The role of the mucinous component as an independent predictive factor in colorectal adenocarcinoma therapy: The potential of artificial intelligence as an adjunct tool to improve treatment outcomes

Mucinous adenocarcinoma of the colon (MAC) is a distinct histological subtype of colorectal cancer (CRC), defined by the presence of ≥ 50% extracellular mucin. This CRC subtype exhibits unique clinical and molecular characteristics, including more frequent localization in the right colon, higher prevalence among younger patients, and associations with microsatellite instability-high (MSI-H) and BRAF gene mutations. Evidence suggests that MAC demonstrates a poorer response to conventional therapies; however, the independent prognostic value of the mucinous component remains unclear. This narrative review aims to evaluate the prognostic significance of the mucinous component in CRC. Twenty relevant studies published between 2018 and 2025 were analyzed. Results indicate that MAC is more often detected at advanced stages and shows a weaker response to fluorouracil-based regimens and neoadjuvant chemoradiotherapy in rectal tumors. Nonetheless, several studies reported no significant difference in overall survival between mucinous and non-mucinous carcinomas. Additionally, immune-related factors such as tumor-infiltrating lymphocytes (TILs) and desmoplasia are gaining importance and may have greater prognostic value than mucinous differentiation itself. MAC also frequently exhibits marked molecular heterogeneity, complicating prognosis and treatment decisions. Consequently, careful patient monitoring and timely molecular profiling are essential for optimizing therapy. Novel artificial intelligence-based models that integrate histological images, molecular biomarkers, and clinical data show promise for personalizing treatment in MAC. In conclusion, although MAC is currently treated according to standard CRC guidelines, its distinct features underscore the need for individualized therapeutic approaches and further clinical research.