Objective: To evaluate the effects of previous cone biopsy and lesion size on detectability of stage 1a/1b cervical cancer using endovaginal T2- and diffusion-weighted magnetic resonance imaging.
Methods: One hundred and thirteen patients with cervical tumor were imaged using an endovaginal coil with T2-weighted (T2-W) and diffusion-weighted single-shot echo-planar sequences; 85 managed surgically (58 with prior cone biopsy/LLETZ) were evaluated. T2-W images and ADC maps viewed simultaneously were scored positive or negative for tumor and compared with histology at surgery. MRI tumor volumes, maximum radiological and histological dimensions were recorded. ROC analysis determined the MRI volume with optimal sensitivity/specificity for identifying tumor in those without and with prior cone biopsy/LLETZ and the maximum histological dimension for correctly identifying tumor with MRI. Mean apparent diffusion coefficients (ADCs) from tumor and adjacent normal epithelium were compared.
Results: Sensitivity and specificity for detecting tumor in those without (100%; 100% respectively) and with (80%; 78.9% respectively) prior cone biopsy/LLETZ were significantly different (p<0.001). Following cone biopsy/LLETZ, MRI tumor volume of 83 mm3 detected tumor with 80% sensitivity, 94.7% specificity; a 5.3mm maximal histological dimension was detected on MRI with 100% sensitivity, 100% specificity. Tumor ADCs were significantly lower (p<0.001) than paired normal epithelial tissue (median, 988×10(-6) mm2/s vs. 1564×10(-6) mm2/s) but neither tumor nor epithelial ADCs differed significantly between patients with or without prior cone biopsy/LLETZ (p=0.48 and 0.15, respectively).
Conclusions: Endovaginal MRI with T2- and diffusion-weighted sequences has significantly lower sensitivity and specificity for tumor detection following cone biopsy/LLETZ.
Copyright © 2010 Elsevier Inc. All rights reserved.