Proteome- and phenome-wide studies identify circulating protein drivers of kidney diseases.
Abstract
INTRODUCTION Proteome-wide Mendelian randomization (PWMR) provides a genetic framework to identify circulating proteins causally implicated in chronic kidney disease (CKD) and prioritize them for drug discovery. However, the diverse etiologies of CKD and clinical presentations make it difficult to translate proteomic findings into actionable targets. Here, we studied protein associations with longitudinal kidney outcomes, CKD subtypes, and phenome-wide traits to identify proteins that are causally relevant and pharmacologically actionable.
Methods
We performed a PWMR study using cis-pQTLs as genetic instruments for 2807 circulating proteins from four proteogenomic cohorts (ARIC, Fenland, deCODE, and UK Biobank) and tested their effects across four kidney outcomes (eGFR, incident CKD, incident kidney failure, annualized eGFR slope) in up to 464,631 of Million Veteran Program participants. Proteins meeting proteome-wide significance underwent sensitivity testing, Bayesian colocalization, and external replication in CKDGen Consortium information. The same instruments were re-tested against nine diagnosis code-defined CKD subtypes using genome wide association studies meta-analysis across the Million Veteran Program participants, UK Biobank, and FinnGen, and against 1020 phecodes to distinguish kidney-restricted from systemic effects. Druggability was evaluated using ChEMBL annotations layered onto high-confidence STRING protein-protein interaction networks.
Results
We identified 93 proteins with proteome-wide significance, and twelve remained robust across sensitivity testing (ANGPTL3, FAM213A, GCKR, GSTA1, IDI2, INHBA, INHBC, MANBA, MICB, MST1, NRG4, UMOD). Forty-seven proteins mapped to at least one of nine kidney disease subtypes, with distinct patterns across diabetic, hypertensive, and glomerular kidney diseases. Phenome-wide testing separated seven proteins with kidney-restricted associations from those with broader systemic effects. Thirty-four proteins were pharmacologic targets, nominating candidates such as IL-1 pathway blockade for diabetic kidney disease and ANGPTL3 inhibition for podocyte protection.
Conclusions
Our integrative PWMR framework resolves circulating protein targets across CKD etiologies, separates kidney-restricted from systemic effects, and delivers actionable targets that may accelerate therapeutic innovation for CKD. About one-third of the 93 proteins could be targeted with existing or investigational medications.