Transrectal ultrasonography: why are estimates of prostate volume and dimension so inaccurate?

To determine why there are variations in the volumes of the prostate obtained by step planimetry and those calculated from the prolate ellipsoid formula using the measured maximum gland dimensions, to assess the reproducibility of estimates of prostate volume from transrectal ultrasonography (TRUS) and to determine the effect of the angle of the transrectal probe, the human error in the clinical setting and the benefit of computer enhancement of the ultrasonograms.

Forty-five men (mean age 72 years, range 43-89) with symptoms of prostatic enlargement were divided randomly into three groups; those in group 1 had their prostate volume estimated three times by measuring the maximal gland dimensions and calculating the volume using the prolate ellipsoid formula (dimensional method) and by step planimetry, to assess the reproducibility of TRUS; men in group 2 had their prostate volume estimated using the dimensional method and by step planimetry with the probe in the optimal axis, and then from scans repeated after deflecting the transrectal probe 5 degrees anteriorly and posteriorly; men in group 3 had their prostate volume estimated once by the dimensional method and by step planimetry. Step planimetry was carried out using a specially designed indexer firmly attached to the examination couch. All 'frozen' images and planimetry slices were videotaped for later computer enhancement. The volumes and dimensions obtained by each method were compared.

The volumes obtained by step planimetry were 17% greater than those obtained by the dimensional method because the craniocaudal, anteroposterior and transverse diameters were underestimated significantly by the latter (by 13%, 2% and 7%, respectively). The largest anteroposterior and transverse diameters were included in the same axial scan in only 44%, while the largest craniocaudal diameter was in the midline in only 38%, of the patients. The human error in the clinic was negligible (2%) with a good correlation between the dimensions and volumes obtained in the clinic and those from computer-enhanced images (r = 0.94). There were no advantages in computer enhancing the TRUS images. Volumes calculated from the dimensional method using the maximum dimensions obtained from planimetric contours correlated well with planimetric volumes (r = 0.93). The reproducibility of TRUS showed a mean error of 8% by the dimensional method and 1% by step planimetry. Anterior deflection of the transrectal probe reduced the estimated volume by 2% and posterior deflection increased the volume by 4%, using the dimensional method, while there were no significant volume changes when estimated by step planimetry.

This study confirms that to estimate accurately the volume of the prostate using the prolate ellipsoid formula, the current methodology needs to be changed. The largest anteroposterior and transverse diameters may need to be measured in different transverse scan slices and the largest craniocaudal diameter in a sagittal scan away from the midline. If volume estimation is to be repeated then step planimetry is reliable and TRUS using the prolate ellipsoid formula is not.