Mendelian randomization across ~325K individuals flagged two candidate drug targets for ED, both expressed in penile endothelial and smooth muscle cells
Journal: Andrology | Published: 2025-06-15 | Type: Journal Article | PMID: 40518741 Authors: Xiang Bo-Yu et al. — Department of Urology and Neurosurgery, Xiangya Hospital, Central South University, Changsha, China Funding/COI: National Clinical Research Fund for Geriatric Diseases Research Center. No COI listed.
This is a computational fishing expedition, not a clinical trial — the authors used Mendelian randomization (MR) to mine genetic data for proteins that might make useful drug targets in ED, motivated by the fact that a meaningful fraction of ED patients don't respond to PDE5 inhibitors. They identified two candidates that replicated across two independent GWAS cohorts and survived co-localization filtering. Both targets are expressed primarily in endothelial cells and smooth muscle cells, which are the cell types that actually govern erection physiology. What those two targets are, however, the abstract conspicuously refuses to say.
The MR framework is methodologically appropriate here: using genetic variants as instrumental variables for gene expression levels sidesteps the confounding that plagues observational studies of drug targets. Replication across two independent GWAS cohorts is a genuine strength. Co-localization is the right follow-up step — it rules out the common MR failure mode where linkage disequilibrium, not a true causal variant, drives the signal. PheWAS for off-target effects is also good practice.
The scRNA-seq analysis is descriptive only — showing where the genes are expressed, not demonstrating mechanism. Molecular docking produces pretty figures but is a notoriously poor predictor of in vivo drug efficacy; it validates nothing clinically. The entire pipeline from genetic signal to "drug target" is inferential, and the distance between "this gene's expression correlates with ED risk via genetic instruments" and "blocking this protein improves erections" is enormous. No wet-lab validation, no animal model, no human tissue experiment.
This is a competent computational study that uses the right statistical tools to nominate two genetic drug targets for ED — and then declines to name them in the abstract, which undermines the entire point of publishing. The MR-plus-colocalization approach is legitimate hypothesis generation; the molecular docking and scRNA-seq layers add biological plausibility but no causal weight. Nothing here tells us a new ED drug works or even that it should work — it tells us two proteins are worth investigating. That's a valid and modest contribution, if the full text delivers what the abstract teases. The missing target names are an irritant that reviewers should have pushed back on.