Cross-Tissue Concordance of a T1D Blood-Derived B-Cell Signature with Hashimoto Thyroiditis Tissue: Exploratory Spatial Immune-Topic Analysis

Type 1 diabetes (T1D) and Hashimoto thyroiditis (HT) share autoimmune features, but cross-tissue concordance of B-cell transcriptional programs remains uncertain. Using 24 retained cohort–pathway effect estimates, we recalculated T1D blood summary-level meta-analysis statistics and separately report the previously generated frozen 14-gene HT thyroid association. The T1D meta-analysis yielded Hedges’ g = 0.51 (95% CI 0.18–0.85; q = 0.022; I2 = 0%). In HRA001684 (16 HT, 50 non-HT), the frozen score was associated with HT (HC3-adjusted β = 2.65, 95% CI 2.37–2.94; permutation P = 2 × 10−5). Single-cell data (6512 cells; three HT and three immune-checkpoint-inhibitor-associated thyroiditis libraries) lacked healthy controls and were analyzed descriptively. In five spatial sections, the frozen-signature contrast was nonsignificant (Δ = 0.459; exact P = 0.10; matched-gene P = 0.085), with no preferential correspondence to B/plasma-cell topics (reported P = 0.90). Independently inferred thyroid immune topics showed regional variation but did not localize the frozen score. These reported findings are compatible with partial T1D blood–HT thyroid transcriptional concordance but do not establish systemic HT activation, a shared cell-intrinsic program, or validated spatial localization. In the reported sensitivity analysis, joint B-cell/plasma-cell marker adjustment attenuated the HT association by approximately 90%, with the fully adjusted confidence interval including zero. Comparisons with equal weighting and other B-cell reference signatures did not establish a statistically significant advantage of the T1D-derived weights. The public GitHub workflow verifies summary-level pooling of retained effects only; source-to-result reconstruction of the HT bulk, spatial, single-cell, and marker/weight analyses has not been established.