Non-small cell lung cancer (NSCLC) remains the leading global cause of cancer-related mortality, with emerging evidence implicating lipid metabolism and immune dysregulation in its pathogenesis. While previous studies have established associations between lipid alterations and established tumors, the critical question of whether pre-diagnostic lipidome influences NSCLC development through immune modulation remains unresolved. This Mendelian randomization study aimed to: (1) establish causal relationships between plasma lipid species and NSCLC subtypes (adenocarcinoma (LUAD), squamous cell carcinoma (LUSC)), (2) identify mediating immune cell traits, and (3) characterize structural features of carcinogenic lipids. The investigation addresses a fundamental gap in understanding early mechanisms of NSCLC development, with potential implications for early detection and immunometabolic therapeutics.

We conducted a comprehensive two-sample Mendelian randomization (MR) analysis integrating genome-wide association study (GWAS) data from three European cohorts: (1) Lipidome data comprising 179 species across 4 classes (glycerolipids, glycerophospholipids, sphingolipids, sterols) from 7,174 Finnish individuals; (2) 731 immune cell phenotypes from 3,757 Sardinians; and (3) NSCLC outcome data (5,820 cases/345,118 controls) from the FinnGen consortium. Instrumental variables were selected using stringent criteria (p<5×10-8 for lipids, p<1×10-5 for immune traits; LD r2<0.001), with F-statistics >10 ensuring robust instruments.

Primary analyses employed inverse-variance weighted (IVW) regression for multi-SNP exposures and Wald ratios for single-SNP associations. Mediation analysis quantified immune-mediated effects using multivariable MR and product-of-coefficients methods. Bayesian colocalization (PP4>0.8 threshold) validated shared causal variants, while sensitivity analyses (MR-Egger, MR-PRESSO, leave-one-out) addressed pleiotropy and heterogeneity. Reverse MR (p<5×10-6 for LUSC, p<5×10-8 for NSCLC) was utilised to exclude potential reverse causality. All p-values were FDR-corrected (PFDR<0.05 significance threshold).

1. Lipidome effects on NSCLC

Twenty-seven lipid species showed causal associations with NSCLC (PFDR < 0.05), including 11 specific to LUSC. Risk-enhancing lipids contained polyunsaturated fatty acids (PUFAs) with ≥4 double bonds, such as phosphatidylcholine(17:0_20:4) (OR = 1.15, PFDR = 1.25×10⁻³), while protective lipids featured ≤3 double bonds (e.g., phosphatidylethanolamine(O-18:2_18:2): OR = 0.54, PFDR = 1.44×10⁻⁶). Sterol esters with long-chain PUFAs (e.g., sterol ester(27:1/20:5): OR = 1.19, PFDR = 2.19×10⁻³) exhibited strong oncogenic effects. No significant associations were observed for LUAD. Reverse MR excluded reverse causality (p > 0.05 for all significant lipids).

2. Colocalization evidence

Phosphatidylcholine (16:0_18:3) had strong evidence for colocalization (PP.H4 = 0.83) with LUSC at SNV rs1260326 , which is a coding sequence variant for the glucokinase regulator (GCKR) gene.  Phosphatidylethanolamine (18:2_0:0) demonstrated moderate colocalization evidence with both non-small cell lung cancer (NSCLC; PP.H4 = 0.57) and LUSC (PP.H4 = 0.54) at SNV rs174560, situated in the Fatty Acid Desaturase Ⅰ(FADS1)  gene.

3. Immune cell contributions

Thirty-two immune traits were causally linked to LUSC (PFDR < 0.05), including CD28+ CD45RA- CD8+ T cells (%T cell: OR = 0.96, PFDR = 0.017) and CD86+ plasmacytoid dendritic cells (%DC: OR = 0.86, PFDR = 0.003). For NSCLC, 25 immune phenotypes showed causal effects, such as terminally differentiated CD4+ T cells (%CD4+: OR = 0.91, PFDR = 0.008) and CD14+ monocytes (OR = 1.09, PFDR = 0.017).

4. Mediation by immune cells

Multivariable Mendelian randomization (MVMR)  adjusting for lipidomic effects revealed 8 immune traits-NSCLC and 5 immune traits-LUSC associations retaining statistically significant causality (P<0.05). Among these, three mediator-outcome pairs demonstrated statistically significant mediation effects (P<0.05). 

Phosphatidylcholine (18:2_20:3) exhibited a 12.02% (95%CI:0.56%-23.48%) mediation proportion through CD86+ plasmacytoid dendritic cells in LUSC pathogenesis (mediation effect: -0.0612, P=0.0397). CD33 expression on CD66b++ myeloid cells mediated 10.78% (0.06%-21.51%) of Phosphatidylethanolamine (O-18:2_18:2)'s effect on LUSC risk (mediation effect: -0.0717, P=0.0487). Phosphatidylethanolamine (18:2_0:0) showed borderline significant mediation via CD86+ plasmacytoid dendritic cells in LUSC (mediation effect: -0.0303, P=0.0500).

This study establishes causal links between specific lipid species, immune cell traits, and NSCLC pathogenesis. The identification of PUFA-enriched phospholipids as risk factors and their mediation through immune phenotypes provides mechanistic clarity to epidemiological associations.