Prostate cancer: correlation of MR images with tissue optical density at pathologic examination.

To explain the variability in detection of prostate cancer with magnetic resonance (MR) imaging, the authors correlated preoperative MR findings in 28 patients with tissue optical density (TOD) measurements on whole-mount pathologic slides prepared from radical prostatectomy specimens. TOD was used as an indicator of the degree of tissue compactness or openness. TOD measurements from proved cancers and from pathologic regions corresponding to MR lesions (areas of low signal intensity seen at T2-weighted MR imaging) were compared with TOD measurements from adjacent, nonmalignant tissue. TOD measurements corresponding to MR lesions were higher than noncancerous tissue measurements in all cases (P less than .005). Although most of these lesions represented cancers (21 of 30), nine of 30 represented benign tissue that was composed mainly of densely packed fibromuscular stroma (30% false-positive results). Thus, signal intensity appeared to be related to TOD rather than to a specific histologic tissue type, and the finding of a peripheral zone lesion with low signal intensity did not necessarily indicate the presence of a cancer.

Carcinoma of the prostate: MR images obtained with body coils do not accurately reflect tumor volume.

MR imaging with a body coil is unreliable in directly demonstrating tumor spread through the prostatic capsule. However, the likelihood of extracapsular spread of prostatic cancer rises with increasing tumor volume. The aim of our study was to assess the accuracy of MR with a body coil in diagnosing capsular penetration indirectly via an estimation of prostatic tumor volumes. Twenty-six patients with proved prostatic cancer that was clinically confined to the gland underwent MR imaging before radical prostatectomy and whole-mount pathologic sectioning of the specimen. Twenty of 31 lesions prospectively outlined on the MR images corresponded to cancers outlined on the pathology slides, and tumor volumes were calculated by using a voxel summation technique. On MR, tumor volume was underestimated in 11 of 20 cases and overestimated in nine of 20 cases. Only two of 20 size estimates based on MR findings were within 10% of actual tumor volume. Overlap in MR tumor volumes was significant between lesions with and without capsular penetration at microscopy. Factors contributing to inaccuracies in measurements of tumor volume on MR images included the variable histologic make-up of the tumors. Our results show that, although knowledge of the size of a prostatic lesion is important in predicting the behavior of the tumor, MR imaging with a body coil is not reliable for accurate estimation of tumor volume.