Answers are grounded in this paper and cite their sources.
Background: Systemic lupus erythematosus (SLE) is a genetically complex autoimmune disease. Whole-blood expression quantitative trait locus (eQTL) analyses capture average regulatory effects across leukocytes, whereas SLE involves immune processes that depend on cell context. Therefore, a cell-resolved screen requires rigorous verification of instrument identity, allele orientation, and the multiple-testing denominator before biological interpretation. Aims: To conduct an identity-first, immune-cell-resolved cis-eQTL Mendelian randomization (MR) screen across SLE and related immune phenotypes and distinguish statistical signal counts from variant-level, regional, and cross-cell evidence. Study Design: Two-sample summary-data cis-eQTL MR and secondary evidence synthesis using 15 primary human immune-cell types from Database of Immune-Cell Expression (DICE) and eight disease outcomes from FinnGen release 12. Methods: All 19,009 gene-cell-type instruments were audited before outcome testing. Only 6,525 instruments with exact variant identities and concordant DICE effect information were retained for analysis. Wald ratios were calculated for 48,958 outcome-instrument tests, with the Benjamini-Hochberg (BH) procedure applied separately to each outcome to control the false discovery rate (FDR). Both prespecified infection-negative controls were then matched exactly by cell type, gene, and rsID, with each contributing 6,119 tests. For the sensitivity analysis, each chi-square statistic was divided by its outcome-specific genomic control inflation factor (lambda GC), and the BH procedure was reapplied within each outcome. Audited signals were summarized by unique single-nucleotide polymorphisms (SNPs), linkage disequilibrium (LD)-independent index signal, gene-outcome relationship, and MHC location. Regional colocalization, bounded multivariable MR, and Gene Ontology analyses provided additional layers of evidence. Results: The identity-first audit retained 6,525 of 19,009 instruments. Of these, 6,122 matched at least one outcome, whereas 403 did not. Among the 48,958 strict tests, 1,186 rows reached an outcome-specific FDR of less than 0.05, corresponding to 304 genes, 494 SNPs, 868 cell-gene pairs, and 424 gene-outcome relationships. LD auditing resolved 670 outcome-specific lead-SNP records into 639 reference-matched records and 31 separately reported records, yielding 288 LD-independent index signals. The primary SLE analysis identified 55 FDR-significant signals, 23 genes, 29 SNPs, and 11 index signals; 16 of the 23 genes and 6 of the 11 index signals were located in the MHC. The second SLE definition showed a similarly strong concentration of signals in the MHC. Most non-MHC disease signals were associated with rheumatoid arthritis (RA) and autoimmune hypothyroidism (AIH). Lambda GC ranged from 1.17 to 2.03 across disease outcomes. Following outcome-specific lambda adjustment, 483 disease rows remained FDR-significant. The greatest reductions in significant rows occurred for AIH (691-211), RA (213-103), and strict MI (83-11). The negative controls, other sepsis and bacterial infection not otherwise specified (NOS), each matched 6,119 tests and had lambda GC values of 1.06 and 1.08, respectively. No FDR-significant rows were identified for bacterial infection NOS. Conclusion: This revised analysis provides a graded and auditable screening framework rather than a definitive list of SLE risk genes. The primary SLE evidence is concentrated in the 6p21/MHC region, whereas broader non-MHC candidate signals arise predominantly from related immune phenotypes. These findings are hypothesis-generating and require independent replication and experimental validation.
No takes yet. Share an insight, caveat, or question.
Wu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: