**Background:** Both depression and breast cancer (BC) contribute substantially to global morbidity and mortality among women. Prior observational studies have reported inconsistent associations between depression and BC risk, potentially due to confounding and reverse causality. This study aimed to comprehensively characterize both the phenotypic and genetic relationships between depression and BC using the largest available datasets and advanced statistical genetic methods.
**Methods:** The observational analysis used longitudinal follow-up data from the UK Biobank, including 250,294 women of European descent (after excluding 6,856 with baseline BC). Depression was defined using ICD-10 codes (F32, F33, F34, F38, F39) and BC using ICD-10 code C50 and ICD-9 code 174. Cox proportional hazards models with time-dependent depression exposure were adjusted in three stages: Model 1 (age, assessment center, 40 genetic PCs); Model 2 (plus income, Townsend deprivation index, BMI, smoking, drinking, physical activity, sleep duration, education); Model 3 (plus family history of BC, reproductive factors, other mental health diagnoses, antidepressant/antipsychotic medication). For genetic analyses, the study leveraged GWAS summary statistics from the largest available studies: depression (N = 500,199, meta-analysis of UKBB, 23andMe, and PGC), overall BC (N = 247,173 from BCAC and 11 other studies), and ER+ BC (N = 175,475) and ER− BC (N = 127,442). Global genetic correlation was estimated using cross-trait LD Score Regression (LDSC). Local genetic correlation was assessed using ρ-HESS and GWAS-PW across 1,703 LD-independent regions. Cross-trait meta-analysis (CPASSOC) identified pleiotropic loci. Fine-mapping (FM-summary), colocalization (Coloc), and transcriptome-wide association studies (TWAS using FUSION with GTEx v8 data across 49 tissues) were performed. Bidirectional two-sample Mendelian randomization (MR) used inverse-variance weighted (IVW) as primary analysis, with MR-Egger, weighted median, and MR-PRESSO as sensitivity analyses, plus multivariable MR adjusting for BMI, smoking, alcohol, physical activity, sleep duration, and educational attainment.
**Key Results:** Over 3,014,168 person-years of follow-up (mean 11.41 ± 2.95 years), 529 depression patients and 9,516 depression-free individuals developed BC. The fully adjusted observational model showed HR = 1.096 (95% CI 0.951–1.264). Global genetic correlation was significant for depression with overall BC (rg = 0.08, P = 3.00 × 10⁻⁴), ER+ BC (rg = 0.06, P = 6.30 × 10⁻³), and ER− BC (rg = 0.08, P = 7.20 × 10⁻³). Local genetic correlation identified one significant region at 6p22.1 (ZSCAN12) for overall BC (P_ρ-HESS = 1.83 × 10⁻⁵) and four regions for ER+ BC, including 6p22.2 (ABT1, P_ρ-HESS = 7.30 × 10⁻⁶, PPA_3 = 0.73), 6p22.1 (PPA_3 = 0.71), 6p22.3–22.2 (PPA_3 = 0.89), and 9q31.2 (KLF4, PPA_3 = 0.52). Cross-trait meta-analysis identified 17 pleiotropic loci (12 for overall BC, 9 for ER+ BC), including one novel locus at 14q32.32 (rs56101042 near snoU13 and TRAF3, P_CPASSOC = 6.53 × 10⁻⁹). The strongest shared signal was rs2403907 at 21q21.1 (P_CPASSOC = 1.55 × 10⁻³³ for overall BC; 1.91 × 10⁻³² for ER+ BC). Colocalization analysis showed 5 of 12 loci for overall BC and 5 of 9 for ER+ BC shared causal variants (8 with PPH4 > 0.90). TWAS identified 5 shared genes (FLOT1, HLA-S, ENSG00000247934.4, ZSWIM1, GRAP2) enriched in brain, blood, uterus, colon, artery, and heart tissues. MR analysis demonstrated a causal effect of genetic liability to depression on overall BC risk (IVW OR = 1.12, 95% CI 1.04–1.19, P = 1.40 × 10⁻³), directionally consistent in MR-Egger and weighted median, with no horizontal pleiotropy (MR-Egger intercept P = 0.86). MR-PRESSO confirmed the result (OR = 1.13, 95% CI 1.06–1.21, P = 5.00 × 10⁻⁴). Subtype analyses showed non-significant associations for ER+ BC (IVW OR = 1.08, 95% CI 0.99–1.18, P = 6.98 × 10⁻²) and suggestive for ER− BC (IVW OR = 1.12, 95% CI 1.01–1.24, P = 3.77 × 10⁻²). No evidence of reverse causality was found (BC on depression: IVW OR = 1.01, P = 0.49).
**Clinical Implications:** This study provides the most comprehensive evidence to date that depression is associated with a modest increase in BC risk, supported by both observational and genetic analyses. The identified shared genetic architecture and putative causal relationship suggest that biological mechanisms—particularly involving immune, inflammatory, and stress-response pathways—may underlie the depression-BC link. These findings highlight the potential importance of mental health management in BC prevention strategies and suggest that women with depression may represent a target population for enhanced BC screening. The novel pleiotropic loci and shared genes identified may inform future research into therapeutic targets and biomarkers. However, the modest effect sizes and the restriction to European-ancestry populations limit generalizability, and further studies in diverse populations and experimental validation are needed.