Site-Resolved Plasma-Protein Architecture of Infection Susceptibility: A Cis-pQTL Mendelian Randomization and Colocalization Study.

Background/Objectives: Infection susceptibility and systemic infection severity arise from distinct tissue barriers and host-response compartments, but plasma-protein genetic studies often treat infection as a broad phenotype. We aimed to define a site-resolved host plasma-protein architecture of infection susceptibility across urinary, lower-respiratory, systemic, and skin/soft-tissue infection outcomes. Methods: We performed proteome-wide two-sample cis-protein quantitative trait locus (cis-pQTL) Mendelian randomization (MR) using UK Biobank Pharma Proteomics Project (UKB-PPP) instruments and registry-derived FinnGen R13 infection endpoints. Prioritized protein-outcome pairs underwent regional colocalization, prior-sensitivity analysis, Steiger-like directionality sensitivity, Finnish and broader European reference-panel linkage disequilibrium (LD) assessment, candidate-level deCODE pQTL-source follow-up where feasible, and druggability and safety-domain annotation. Results: Across 18,766 harmonized tests, evidence was strongly site-resolved. PSCA emerged as the principal novel cystitis-specific candidate (OR 0.977, 95% CI 0.967-0.987), with strong colocalization and high Finnish reference-panel linkage disequilibrium between the MR instrument and the top colocalization single-nucleotide polymorphism (SNP) (r2 = 0.980). UMOD anchored urinary host-defense coherence. APOE showed strong evidence for the registry-derived implicit sepsis endpoint but was interpreted as a pleiotropy-prone broad host-response or severity-associated locus. TNFSF8 retained a robust pneumonia MR signal, but colocalization was moderate (PP.H4 = 0.544) and became weak under stringent priors (PP.H4 = 0.107), despite high-LD deCODE proxy-instrument follow-up. Skin/soft-tissue infection lacked a convergent lead. Conclusions: This study supports a site-resolved architecture of genetically anchored plasma-protein candidates for infection susceptibility. PSCA and UMOD define complementary urinary signals, whereas APOE and TNFSF8 require cautious interpretation as broad host-response and lower-respiratory follow-up hypotheses rather than established therapeutic targets.