Immune Cell-Specific Regulatory Gene Prioritization in Low Back Pain Using Single-Cell eQTL-Informed Mendelian Randomization and Colocalization.

Objective: Low back pain (LBP) is a heterogeneous clinical pain condition, but the contribution of inherited immune cell-specific regulatory variation to LBP susceptibility remains poorly understood. We aimed to prioritize peripheral immune cell-specific regulatory genes associated with LBP by integrating single-cell eQTL-informed Mendelian randomization (MR), Bayesian colocalization, consistency analyses across related spinal phenotypes, and transcript-level inflammatory-response assessment. Methods: Single-cell eQTL data from the OneK1K resource across 14 peripheral immune cell types were integrated with FinnGen release 12 genome-wide association study summary statistics for LBP. Each gene-cell type pair was analyzed as a separate exposure using cell-type-resolved MR, with multiple testing controlled by the Benjamini-Hochberg false discovery rate. FDR-significant associations were further prioritized using Bayesian colocalization and assessed for directional consistency in two related FinnGen phenotypes, lower back pain or/and sciatica and lumbar disc prolapse. High-confidence genes were functionally annotated and ranked at the gene level. Selected candidates were subsequently assessed by qPCR in an LPS-stimulated THP-1-derived macrophage-like model to examine their transcriptional responsiveness to inflammatory stimulation. Results: Cell-type-resolved MR identified 254 FDR-significant gene-cell type associations with LBP, of which 130 remained outside the extended major histocompatibility complex region. Among 114 MR-prioritized gene-cell type pairs evaluated by colocalization, 24 showed strong evidence of a shared genetic signal (PP.H4 ≥ 0.8), with most high-confidence signals located outside the MHC region. Recurrent candidates included C1orf86, RGS12, NT5E, CLLU1/CLLU1OS, RAB38, and LTBP1. All 24 strongly colocalized pairs showed concordant effect directions and nominal associations in both related FinnGen phenotypes, providing consistency support across related spinal pain and disc phenotypes rather than independent replication. In the THP-1 inflammatory-response model, LPS robustly induced canonical inflammatory markers, while C1orf86, RGS12, and NT5E showed significant transcriptional responses after multiple-testing correction. Conclusion: This study prioritizes a focused set of peripheral immune cell-specific regulatory candidates associated with LBP using convergent MR, colocalization, and related-phenotype evidence. The qPCR findings provide complementary evidence of inflammatory responsiveness for selected candidates but do not establish the cell-type-specific or molecular mechanisms underlying LBP. C1orf86, RGS12, and NT5E warrant further investigation in the immune cell populations and pain-relevant tissues highlighted by the genetic analyses.