PCOS and EC risk. Polycystic ovary syndrome (PCOS) is the most common endocrinopathy in women, affecting 13% worldwide. Women with PCOS have a 2.7-fold increased risk of developing endometrial cancer (EC), driven by hormonal imbalances including hyperandrogenemia, hyperinsulinemia, and chronic anovulation leading to unopposed estrogen exposure.
Sedentary behavior and cancer. Sedentary behavior is linked to 22% of all-cause mortality and 14% of cancer mortality. Physical inactivity increases EC risk by promoting obesity and hormonal dysregulation. Conversely, diet and exercise reduce EC risk and prevent recurrence, suggesting a shared metabolic pathway between sedentary behavior and EC pathogenesis.
The aromatic compound connection. Aromatase - the enzyme catalyzing androgen-to-estrogen conversion - is the rate-limiting step in estrogen biosynthesis. Its expression is reduced in PCOS granulosa cells and is regulated by skeletal muscle, the primary site of sedentary behavior analysis. This creates a biological bridge between PCOS, physical inactivity, and estrogen-driven EC.
WGCNA as a discovery tool. Weighted gene co-expression network analysis (WGCNA) identifies biologically coherent gene modules correlated with clinical traits. The authors applied WGCNA to GEO gene expression datasets from PCOS and sedentary individuals to discover shared molecular mechanisms and test their relevance in EC.
Dataset selection. Four GEO datasets from skeletal muscle were selected: GSE6798 and GSE9103 as the discovery cohort (PCOS and sedentary, respectively) and GSE8157 and GSE9405 as the validation cohort. All datasets required samples from cases and controls with a minimum of 10 per group. TCGA provided 554 EC and 23 normal endometrial samples for cancer validation.
WGCNA module identification. Genes with low variability were removed before WGCNA analysis in R 4.0.4. Power values were selected via the scale-free network criterion, and TOM (topological overlap matrix) matrices were computed. Module-trait correlation heatmaps identified PCOS-related modules (green: r=0.61, grey: r=-0.85) and sedentary-related modules (light cyan: r=0.51, light green: r=0.57).
Shared gene identification. The overlap of positively correlated PCOS and sedentary modules yielded 50 shared genes (gene set 1, GS1). Venn diagram analysis of validated DEGs from GSE8157 and GSE9405 produced gene set 2 (GS2) with 20 DEGs, of which 6 overlapped with GS1: ZHX3, NAA15, CYB561, B3GALT6, FAM217B, and CYP27A1.
Immune landscape and pathway analysis. CIBERSORT deconvolution assessed 22 immune cell type proportions in 554 TCGA EC samples. Unsupervised consensus clustering of 27 cellular aromatic compound metabolic process pathway genes stratified patients into ACMP-high and ACMP-low subtypes. Survival analysis, immune infiltration, and immune checkpoint expression were compared between subtypes.
Top shared pathway. ClueGO enrichment analysis of GS1 genes revealed the cellular aromatic compound metabolic process (ACMP) pathway as the dominant shared GO term, accounting for 72.73% of all GO keywords in PCOS and linked to 16 genes. This pathway reflects the central role of aromatase in converting androgens to estrogen.
miRNA-mRNA network. Six miRNAs were found to be common to PCOS and sedentary people (has-miR-31-3p, has-miR-193a-5p, has-miR-127-5p, has-miR-520a-5p, has-miR-199b-3p, has-miR-222-5p). These miRNAs are enriched in the ACMP pathway. A miRNA-mRNA network of 18 nodes and 20 edges was constructed, with hsa-miR-520, hsa-miR-193, and hsa-miR-199 proposed to upregulate aromatic compound-induced genes.
Validation cohort confirmation. In GSE8157 and GSE9405, 14 common upregulated genes were found in PCOS and sedentary individuals, with functional enrichment again dominated by ACMP, cellular component organization, and cellular biosynthetic processes - consistent with the discovery cohort findings and confirming methodological reliability.
ACMP pathway disease bridge. The study proposes ACMP as a molecular bridge between sedentary behavior, PCOS, and EC: skeletal muscle produces aromatase, sedentary behavior increases aromatase activity, aromatase deficiency characterizes PCOS, and excess estrogen drives EC. This mechanistic model is supported by the gene co-expression, miRNA, and immune landscape data.
EC subtype classification. Unsupervised consensus clustering of TCGA EC samples using 27 ACMP pathway genes yielded two stable subtypes (k=2): ACMP-high and ACMP-low. Survival analysis showed significant differences between the two subtypes, with the ACMP-high subtype demonstrating different survival outcomes.
Immune landscape differences. CIBERSORT analysis showed that ACMP-high patients had significantly higher proportions of regulatory T cells (Tregs) and dormant dendritic cells, and lower active dendritic cells. HLA gene expression was broadly elevated in the ACMP-high subtype, indicating a more immunologically active tumor microenvironment - an immune-hot phenotype.
Immune checkpoint upregulation. Three key immune checkpoints (CD274/PD-L1, PDCD1/PD-1, and PDCD1LG2/PD-L2) were significantly differentially expressed between ACMP-high and ACMP-low groups. This suggests ACMP-high EC may be more responsive to checkpoint inhibitor immunotherapy, potentially offering a new therapeutic stratification approach.
ESTIMATE and stromal scores. The ACMP-high subtype showed higher immune, stromal, and ESTIMATE scores, with lower tumor purity compared to ACMP-low. This global immune activation is consistent with better immune surveillance and potential sensitivity to immunotherapy in ACMP-high patients.
Hub gene selection. Among the six overlapping genes in GS1 and GS2, ZHX3, NAA15, and FAM217B had not been studied in EC. STRING protein interaction analysis showed NAA15 had the most interaction partners. NAA15 expression was significantly elevated in the ACMP-high subtype and higher EC expression predicted shorter survival.
Expression validation. The Human Protein Atlas confirmed that NAA15 protein is consistently overexpressed in EC tissue compared to normal endometrium. qRT-PCR confirmed elevated NAA15 mRNA in 5 EC cell lines (HEC-1A, RL95-2, Ishikawa, KLE, AN3CA) relative to normal hESC controls.
Functional gain and loss experiments. si-NAA15 knockdown in Ishikawa cells reduced proliferation (CCK-8 and colony assay), while NAA15-cDNA overexpression in HEC-1A cells promoted growth. Transwell invasion assays confirmed that NAA15 knockdown decreased invasion and overexpression increased it - establishing NAA15 as a functional oncogene in EC.
Clinical relevance of NAA15. NAA15 is a subunit of NatA, the major N-terminal acetyltransferase complex, which broadly regulates protein stability, function, and degradation. Its overexpression in EC places it at the intersection of metabolic dysregulation and oncogenic signaling, making it a candidate diagnostic biomarker and therapeutic target.
Primary finding. WGCNA-based analysis of PCOS and sedentary gene expression datasets identified the cellular aromatic compound metabolic process pathway as the shared molecular mechanism linking both risk factors to EC pathogenesis. This pathway stratifies EC patients into immunologically distinct subtypes with different survival outcomes.
NAA15 as a translational target. NAA15 was identified as a novel hub gene functionally validated to promote EC proliferation and migration. Its consistent overexpression in EC tissue and correlation with ACMP-high immune-hot subtype make it a promising biomarker for EC diagnosis and a potential target for therapeutic intervention.
Immunotherapy implications. The ACMP-high subtype's immune-hot phenotype, elevated immune checkpoints, and Treg infiltration suggest that pathway-based molecular stratification could identify EC patients most likely to benefit from checkpoint inhibitor therapies - an increasingly important consideration as immunotherapy expands in gynecologic oncology.
Study limitations. The skeletal muscle transcriptomes used for PCOS and sedentary datasets are not directly comparable to endometrial cancer transcriptomes, and connections were inferred rather than directly measured. Future studies should validate ACMP pathway and NAA15 expression in prospective EC clinical cohorts with treatment outcome data.