Abstract Background Crohn’s disease (CD) leads to fistula formation in about 40% of patients due to persistent transmural inflammation. Despite advanced therapies, recurrence remains frequent and the molecular mechanisms remain poorly defined, partly because intact tracts are difficult to profile and highly heterogeneous. We integrated multi-modal spatial transcriptomics, single-cell profiling and matrix imaging to define the cellular and molecular architecture of intestinal fistulae. Methods We constructed a comprehensive atlas of 92 human intestinal fistulae and control tissues using Visium spatial transcriptomics, subcellular-resolution Xenium profiling (480 and 5100-plex), single-cell RNA sequencing, multiplexed immunohistochemistry and quantitative collagen imaging. Over 90,000 spatial transcriptomic spots and seven million segmented cells were analysed. Computational integration mapped fistula-linked single-cell states and conserved cellular niches across diverse lesions, including CD and non-CD fistulae. Functional consequences of aberrant fibroblast re-programming was explored via transduction of primary colonic fibroblasts with transcriptional regulators we identified as abnormally induced in fistula tracts. Results We identified a fibroblast population uniquely associated with fistulae, marked by TWIST1, RUNX2, OSR2 and PRRX1, termed fistula-associated stromal (FAS) fibroblasts. These cells expressed developmental morphogens (e.g., WNT5A) and matrix-remodelling mediators (MMP1/3/9, IL11, COL7A1) promoting invasion, propagation and fibrosis. In vitro assays showed that TWIST1 and OSR2 drive distinct functional programs in FAS fibroblasts. Spatial analyses revealed concentric FAS zones around tracts: a proliferative lumen-adjacent zone (FAS-LAZ), a matrix-remodelling active lesion core (FAS-ALC), and a fibrotic outer zone (FAS-FOZ). Collagen mapping demonstrated matching extracellular matrix zonation, from hypodense collagen at proliferative areas to high anisotropy in fibrotic zones. Stromal–immune co-localisation identified interactions between FAS fibroblasts and SPP1+ macrophages driving immune recruitment (CXCL5/9) and matrix remodelling. FAS-like fibroblasts were also found at ulcer bases in non-fistulating CD. Conclusion This study defines fistula-associated stromal and immune states as key mediators of invasion, immune crosstalk and fibrotic remodelling. Aberrant developmental transcriptional programmes in FAS cells lead to abnormal zonation of growth factors and morphogens, reshaping collagen architecture and sustaining fistula persistence over normal healing. These findings provide new mechanistic insights into fistula biology and information to guide model design and preventatives. Conflict of interest: McGregor, Colleen: No conflict of interest Qin, Xiao: No conflict of interest Jagielowicz, Marta: No conflict of interest Gupta, Tarun: No conflict of interest Yin, Zinan: No conflict of interest Lentsch, Verena: No conflict of interest Fawkner-Corbett, David: No conflict of interest Lai, Vy: No conflict of interest Gomez Castro, Paula: No conflict of interest Bridges, Esther: No conflict of interest Lee, Chloe H: No conflict of interest Chuang, Huei-Wen: No conflict of interest Deng, Lei: No conflict of interest Aulicino, Anna: No conflict of interest Teague, Renuka: No conflict of interest Moradi, Sorayya: No conflict of interest Park, Jun Sung: No conflict of interest Woo, Jeongmin: No conflict of interest Xu, Kexin: No conflict of interest Tandon, Ruchi: No conflict of interest Cianci, Nicole: No conflict of interest Bornschein, Jan: No conflict of interest Ho, Ling-Pei: No conflict of interest Siejka-Zielinska, Paulina: No conflict of interest Christoforidou, Zoe: No conflict of interest Hill, Sarah: No conflict of interest Lehmann, Johannes: No conflict of interest Kujawa, Rhea: No conflict of interest Vargas Gutierrez, Paola: No conflict of interest Cheng, Carol: No conflict of interest Greco, Maria: No conflict of interest Baker, Katherine: No conflict of interest Bignell, Mark: No conflict of interest George, Bruce: No conflict of interest Fryer, Eve: None Vieth, Michael: No conflict of interest Antanaviciute, Agne: No conflict of interest Simmons, Alison: No conflict of interest
McGregor et al. (Thu,) studied this question.