Histology-related clinical significance of artificial intelligence-assisted multiplex immunofluorescence features in rectal mucinous and non-mucinous adenocarcinoma

Abstract Background Mucinous adenocarcinoma (MAC) and non-mucinous adenocarcinoma (NMAC) are distinct histological subtypes of rectal cancer, but the spatial patterns and clinical relevance of MAC-associated biomarkers remain poorly defined. Methods Transcriptomic data from TCGA-READ ( n = 177) and the Swedish Uppsala-READ cohort ( n = 281) were analysed using weighted gene co-expression network analysis to identify MAC-associated candidate markers. Selected markers were subsequently evaluated by multiplex immunofluorescence (mIF) using the COMET platform in an independent Southeast Swedish–READ cohort of 155 patients with rectal adenocarcinomas. Associations of mIF-derived image features with histological subtype and disease-free survival (DFS) were assessed using group comparisons, Kaplan–Meier analysis, and Cox regression. Results MAC-associated transcriptomic modules were identified in TCGA- and Uppsala-READ cohorts, with 109 shared genes. MUC1, MUC2, and TCN1 were selected for mIF analysis. A customized AI-assisted pipeline was developed and generated 79 spatial features from mIF images, of which 9 differed between MAC and NMAC after multiple-testing correction and five showed associations with DFS in Kaplan–Meier analyses. Higher intratumoral TCN1 variability was associated with longer DFS in MAC, whereas greater MUC2 variability, higher tumor-region nuclear counts, and higher stromal-compartment MUC1-associated positive nuclear counts were associated with longer DFS in NMAC. The latter two features remained independently associated with DFS after multivariable adjustment in the overall cohort. In addition, exploratory treatment-stratified analyses showed that adjuvant chemotherapy was associated with longer DFS in MAC patients with low MUC2 variability, whereas no similar association was observed in NMAC patients. Conclusion Our transcriptome-guided, AI-assisted spatial mIF profiling revealed the value of examining protein signal variability and regional distribution beyond mean intensity alone, providing further insight into differences between rectal MAC and NMAC and their potential prognostic relevance. These findings, particularly the exploratory observations in MAC, require validation in larger, independent cohorts.

Authors

Institutions

Publication Details

Journal
Journal of Translational Medicine
Published
2026-11-17
DOI
https://doi.org/10.1186/s12967-026-09242-3
Primary Topic
Colorectal Cancer Surgical Treatments
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Histology-related clinical significance of artificial intelligence-assisted multiplex immunofluorescence features in rectal mucinous and non-mucinous adenocarcinoma

Charlotte Stadler, Karolina Vernmark, Annika Knutsen, Chuanwen Fan et al.
Journal of Translational Medicine
Colorectal Cancer Surgical Treatments
article

Histology-related clinical significance of artificial intelligence-assisted multiplex immunofluorescence features in rectal mucinous and non-mucinous adenocarcinoma

Charlotte Stadler, Karolina Vernmark, Annika Knutsen, Chuanwen Fan, Yajun Li, Xiao-Feng Sun, Ke Zhao
article en

Abstract

Abstract Background Mucinous adenocarcinoma (MAC) and non-mucinous adenocarcinoma (NMAC) are distinct histological subtypes of rectal cancer, but the spatial patterns and clinical relevance of MAC-associated biomarkers remain poorly defined. Methods Transcriptomic data from TCGA-READ ( n = 177) and the Swedish Uppsala-READ cohort ( n = 281) were analysed using weighted gene co-expression network analysis to identify MAC-associated candidate markers. Selected markers were subsequently evaluated by multiplex immunofluorescence (mIF) using the COMET platform in an independent Southeast Swedish–READ cohort of 155 patients with rectal adenocarcinomas. Associations of mIF-derived image features with histological subtype and disease-free survival (DFS) were assessed using group comparisons, Kaplan–Meier analysis, and Cox regression. Results MAC-associated transcriptomic modules were identified in TCGA- and Uppsala-READ cohorts, with 109 shared genes. MUC1, MUC2, and TCN1 were selected for mIF analysis. A customized AI-assisted pipeline was developed and generated 79 spatial features from mIF images, of which 9 differed between MAC and NMAC after multiple-testing correction and five showed associations with DFS in Kaplan–Meier analyses. Higher intratumoral TCN1 variability was associated with longer DFS in MAC, whereas greater MUC2 variability, higher tumor-region nuclear counts, and higher stromal-compartment MUC1-associated positive nuclear counts were associated with longer DFS in NMAC. The latter two features remained independently associated with DFS after multivariable adjustment in the overall cohort. In addition, exploratory treatment-stratified analyses showed that adjuvant chemotherapy was associated with longer DFS in MAC patients with low MUC2 variability, whereas no similar association was observed in NMAC patients. Conclusion Our transcriptome-guided, AI-assisted spatial mIF profiling revealed the value of examining protein signal variability and regional distribution beyond mean intensity alone, providing further insight into differences between rectal MAC and NMAC and their potential prognostic relevance. These findings, particularly the exploratory observations in MAC, require validation in larger, independent cohorts.

Journal of Translational Medicine
Linköping University (SE), Sichuan University (CN), Science for Life Laboratory (SE), West China Hospital of Sichuan University (CN), Guangdong Provincial People's Hospital (CN), KTH Royal Institute of Technology (SE)
Colorectal Cancer Surgical Treatments
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon