Coordinated upregulation of Hippo, NF-κB and interferon pathway genes in pancreatic ductal adenocarcinoma: a retrospective tissue qRT-PCR profiling study with external tumour-versus-non-tumour validation, molecular subtype and survival analyses.

Abstract

Objectives: Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal solid tumours. Alterations in Hippo, NF-κB and interferon (IFN)-related signalling have each been implicated in its biology, but the extent to which these transcriptional programmes are coordinately dysregulated in tumour tissue remains incompletely defined. This study therefore set out to determine whether these programmes are coordinately upregulated in PDAC tissue, and whether individual transcripts discriminate tumour from non-tumour tissue reproducibly in an independent cohort.

Methods: In this retrospective study we profiled 17 mRNA targets by real-time quantitative PCR in 50 formalin-fixed paraffin-embedded PDAC specimens and 35 non-tumoural pancreatic tissue samples, covering the Hippo pathway core and output, the inflammation/IL-6 module, the IFN-related axis and the NF-κB cascade, normalised to GAPDH using the 2 - ΔΔCt method. Discrimination was assessed by full-cohort receiver-operating-characteristic analysis without resampling and by nested cross-validation, and associations with clinicopathological variables were tested with Benjamini-Hochberg correction. Findings were evaluated externally in GSE62452 and TCGA-PAAD.

Results: All 17 transcripts were significantly upregulated in PDAC (FDR < 0.05), with median fold-changes from 3.4 (WWTR1) to 18.4 (LATS1). MYD88, ANKRD1, ISG15 and YY1 discriminated best (AUC 0.853-0.911), and nested cross-validation gave a signature AUC of 0.890 (95% CI 0.821-0.959), only marginally better than MYD88 alone. No clinicopathological variable was associated with transcript abundance after correction. In GSE62452 (69 PDAC, 61 adjacent non-tumour tissues) eight transcripts replicated with concordant direction, led by YAP1 (AUC 0.791), TEAD1 (0.760) and MYD88 (0.755), and the fixed four-gene discovery signature retained discrimination (AUC 0.798, 95% CI 0.719-0.877). Replication was gene-specific rather than module-wide: IL6, the remaining NF-κB transcripts and four Hippo members did not replicate, and IFNB1 was discordant. Hippo output genes tracked the basal-like subtype in both external cohorts. In TCGA-PAAD, high YY1 (HR 2.01; p = 0.001) and high YAP1 (HR 1.72; p = 0.010) predicted shorter survival, but both associations lost significance when the analysis was restricted to ductal-type histology.

Conclusions: PDAC tissue shows coordinated upregulation across these four modules, but only eight of the seventeen transcripts replicated externally, most strongly within the Hippo effector axis and for MYD88. YAP1, TEAD1, MYD88 and STK4 reached validation AUCs above 0.70 and are prioritised for prospective study. Because transcript abundance does not measure pathway activity and all data derive from bulk tissue, these associations are hypothesis-generating rather than mechanistic.