Magnetic resonance imaging and transvaginal ultrasonography for the diagnosis of adenomyosis

Objective: To compare the diagnostic potential of magnetic resonance imaging (MRI) and transvaginal ultrasonography (TVS) in the diagnosis of adenomyosis. Design: Double blind set-up. Setting: University medical school. Patient(s): We studied 106 consecutive premenopausal women who underwent hysterectomy for benign reasons. Intervention(s): Transvaginal ultrasonography and MRI were compared with histopathologic examination as the golden standard. Main Outcome Measure(s): Adenomyosis. Result(s): Twenty-two (21%) patients had adenomyosis. The sensitivity and specificity were as follows: sensitivity: MRI 0.70 (0.46-0.87) and TVS 0.68 (0.44-0.86) (P=.66); specificity: MRI 0.86 (0.76-0.93) and TVS 0.65 (0.50-0.77) (P=.03). The combination of MRI and TVS was most sensitive (0.89 [0.64-0.98]), but produced the lowest specificity (0.60 [0.44-0.73]). Adenomyosis was not detected by either TNM or TVS at uterine volumes >400 mL. Exclusion of uteri >400 mL from the analysis improved the diagnostic precision of MRI, but not that of TVS. The diagnostic accuracy at MRI was improved by calculating the maximum difference between the thinnest and thickest junctional zone (JZdif) (i.e., greater than or equal to5-7 mm). Conclusion(s): Magnetic resonance imaging was superior to TVS for the diagnosis of adenomyosis. Magnetic resonance imaging had a higher specificity than TVS, but their sensitivities were in line. The diagnostic accuracy of MRI, as that of TVS, was at an intermediate level, but the diagnostic accuracy of the former improved by exclusion of uteri >400 mL. The combination of MRI and TVS produced the highest level of accuracy for exclusion of adenomyosis, but the low specificity may necessitate further investigation of positive findings, Measurement of the difference in junctional zone thickness may optimize the diagnosis of adenomyosis at MRI. (Fertil Steril(C) 2001;76:588-94. (C) 2001 by American Society for Reproductive Medicine).