The relationship between dietary patterns and premature coronary heart disease (CHD), as well as the underlying proteomic mechanisms, remains incompletely understood. Using data from the Global Burden of Disease (GBD) 2021 study, we calculated the population attributable fraction (PAF) to derive weights for dietary factors and constructed a dietary score specific to premature CHD. This score was evaluated in 14 134 UK Biobank participants with proteomics data. Protein associations with the dietary score and incident premature CHD were assessed using multivariable linear regression and Cox proportional hazards models under strict dual-filtering criteria. Statistical mediation analysis was performed to explore whether proteins may partially account for the observed diet-CHD association. A higher GBD dietary score was associated with a reduced risk of premature CHD. Individuals in the highest quintile (Q5) of the dietary score exhibited a 47.0% lower risk compared with those in the lowest quintile (Q1) (HR: 0.530; 95% CI: 0.350-0.803). From 134 robust proteins identified via stringent intersection analysis as associated with both diet and premature CHD, 133 showed statistical evidence of mediation, with proportions ranging from -3.8% to 27.6%. A higher GBD dietary score is associated with a reduced risk of premature CHD, with proteins potentially mediating this relationship.