Histologic grade and genotype-associated metabolic and proliferation-related transcript patterns in hepatocellular carcinoma: a transcriptomic analysis of public cohorts

Histologic grade and genotype-associated metabolic and proliferation-related transcript patterns in hepatocellular carcinoma: a transcriptomic analysis of public cohorts Overview This record provides the preprint manuscript and supporting research materials for a transcriptomic analysis of hepatocellular carcinoma (HCC). The study examines how hepatic-metabolic and proliferation-related expression profiles reveal biological and outcome-associated diversity within established histologic grades, molecular subclasses and genotype groups. It also proposes exploratory prognostic gene candidates and investigates their relationships with stress/metabolic and vascular–stromal transcription within H-dominant G1/G2–S3 tumors. Study design Fixed target–reference gene pairs defined hepatic-metabolic (H) and proliferation-related (P) contrasts. H summarized urea-, bile- and gluconeogenesis-related transcription, while P summarized replication- and mitosis-related transcription. These measurements were integrated with recorded grade, reconstructed Hoshida subclass, TP53 and CTNNB1 alterations, and AFP/EPCAM/GPC3 expression. Outcome analyses compared continuous P, continuous P–H balance and relative P dominance. The principal TCGA-LIHC population comprised 348 G1–G3 tumors, including 321 overall-survival and 277 source-defined disease-free-survival (DFS) observations. LIRI-JP contributed 169 exact-grade primary-tumor specimens from 164 donors, and the external GSE14520 recurrence analysis included 219 patients. Cohorts were analyzed separately, retaining their source-specific endpoint definitions. TCGA DFS denotes recorded recurrent/progressive events from initial diagnosis. Principal findings Most variation in H and P occurred within, rather than between, recorded grades. TP53 and CTNNB1 alterations showed opposing grade-adjusted metabolic–proliferative associations, while molecular subclass, hybrid-grade profiles and established markers supplied additional biological detail. In TCGA, P dominance was associated with 24-month DFS across nine grade–subclass strata (hazard ratio 3.95; 95% confidence interval 2.27–6.88) and contributed information beyond continuous P in a clinical/subclass-adjusted model (2.67; 1.44–4.97). Associations also persisted within G2 and G3 after TP53 and marker adjustment. Continuous P was associated with overall survival in both external cohorts. GSE14520 showed a direction-consistent but uncertain 24-month recurrence association for P dominance (1.22; 0.77–1.93), without a detectable increment beyond continuous P. Exploratory candidates and tissue-expression relationships Within H-dominant G1/G2–S3 tumors, the 12-month DFS event group showed higher SLC22A31 and lower FAM219B and TRIM66 expression. These three candidates met contrast-specific, but not full-family, false-discovery criteria. DDIT3 provided a related stress-response lead. The 24-month exploration highlighted ESAM, PECAM1, CLEC14A, LAMA4 and SPARC, with coordinated expression involving PGF, TIE1 and DLL4. Further analyses identified coexpression of SLC16A3 and PLAUR with PGF/PECAM1, and of LOXL2 with SPARC/LAMA4. These relationships direct attention beyond the fixed replication/mitosis contrast toward metabolic, cellular-state and vascular–stromal features. A retained HK2/IL6-high index tumor is additionally documented through its accompanying 60-gene expression profile, preserving specimen-level heterogeneity for comparison with future cases. Outcome-expression contrasts and coexpression analyses are reported as distinct lines of evidence. Biological implications The findings motivate the hypothesis that transport/chromatin and stress-metabolic states in some H-dominant G1/G2–S3 tumors interact with vascular–stromal remodeling and contribute to conditions permissive for intrahepatic dissemination. A tissue state established before resection could influence when subsequent events become clinically apparent. Single-cell and spatial studies should resolve the cellular origins of the candidate transcripts and distinguish cell abundance from within-cell expression state. Prospective resection cohorts should evaluate candidate combinations and their incremental prognostic information alongside established clinicopathological and H/P measurements, including their potential use in targeted bulk-tissue assays. Deposit contents CS9-manuscript_hcc_ymp-26Oct07.pdf is the reader-facing manuscript. CS9-manuscript_hcc_ymp-26Oct7.zip contains the manuscript in Word and PDF formats; Supplementary Methods SM1–SM18 and Results SR1–SR16; Supplementary Tables S1–S24, including the separately supplied S24; Supplementary Figures S1–S13 with images and legends; Main Figures 1–12 in PNG format; and source-data files containing analysis-specific inputs, model outputs, figure-source tables and supporting computational materials. Author: Young Min Park, M.D., Ph.D.Affiliation: Liver Research Initiative (Independent Researcher), Republic of Korea.Preprint / manuscript archive DOI: 10.5281/zenodo.23214311.