Topographical sensitivity and specificity of endorectal coil magnetic resonance imaging for prostate cancer detection.

OBJECTIVES

Endorectal coil magnetic resonance imaging (EC-MRI) is useful to evaluate prostate cancer localization. Herein, we evaluate sensitivity and specificity of EC-MRI in different regions of the prostate by comparing the acquired images to whole-mount sections of the prostate after radical prostatectomy.

METHODS

69 patients with localized prostate cancer were included. After virtually dividing the prostate into 12 sectors, results of EC-MRI were compared to corresponding whole-mount sections by contingency analysis. Sensitivity and specificity were calculated for each of the 12 areas as well as for the dorsal and ventral region.

RESULTS

Sensitivity right/left was dorsal apex/mid/base 41/41, 60/67 and 73/79%; ventral 33/52, 43/42 and 47/52%. Specificity right/left was dorsal apex/mid/base 92/89, 82/75 and 88/69%; ventral 100/100, 100/92 and 88/83%. Local sensitivity and specificity regarding dorsal versus ventral was 88/100 and 65/87%.

CONCLUSIONS

Local sensitivity decreased from basodorsal to apicoventral direction, whereas local specificity increased in the same direction. Therefore, prostate cancers demonstrated by MRI are more prone to be detected in the basodorsal region, whereas less false-positive results are found in the apicoventral region. These variations in topographical specificity and sensitivity need to be considered before radical prostatectomy or MRI-guided biopsy.