Preoperative Magnetic Resonance Imaging for Assessment of Myometrial and Cervical Invasion in Endometrial Cancer: Is Gadolinium Necessary?

Background: Multiparametric magnetic resonance imaging (mpMRI), including dynamic contrast-enhanced (DCE) imaging, can assess deep myometrial invasion (DMI) and cervical stromal invasion (CSI) in endometrial cancer (EC). However, the need for gadolinium injection increases potential risks. Thus, whether comparable diagnostic performance can be achieved without gadolinium becomes a critical clinical question. Methods: We retrospectively analyzed 138 patients with EC from the first hospital of Putian City. T2-weighted imaging (T2WI) was combined with diffusion-weighted imaging (DWI) and DCE to form the T2&DWI and T2&DCE groups, respectively. Two senior radiologists independently evaluated each image set according to standardized criteria and were blinded to the pathology. A case was categorized as positive if either DMI or CSI was identified on any sequence within the image set. A more senior radiologist resolved any discrepancies. Two additional combinations were formed to generate a high-sensitivity group and a high-specificity group for secondary analysis. The diagnostic performance of the four image sets was evaluated using postoperative pathology as the reference standard. Results: Among all cases, 31 had DMI, and 12 had CSI. Agreement between the two readers for the same sequence combination and between sequence combinations for the same reader was high. For DMI, the sensitivity, specificity, accuracy, and receiver operating characteristic (ROC) analysis for the area under curve (AUC) were as follows: T2&DWI group, 0.806, 0.935, 0.906, and 0.871; T2&DCE group, 0.710, 0.935, 0.884, and 0.822; high-sensitivity group, 0.839, 0.916, 0.899, and 0.877; high-specificity group, 0.774, 0.944, 0.906, and 0.811. For CSI, the corresponding results were: T2&DWI group, 0.833, 0.984, 0.971, and 0.909; T2&DCE group, 0.917, 0.984, 0.978, and 0.950; high-sensitivity group, 0.917, 0.976, 0.971, and 0.946; high-specificity group, 0.833, 0.992, 0.978, and 0.913. DeLong’s pairwise comparisons showed no statistically significant differences (p > 0.05), with the Holm correction for multiple comparisons yielding consistent results. The McNemar test for paired comparisons among the four sequence combinations also revealed no significant differences between the groups. Conclusion: The T2&DWI group demonstrated diagnostic performance comparable to that of the other groups to evaluate DMI and CSI, with no statistically significant differences, suggesting it may serve as a non-contrast alternative to mpMRI with DCE in specific clinical settings.