Pre-specification analysis plan: two-sample Mendelian randomization of whole-blood cis-eQTLs for ESR1 and three neighbouring skeletal genes against fracture risk

Abstract

This record is the pre-specification analysis plan for a two-sample Mendelian randomization (MR) study of fracture risk, submitted to the Journal of Bone and Mineral Research (JBMR). Scope. Two-sample MR of four pre-specified cis-eQTL instruments from whole blood (eQTLGen Phase-I, N = 31,684) -- ESR1 and its neighbours NCOR1, SREBF1 and CCR1 -- against fracture risk (UK Biobank GCST006980, 53,184 cases / 373,611 controls), with replication in a second UK Biobank phenotype and in FinnGen R11, and multi-GWAS colocalization plus fine-mapping. Why this record exists. The analysis hierarchy, the colocalization decision rule and the conditional-analysis implementation were all fixed before any association estimate was computed. Depositing the plan makes those pre-specification claims independently auditable, as requested by reviewers and consistent with the STROBE-MR reporting standard (Skrivankova et al., JAMA 2021). The plan was frozen on 2026-09-09 and was not registered prospectively in a public registry; the date of freeze is stated explicitly in the document. Contents (7 sections): 1. Study design and the pre-specified estimand hierarchy (primary = IVW on the r2 < 0.001 LD-clumped set; sensitivity = r2 < 0.01 reported in full; single-SNP lead Wald ratio; supporting methods reported but not used for inference). 2. Instrument definition and reconstruction of exposure effects from eQTLGen Z-scores and allele frequencies, with the allele-frequency-source sensitivity check. 3. Outcomes and replication, including the pre-specified reading that FinnGen confirms direction rather than providing independent-instrument corroboration. 4. The pre-specified colocalization decision rule (coloc.abf PP.H4 > 0.80 threshold, priors, and locus-wide sensitivity padding). 5. COJO-equivalent conditional analysis: exact implementation provenance and the conditional results for the two ESR1 instruments. 6. Prior evidence considered at planning time, recorded as context and not as hypothesis tests. 7. Methodological references supplementary to the 50-reference manuscript list. Files in this record: ANALYSIS_PLAN_JBMR.pdf -- the plan, fixed-layout render (register this file) ANALYSIS_PLAN_JBMR.docx -- the same document as editable source Recommended citation of the underlying article will be added to the manuscript record; the manuscript itself cites this deposit by its DOI. Licence: CC BY 4.0.