Photo Credit: DedMityay

Imaging phenotypes of endometrial cancer (EC) are associated with clinical, pathologic, and molecular characteristics and disease-free survival (DFS), according to a retrospective study published in Magnetic Resonance Imaging. Researchers analyzed MRI radiomics features and clinical data from 356 patients with endometrial cancer. Using unsupervised machine learning, the researchers categorized patients into two imaging phenotypes. Phenotype 1 exhibited a stronger correlation with adverse features, including deep myometrial invasion (33.7% vs 13.0%), lymphovascular space invasion (23.8% vs 9.2%), cervical stromal invasion (16.3% vs 3.8%), aggressive histology (36.0% vs 17.4%), and advanced FIGO stage (43.6% vs 22.3%) compared with phenotype 2 (all P<0.001). Additionally, phenotype 1 was associated with higher rates of p53 overexpression (20.2% vs 8.5%; P=0.022). Survival analysis revealed that phenotype 1 patients had significantly worse DFS (log-rank P=0.002). The authors concluded that MRI radiomics-based imaging phenotypes can aid in preoperative risk stratification for EC.