Spatial profiling of protein expression using an artificial intelligence pipeline has revealed distinct prognostic image features in rectal mucinous adenocarcinoma and non-mucinous adenocarcinoma. An investigation published in the Journal of Translational Medicine demonstrates that evaluating protein signal variability and tissue compartment distribution provides prognostic information that extends beyond traditional mean intensity measurements. The findings highlight biological differences between these two histological subtypes of rectal cancer.
Mucinous adenocarcinoma and non-mucinous adenocarcinoma represent clinically and pathologically distinct forms of rectal cancer, yet the spatial distribution and clinical relevance of subtype-specific biomarkers have remained poorly defined. To identify robust candidate markers, investigators first examined transcriptomic datasets from two independent patient cohorts, including 177 cases from TCGA-READ and 281 cases from the Swedish Uppsala-READ cohort. Using weighted gene co-expression network analysis, the research team identified subtype-associated transcriptomic modules, uncovering 109 shared genes across both cohorts.
Guided by these transcriptomic results, the researchers selected three candidate markers for in-depth spatial evaluation: MUC1, MUC2, and TCN1. The team then analyzed these proteins using multiplex immunofluorescence on the COMET platform in an independent validation cohort. This validation series, designated the Southeast Swedish-READ cohort, comprised 155 patients diagnosed with rectal adenocarcinomas, enabling detailed characterization of marker localization across distinct tissue compartments.
To quantitatively evaluate the multiplex imaging data, the researchers constructed a customized artificial intelligence-assisted analytical pipeline. The automated pipeline generated 79 spatial features from the multiplex immunofluorescence images, capturing complex signal variability, regional cell densities, and marker distributions. Following rigorous correction for multiple testing, nine spatial features exhibited statistically significant differences between mucinous and non-mucinous adenocarcinoma samples.
Survival analyses using Kaplan-Meier modeling indicated that five spatial features were significantly associated with disease-free survival, revealing divergent patterns across the histological subtypes. In patients with mucinous adenocarcinoma, higher intratumoral variability of TCN1 expression was associated with longer disease-free survival. Conversely, in non-mucinous adenocarcinoma, favorable survival was linked to greater MUC2 variability, higher nuclear counts within the tumor region, and elevated positive nuclear counts associated with MUC1 in the stromal compartment.
Further multivariable Cox regression adjustment across the overall cohort confirmed that higher tumor-region nuclear counts and increased stromal MUC1-associated positive nuclear counts remained independently associated with improved disease-free survival. In addition, exploratory treatment-stratified assessments revealed that adjuvant chemotherapy correlated with longer disease-free survival specifically among mucinous adenocarcinoma patients displaying low MUC2 variability, whereas no corresponding survival relationship was detected in patients with non-mucinous tumors.
The study authors concluded that transcriptome-guided, machine learning-assisted spatial immunofluorescence profiling underscores the diagnostic and prognostic value of measuring regional signal heterogeneity rather than overall staining intensity alone. However, the researchers emphasized that the observed associations, particularly the exploratory findings involving chemotherapy response in mucinous adenocarcinoma, represent preliminary observations that require validation in larger, independent prospective cohorts before clinical translation.
Subject of Research: Spatial multiplex immunofluorescence profiling in rectal adenocarcinoma subtypes
Article Title: Histology-related clinical significance of artificial intelligence-assisted multiplex immunofluorescence features in rectal mucinous and non-mucinous adenocarcinoma
Article References: Zhao, K., Vernmark, K., Knutsen, A., Li, Y., Stadler, C., Fan, C., & Sun, X.-F. (2026). Histology-related clinical significance of artificial intelligence-assisted multiplex immunofluorescence features in rectal mucinous and non-mucinous adenocarcinoma. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-09242-3
Image Credits: AI Generated
DOI: 10.1186/s12967-026-09242-3
Keywords: rectal cancer, mucinous adenocarcinoma, non-mucinous adenocarcinoma, multiplex immunofluorescence, spatial profiling, artificial intelligence, disease-free survival, MUC1, MUC2, TCN1, COMET platform, transcriptomics, TCGA-READ, prognostic biomarkers, adjuvant chemotherapy, translational medicine
News Source: Nathaniel Bowman. (November 18, 2026). AI Spatial Profiling Identifies Prognostic Protein Markers in Rectal Cancer Subtypes. Scienmag.



