Multimodal biomarkers and prediction models for stricturing and penetrating progression in Crohn's disease.

Crohn's disease can progress from non-stricturing, non-penetrating inflammation (B1) to stricturing (B2) or penetrating (B3) complications, yet risk is commonly recognized only after structural damage is established. This narrative review examines clinical, endoscopic, genomic, circulating, stool, tissue, imaging, radiomic, and computational biomarkers relevant to behavioral-progression risk and separates direct B1-to-B2 or source-defined B3 evidence from related prognostic and mechanistic evidence, while distinguishing internal penetrating from perianal fistulizing disease when source reporting permits. Structured searches of PubMed/MEDLINE, Embase, and Web of Science, supplemented by citation tracking, informed the selection of longitudinal human studies, prediction models, and mechanistic studies relevant to candidate feature selection. Ileal involvement, perianal fistulizing disease, persistent inflammatory burden, and endoscopic severity provide reproducible risk enrichment but insufficient individual-level precision. Genomic scores, plasma and stool proteins, tissue transcriptomics, shear-wave elastography, radiomics, and machine-learning models suggest potential value for multimodal risk stratification, but most lack independent external validation, calibration, transportability, and demonstrated incremental value over routinely available clinical variables. Direct prospective evidence is substantially more developed for B2 stricturing progression than for incident internal B3 penetrating progression; internal penetrating disease and perianal fistulizing disease are treated as distinct outcomes. Mechanistic studies suggest divergent fibrostenotic and tissue-destructive programs that provide biological guidance for biomarker selection. No biomarker or multimodal model identified in this review was ready to support routine biomarker-guided treatment escalation. Future studies should use well-characterized B1 baselines, prespecified structural outcomes, and prospective decision-impact testing that quantifies overtreatment.