The Origins of Allergy: An Outside‐In, Bioenergetic Hypothesis

ABSTRACT Allergic disease has risen too rapidly for inherited susceptibility to explain it, and the risk variants so far reported lie in the epithelial barrier and in type‐2 immunity. We propose a five‐layer model in which epithelial barrier and bioenergetic dysfunction may precede and promote type‐2 immune deviation rather than arise solely as its consequence, progressing from (1) barrier and bioenergetic dysfunction, with mitochondrial DNA escaping into the cytosol, through (2) danger sensing and (3) epithelial–immune communication, to (4) type‐2 immune amplification and (5) a systemic layer of circulating IgE, neurogenic amplification and the atopic march, with reciprocal inflammatory feedback. Layer 1 is also the least measured: standard immune panels carry almost none of these genes, and no genome‐wide association study has implicated the mitochondrial genome or its nuclear regulators. From 1856 catalogued transcriptomic series, 985 of them expression series, we drew 116 case–control cohorts from 110 series, 3776 patients and 2164 controls, with Layer 1 prespecified as 350 genes across 19 domains. Of 212 module estimates nine survived false‐discovery correction: two type‐2 positive controls, two cytosolic sensing and inflammasome modules, raised in oesophageal tissue (Hedges' g +1.70) and atopic skin (+1.27), and five bioenergetic modules in oesophageal tissue. In eosinophilic oesophagitis the mitochondrial clearance limb was reduced in every one of thirteen cohorts (HIF1A −1.94, BNIP3–1.98). STING1 increased in all thirteen skin cohorts that measured it while MAVS fell in ten of thirteen oesophageal cohorts, and a STING1 minus MAVS contrast separated the allergic epithelia from psoriasis, chronic obstructive pulmonary disease and asthma without exception (exact p 0.029). In one series of clinically uninvolved atopic skin KEAP1 increased while NFE2L2, claudin‐1 and filaggrin decreased ( p < 0.001), in tissue that is not inflamed; pooled across nine series the directions largely hold without reaching significance. Broad suppression of mitochondrial programmes across allergic tissues was not supported: the phenotype is tissue‐dependent, present in eosinophilic oesophagitis and atopic dermatitis, absent in nasal polyposis and established asthma and conditional in allergic rhinitis. Prospective and experimental studies must establish temporal and causal relationships.