Abstract Background Progression to colitis-associated colorectal cancer among Inflammatory Bowel Disease (IBD) patients remains difficult to predict. Whilst genetic changes such as copy number alteration (CNA) burden is a strong predictor of advanced neoplasia (AN) in IBD patients with low-grade dysplasia (LGD), the role of the immune microenvironment remains underexplored. 1 Here, we examine both the genetics and the microenvironment of four IBD-LGD cohorts. Methods Using a convolutional neural network-based cell classifier on pathologist-annotated haematoxylin and eosin-stained tissue sections, we assessed the cellular immune microenvironment, quantifying neutrophils, lymphocytes, plasma cells, eosinophils and macrophages, in LGD and matched non-dysplastic tissue in three UK cohorts (St Marks, Oxford, Newcastle, total n = 128 Ulcerative Colitis (UC) patients, n = 117 LGD lesions), and a Dutch validation cohort (n = 52 UC and Crohn’s colitis patients, n = 82 LGD lesions). All samples underwent lpWGS, followed by copy number and MSI calling. 1, 2 Neutrophil elastase protein expression was digitally quantified in 22 samples, another 8 samples underwent comprehensive cell phenotyping using 37 markers via multiplex immunofluorescence (mIF) on the Akoya Phenocycler. 3 The T-cell receptor (TCR) repertoire was characterised in 17 samples using FUME-TCRseq. 4 Results Neutrophil density was significantly higher in LGD lesions of Progressors (P) than in LGD lesions of Non-Progressors (NP) (p = 0. 017, St. Marks), and a non-significant trend was also observed in the Dutch validation cohort (p = 0. 088). Interestingly, the validation cohort showed that macrophage density was significantly higher in P (p = 0. 002). There was also a trend towards higher neutrophil elastase-positive cell counts in progressor LGD (p = 0. 073). Detailed immune phenotyping by mIF revealed a shift towards myeloid cells, particularly CD68+ iNOS- macrophages (p = 0. 036, 3 P vs. 5 NP), and fewer CD38+ plasma cells in P. There were no significant differences in T-cell subtypes by mIF, but TCRseq revealed a trend towards higher unique clonotype counts in P (p = 0. 053). Lastly, we validated our previous genomic findings in this independent Dutch validation cohort, finding a significantly higher number of segments of genome altered in progressor lesions (p 0. 001). Survival analysis revealed that higher macrophage and neutrophil density in LGD was associated with a shorter time to progression (p 0. 001; p = 0. 05; Utrecht). Conclusion In the IBD-LGD setting, P lesions exhibit higher neutrophil, and macrophage densities, alongside a myeloid-skewed environment with immunosuppressive M1 polarisation of macrophages. Combining genomic alterations with immune features improves prediction of progression. References: 1. Bakir IA, Curtius K, Cresswell GD, et al. Low-coverage whole genome sequencing of low-grade dysplasia strongly predicts advanced neoplasia risk in ulcerative colitis. Published online January 29, 2025. doi: 10. 1136/gutjnl-2024-333353 2. Guo Q, Househam J, Lakatos E, et al. Long deletion signatures in repetitive genomic regions track somatic evolution and enable sensitive detection of microsatellite instability. bioRxiv. Preprint posted online October 4, 2024: 2024. 10. 03. 616572. doi: 10. 1101/2024. 10. 03. 616572 3. Donovan ML, Jhaveri N, Ma N, et al. Protocol for high-plex, whole-slide imaging of human formalin-fixed paraffin-embedded tissue using PhenoCycler-Fusion. STAR Protoc. 2024;5 (3): 103226. doi: 10. 1016/j. xpro. 2024. 103226 4. Baker AM, Nageswaran G, Nenclares P, et al. FUME-TCRseq Enables Sensitive and Accurate Sequencing of the T-cell Receptor from Limited Input of Degraded RNA. Cancer Res. 2024;84 (10): 1560-1569. doi: 10. 1158/0008-5472. CAN-23-3340 Conflict of interest: Bräutigam, Konstantin: KB was funded by the Swiss National Science Foundation (P500PM₂17647 / 1). Mandal, Soham: No conflict of interest Grant, Heather: No conflict of interest Trahearn, Nicholas: No conflict of interest Mossner, Max: No conflict of interest Yasmin, Suhana: No conflict of interest Louwers, Jonas: No conflict of interest Guo, Qingli: No conflict of interest Sakr, Chirine: No conflict of interest Andersen, Jesper Winkler: I have no conflicts of interest to declare Fisher, Jennifer: No conflict of interest Curtius, Kit: No conflict of interest Oldenburg, Bas: No conflict of interest Leedham, Simon: No conflict of interest Hart, Ailsa: Grant: Takeda Personal Fees: Abbvie, Amgen, Arena, AZ, Falk, Celltrion, Eli Lilly, Ferring, Genentech/ Roche, GSK, Pfizer, Takeda, Napp, Pharmacosmos, Janssen (J & J), Bristol-Myers Squibb, Gilead, Galapagos, Alfasigma Baker, Ann-Marie: No conflict of interest Graham, Trevor: Personal Fees: I have received speaking fees from Genentech in 2023, on a topic unrelated to work I will present at ECCO.
Bräutigam et al. (2026) studied this question.