Hepatocyte-enriched LEAP2 in relation to heel bone density and osteoporosis-related phenotypes: an exploratory genetic and translational study

Background Unbiased evaluation of liver-derived factors may clarify endocrine links between hepatic metabolism and skeletal fragility; however, hepatic transcript abundance cannot be assumed to represent circulating protein or direct bone-cell signaling. Methods We screened 168 liver-secreted genes and 202 independent liver cis -eQTL signals across seven skeletal outcomes using outcome-wise false-discovery-rate control, Steiger direction checks, bilateral sum of single effects (SuSiE) fine-mapping, and a three-edge analysis of liver expression, circulating protein, and heel estimated bone mineral density (eBMD). LEAP2 was then examined in human liver cell-resolved, spatial and bulk datasets, public bone transcriptomes, an exploratory hospital serum cohort [30 osteoporosis (OP) and 30 nonosteoporosis (NOP) participants], and a 2 × 2 high-fat diet (HFD) × orchidectomy (Orx) mouse experiment. Results The confirmatory genetic screen was null. In the separately defined exploratory layer, LEAP2–heel eBMD was the sole pair meeting the prespecified bilateral SuSiE criteria (PP.H4 = 0.915327); heel eBMD is quantitative-ultrasound-derived rather than a dual-energy X-ray absorptiometry (DXA) outcome. The MR association remained significant after a post hoc Bonferroni correction across all 1,414 signal-outcome combinations (adjusted p = 1.39 × 10 −14 ). Colocalization was prior-sensitive: PP.H4 was 0.8439 at p 12 = 5 × 10 −6 and 0.1903 under a deliberately conservative p 12 = 1 × 10 −5 /46 stress test, so no formal family-wise colocalization claim was made. The liver eQTL–circulating pQTL edge was analytically indeterminate because the pQTL model did not converge, and the complete genetic triangle was not supported. Hepatocytes dominated LEAP2 signal in all three cell-resolved datasets. Serum LEAP2 was lower in OP than NOP participants (median [IQR], 5.336 [3.780–6.083] vs. 8.153 [7.015–9.458] ng/mL; U = 854, p = 2.33 × 10 −9 ), but age imbalance and unavailable metabolic covariates preclude an independent association claim. In mice, HFD and Orx were associated with adverse trabecular phenotypes and lower hepatic Leap2 and circulating LEAP2. Conclusion A null confirmatory screen and prior-sensitive exploratory colocalization nominate LEAP2 for further study but do not establish circulating mediation, direct bone-cell GHSR signaling, diagnostic utility, or a causal effect on bone.