Transrectal ultrasound in the evaluation of cervical carcinoma and comparison with spiral computed tomography and magnetic resonance imaging.

38 women with biopsy proven untreated cervical carcinoma were prospectively studied with transrectal ultrasound (TRUS), spiral computed tomography (SCT) and magnetic resonance imaging (MRI). 20 women had radical hysterectomy and pelvic lymphadenectomy with detailed histological evaluation of the parametra. The echographic features of cervical carcinoma on TRUS are a hypoechoic (60%) or isoechoic (40%) (relative to normal uterine muscle/cervical stroma), poorly defined mass lesion with indistinct margins in an enlarged cervix. This relatively high percentage of isoechoic tumours and relative lack of contrast resolution may pose a problem in the identification of some tumours, and to our knowledge has not been previously reported. Further limitations of TRUS are in the evaluation of advanced cervical cancer, due to bulky tumours rendering poor access to the parametrium and pelvic sidewall. The overall accuracy in staging of early cervical cancer (less than stage 2b) was 85% for examination under anaesthesia (EUA), 75% for TRUS, 65% for MRI and 50% for SCT. The positive predictive value in evaluating the parametra in this group of patients was also lower for SCT (14%) and MRI (33%) compared with TRUS (100%). In the evaluation of advanced cervical cancer (stage 2b or higher), there was poor correlation between TRUS and EUA, with MRI showing the best correlation with EUA. We conclude that SCT is inferior to both TRUS and MRI in the staging of early stage cervical cancer.