Genetic mimicry of haematopoietic growth factor signalling: efficacy–safety trade-offs for cytopenias by drug-target Mendelian randomisation
Higher MPL expression is linked to a lower risk of thrombocytopenia, but G-CSF and EPO pathways raise concerns about thromboembolic risks.
Where it sits
this study against the rest of the erythropoietin (epo) corpusSummary and findings
This study investigated the efficacy and safety trade-offs of erythropoietin (EPO) and other haematopoietic growth factors for cytopenias using drug-target Mendelian randomisation. The analysis involved large datasets from various sources, including over 500,000 individuals in FinnGen. The findings indicated that higher MPL expression was associated with lower thrombocytopenia risk, while EPO variants did not show clear forward agonism.
Abstract
<b>Objectives: </b>Recombinant haematopoietic growth factors are mainstay treatments for cytopenias, whose efficacy and safety are known from separate trials. Drug-target Mendelian randomisation (MR) can compare these pathways when instruments reflect forward pathway activation rather than protein clearance or reverse physiological feedback. <b>Methods: </b>We instrumented erythropoietin (EPO), thrombopoietin (MPL) and granulocyte colony-stimulating factor (G-CSF/CSF3R) using plasma proteomic platforms (deCODE, N=35,559; UK Biobank Pharma Proteomics Project, N=54,219) and whole-blood expression (eQTLGen, N=31,684), testing efficacy endpoints, seven safety domains in FinnGen R13 (~500,000) and blood counts (746,667). Analyses used inverse-variance weighted MR with sensitivity models, colocalisation, and a physiological-direction check requiring the protein-raising allele to move downstream cell counts in the biologically established direction. <b>Results: </b>Higher MPL expression lowered thrombocytopenia risk (OR 0.65, 95% CI 0.49–0.86, P=0.0026), promoted multilineage haematopoiesis sparing lymphocytes, and conferred no thrombotic excess. G-CSF pathway instruments showed no protection against agranulocytosis (OR 1.07, P=0.57) or infection outcomes, whereas receptor instruments increased venous thromboembolism (OR 1.14) and coronary heart disease (OR 1.09). EPO lead variants mapped to the neighbouring genes GIGYF1 and TFR2, reflecting iron-restricted compensatory feedback rather than forward agonism and rendering the axis genetically non-estimable. <b>Conclusions: </b>Physiological-direction checking should become mandatory in cis-pQTL drug-target MR.
Background
Not reported in abstract.
Methods
Not reported in abstract.
Results
Not reported in abstract.
Interpretation
Not reported in abstract.
Limitations
Not reported in abstract.