EP22.08: Does the prevalence of deep myometrial infiltration affect the diagnostic accuracy of transvaginal ultrasound and MRI in endometrial cancer?

It has been demonstrated that there is no significant difference between the specificities and sensitivities of transvaginal ultrasound (TVS) and magnetic resonance imaging (MRI) in deep myometrial infiltration (DMI) detection. However, literature suggests that DMI prevalence might influence diagnostic accuracy. This study aims to compare TVS and MRI in different prevalence groups of DMI. A comprehensive search was conducted in three medical databases for original articles comparing TVS to MRI on the same patient cohort with final histopathological confirmation of DMI as the reference standard. Positive and negative predictive values were calculated for both imaging methods across different DMI prevalence groups. In the first group (<30% of patients with DMI), TVS had a positive predictive value (PPV) of 42% (Confidence interval (CI) 95%, 28-58%) and MRI had 55% (CI 95%, 36-72%), with non-significant differences (p = 0.296). Negative predictive values (NPV) for TVS and MRI were 91% (CI 77-96%) and 94% (CI 84-98%), respectively, also non-significant (p = 0.066). In the second group (30-40%), TVS exhibited a PPV of 79% (CI 95%, 66-87%) compared to MRI's 74% (CI 95%, 65-80%) (p = 0.588). NPV were also comparable at 86% (81-90%) for TVS and 86% (79-91%) for MRI (p = 0.580). In the third group (40-50%) TVS showed a PPV of 77% (CI 95%, 61-88%), while 78% (CI 95%, 63-89%) value was for MRI (p = 0.813). NPV for both modalities were also similar at 83% (69-91%) for TVS and 88% (75-95%) for MRI (p = 0.328). In the fourth group (>50%) TVS demonstrated a PPV of 96% (CI 95%, 9-100%) compared to 86% (CI 95%, 64-96%) of the MRI (p = 0.408). NPV were also comparable at 87% (63-96%) for TVS and 84% (59-95%) for MRI (p = 0.995). The prevalence of DMI does not significantly influence the diagnostic accuracy of TVS and MRI in endometrial cancer. Both modalities demonstrate comparable PPV and NPV across different prevalence ranges. Further prospective studies are needed to validate these findings.