatin F, et al. Mannose impairs tumour growth and enhances chemotherapy. Nature. 2018;563(7733):719–23. pmid:30464341
- 72. Luan Y, Xian D, Zhao C, Qing X, He H, Zheng K, et al. Therapeutic targets for lung cancer: genome-wide Mendelian randomization and colocalization analyses. Front Pharmacol. 2024;15:1441233. pmid:39529882
- 73. Sabat R, Wallace E, Endesfelder S, Wolk K. IL-19 and IL-20: two novel cytokines with importance in inflammatory diseases. Expert Opin Ther Targets. 2007;11(5):601–12. pmid:17465720
- 74. Zhang Z, Wang J, Teng M, Yan X, Liu Q. The role of serum interleukins in Cancer: A Multi-center Mendelian Randomization study. Int Immunopharmacol. 2024;137:112520. pmid:38901247
- 75. Fang X, Liu D, Zhao J, Li X, He T, Liu B. Using proteomics and metabolomics to identify therapeutic targets for senescence mediated cancer: genetic complementarity method. Front Endocrinol (Lausanne). 2023;14:1255889. pmid:37745724
- 76. Hsing C-H, Chiu C-J, Chang L-Y, Hsu C-C, Chang M-S. IL-19 is involved in the pathogenesis of endotoxic shock. Shock. 2008;29(1):7–15. pmid:18246602
- 77. Murata D, Azuma K, Murotani K, Matsuo N, Matama G, Tokito T, et al. Survival and soluble immune mediators of immune checkpoint inhibitor-induced interstitial lung disease in patients with non-small cell lung cancer. Lung Cancer. 2023;184:107351. pmid:37639819
- 78. He Y, Luo W, Liu Y, Wang Y, Ma C, Wu Q, et al. IL-20RB mediates tumoral response to osteoclastic niches and promotes bone metastasis of lung cancer. J Clin Invest. 2022;132(20):e157917. pmid:36006737
- 79. Rong B, Liu Y, Li M, Fu T, Gao W, Liu H. Correlation of serum levels of HIF-1α and IL-19 with the disease progression of COPD: a retrospective study. Int J Chron Obstruct Pulmon Dis. 2018;13:3791–803. pmid:30538441
- 80. Wang Y, Sun S, Wang K, Zhang M, Li M, Zan Y, et al. Interleukin-19 Aggravates Pulmonary Fibrosis via Activating Fibroblast through TGF-β/Smad Pathway. Mediators Inflamm. 2022;2022:6755407. pmid:35281428
- 81. Kajdácsi E, Jandrasics Z, Veszeli N, Makó V, Koncz A, Gulyás D, et al. Patterns of C1-Inhibitor/Plasma Serine Protease Complexes in Healthy Humans and in Hereditary Angioedema Patients. Front Immunol. 2020;11:794. pmid:32431708
- 82. Yang C, Fagan AM, Perrin RJ, Rhinn H, Harari O, Cruchaga C. Mendelian randomization and genetic colocalization infer the effects of the multi-tissue proteome on 211 complex disease-related phenotypes. Genome Med. 2022;14(1):140. pmid:36510323
- 83. Zhang K, Wang K, Zhang C, Teng X, Li D, Chen M. Exploring the potential mechanism of emetine against coronavirus disease 2019 combined with lung adenocarcinoma: bioinformatics and molecular simulation analyses. BMC Cancer. 2022;22(1):687. pmid:35733175
- 84. O-charoenrat P, Rusch V, Talbot SG, Sarkaria I, Viale A, Socci N, et al. Casein kinase II alpha subunit and C1-inhibitor are independent predictors of outcome in patients with squamous cell carcinoma of the lung. Clin Cancer Res. 2004;10(17):5792–803. pmid:15355908
- 85. Prieto-Garcia C, Tomašković I, Shah VJ, Dikic I, Diefenbacher M. USP28: Oncogene or Tumor Suppressor? A Unifying Paradigm for Squamous Cell Carcinoma. Cells. 2021;10(10):2652. pmid:34685632
- 86. Maier CR, Hartmann O, Prieto-Garcia C, Al-Shami KM, Schlicker L, Vogel FCE, et al. USP28 controls SREBP2 and the mevalonate pathway to drive tumour growth in squamous cancer. Cell Death Differ. 2023;30(7):1710–25. pmid:37202505
- 87. Zhang L, Xu B, Qiang Y, Huang H, Wang C, Li D, et al. Overexpression of deubiquitinating enzyme USP28 promoted non-small cell lung cancer growth. J Cell Mol Med. 2015;19(4):799–805. pmid:25656529
- 88. Prieto-Garcia C, Hartmann O, Reissland M, Braun F, Bozkurt S, Pahor N, et al. USP28 enables oncogenic transformation of respiratory cells, and its inhibition potentiates molecular therapy targeting mutant EGFR, BRAF and PI3K. Mol Oncol. 2022;16(17):3082–106. pmid:35364627
- 89. Song N, Shi P, Cui K, Zeng L, Wang Z, Di W, et al. Potential drug targets for tumors identified through Mendelian randomization analysis. Sci Rep. 2024;14(1):11370. pmid:38762700
- 90. Nie F, Chen Y, Hu Y, Huang P, Shi X, Cai J, et al. TREM1/DAP12 based novel multiple chain CAR-T cells targeting DLL3 show robust anti-tumour efficacy for small cell lung cancer. Immunology. 2024;172(3):362–74. pmid:38469682
- 91. Mazidi M, Wright N, Yao P, Kartsonaki C, Millwood IY, Fry H, et al. Plasma Proteomics to Identify Drug Targets for Ischemic Heart Disease. J Am Coll Cardiol. 2023;82(20):1906–20. pmid:37940228