ANABOLIC FAILURE, NOT CATABOLIC ACTIVATION, UNDERLIES GLYCOGEN DEPLETION IN CERVICAL CARCINOGENESIS: A MULTI-COHORT TRANSCRIPTOMIC ANALYSIS
Abstract Background. Cervical epithelium loses its glycogen as disease grade rises, an observation old enough to underpin iodine-based visual screening, but which arm of glycogen metabolism accounts for the loss has not been established. The expectation from cancer metabolism is catabolic: that stored glycogen is mobilized and glycolysis activated. Methods. We analyzed 238 specimens from four publicly available cohorts across three sample types: 128 laser-capture-microdissected epithelial samples, 35 bulk-tissue samples, and 75 exfoliated cytology samples in two sets (44 and 31; GSE27678). We prespecified a 33-gene panel comprising four modules: glycogen synthesis, glycogen degradation, glycolysis, and squamous differentiation. For each specimen, we calculated module scores using single-sample gene set enrichment. Differential expression was assessed with limma, pathway enrichment with fgsea, and classification performance with bootstrap confidence intervals for ROC curves. We documented all analytical decisions and their rationale. Results. Synthesis and differentiation fell with grade in all four cohorts, while degradation and glycolysis stayed at their normal-tissue baseline. All seven differentiation genes were reduced in carcinoma, and all seven occupied the top 5.1% of the ranked transcriptome, with CRNN the most down-regulated of 21,755 genes. Synthesis failed selectively, through GYS2 and PPP1R3C only, with the ubiquitous synthase GYS1 unchanged. The glycolytic module separated from normal epithelium at no grade (p 0.45 to 0.62), classified CIN3 or worse at an area under the curve of 0.569 with an interval spanning chance, and showed no enrichment in carcinoma. Synthesis separated from normal first, at CIN2, and differentiation at CIN3. The composite index tracked grade in the discovery cohort (Spearman rho 0.652, p 7.9e-17) but offered no discriminatory advantage over its differentiation component alone (0.832 against 0.832). Conclusions. Glycogen depletion in cervical carcinogenesis reflects failure of a tissue-restricted synthetic axis, GYS2 and PPP1R3C, within a collapsing squamous differentiation program, not catabolic activation. Measuring all four candidate mechanisms on one scale in the same specimen is what allows the catabolic explanation to be excluded rather than merely left untested. The pattern holds across two array chemistries and three specimen preparations, including the exfoliated cytology that screening already collects, and the synthesis and differentiation modules together are the molecular counterpart of what iodine staining reports. These are transcript-level findings and define no clinical threshold.