Prediction of pathological tumor volume in clinically localized prostate cancer: Value of endorectal coil magnetic resonance imaging

The purpose of this study was to determine whether endorectal coil magnetic resonance imaging (MRI) enables accurate assessment of pathologic tumor volume in patients with clinically localized prostate carcinoma. Twenty-four patients with biopsy-proved prostate carcinoma underwent MRI at 0.5 T before radical prostatectomy. Tumor volumes were determined independently on axial fast-spin-echo (SE) T2-weighted MR images and whole- mount pathology slides of the surgical specimens. At pathology, tumor volumes ranged from 0.17 to 9.42 cm3 (mean ± SD, 3.11 ± 2.99 cm3): A strong correlation (r = .944) was found between measurements of tumor volume based on MR images and pathological specimens. The error was less than 0.5 cm3 in 14 cases, in the range of 0.5-1 cm3 in 7 cases, and more than 1 cm3 in 3 cases. By using an MR tumor volume of 2 cm3 as cutoff value, extracapsular tumor spread could be predicted with a sensitivity of 81.2%, a specificity of 100%, and an accuracy of 87.5%. Tumor volume determinations based on MR images seem to be accurate enough to be helpful in clinical decision-making.