Multilayer h-Backbones Identify KLRB1 as a Stable Immune-Ageing h-Focus across Populations, Tissues and Cell Lineages

Immune-ageing signals are observed at different biological scales, but population estimates, multi-organ localisation and direct age contrasts are not statistically interchangeable. We assembled a gene-aligned evidence matrix in which Sayed et al. supplied continuous population-age estimates, Tabula Sapiens v1 supplied tissue and immune-lineage localisation, and Wells et al. supplied an independent direct ageing reference. Across a locked panel of 295 harmonised immune genes, a deterministic one-order h-cutoff produced an h-core subnet, weak-bridge restoration produced an h-backbone, and a two-order h-cutoff selected h-foci independently in each layer. The nine discovery-layer h-backbones retained 4,604–7,741 of 21,770–37,365 positive candidate edges and restored 199–361 weak bridges. Two hundred subject bootstraps and 55 exhaustive Tabula leave-one-donor-out perturbations showed that KLRB1 alone met the prespecified ≥0.80 stability criterion across population, tissue and cell families. KLRB1 was selected in 93.5% of Sayed bootstraps and all tissue and cell perturbations, decreased with age in Sayed (Spearman ρ = −0.191; Benjamini–Hochberg q = 0.0389), and showed concordant negative age effects in Wells (mean z = −0.881; two profiles q < 0.05). This estimator–localiser–reference design identifies KLRB1 as a stable ageing-associated h-focus while preserving the inferential boundary between localisation, replication and causality.