Prostate Cancer and Prostatic Diseases (2002) 5, 273–278
ß 2002 Nature Publishing Group All rights reserved 1365–7852/02 $25.00
www.nature.com/pcan
Accuracy and repeatability of
prostate volume measurements by
transrectal ultrasound
LM Eri1*, H Thomassen1, B Brennhovd1 & LL Håheim2
1
Department of Urology, Aker University Hospital, Oslo, Norway; and 2Department of
Statistics, Ullevaal University Hospital, Oslo, Norway
We evaluated six alternative methods of prostate volume determination by
transrectal ultrasound, three based on planimetry and three based on measurement of prostate diameters. Prostate volume measurements were made on an
average of 6.5 occasions over a 3 y period on 41 patients with benign prostatic
hyperplasia, using standard techniques. We defined the average of multiple
planimetries as the prostate reference volume. Agreement with the reference
volume and reproducibility at repeat testing was in the same range for single
planimetry and volume determinations based on the formulas height (H)6width
(W)6length (L)6p/6 and W6W6H6p/6, but was poorer using the formula
W6W6W6p/6. Using the average result of two successive planimetry measurements increased the reproducibility of planimetry, being statistically significantly better than for one single planimetry (P ¼ 0.024) or for the formula
W6W6H6p/6 (P ¼ 0.048). Our study suggests that the simple formula based
methods of prostate volume determination provide results that are only marginally inferior to one single planimetry, but results are improved by performing
two planimetry measurements.
Prostate Cancer and Prostatic Diseases (2002) 5, 273–278. doi:10.1038/sj/pcan/4500568
Keywords: ultrasonography; prostate; reproducibility of results; planimetry;
prostate volume
Introduction
Prostatic volume estimation by transrectal ultrasound is a
common clinical procedure. Uses include pre-treatment
assessment of prostate size, interpretation of elevated
prostate specific antigen (PSA) levels (PSA density) and,
in the field of research, measurement of the effects of
prostate shrinking drugs.
Step section planimetry is assumed to be the most
accurate method of prostate volume determination, but
it is time consuming and requires cumbersome special
equipment.1 – 5 One-dimensional measurements are preferable in the clinic. The prolate elipsoid formula, multiplying the largest anterioposterior (height), transverse
(width) and cephalocaudal (length) prostate diameters
*Correspondence: LM Eri, MD, PhD, Department of Urology, Aker
University Hospital, 0514 Oslo, Norway.
Received 14 August 2001; revised 28 November 2001; accepted 12
December 2001
by 0.524 (H6W6L6p/6) is probably the most commonly used method,6,7 since it is rapid, reproducible,
and has been shown to have high correlation with the
actual prostate volume.5 However, it has been questioned
whether the prolate elipsoid formula gives accurate
results in large prostates.5,8 The prolate spheroid formula
W6W6H6p/6 might be equally accurate,5 and has the
advantage of requiring measurements in the transversal
plane only.
We present a methodological study with the purpose of
evaluating six alternative methods of prostate volume
determination, three based on planimetry and three
based on measurement of prostate diameters. We studied
agreement with the true prostate volume (average of multiple planimetries), and determined reproducibility of
repeat examinations performed at 8 – 24 weeks’ interval
for the various methods.
Apparent substantial intraindividual variations in
prostate volume estimates over time has been attributed
to retention of prostatic secretion, oedema, vascular congestion, etc.9 – 12 We evaluated whether reproducibility of
two volume determinations performed during the same
session was different from measurements performed 8 –
24 weeks apart.
Accuracy of prostate volume measurements
LM Eri et al
274
Materials and methods
Forty-one patients with benign prostatic hyperplasia
(BPH) who were placebo patients in a clinical trial of
hormonal treatment of BPH were studied. Prostate
volume measurements by transrectal ultrasound (TRUS)
were performed at 8 – 24 weeks’ interval on up to 10
occasions over a 3-y period, using the Bruel and Kjaer
1846 ultrasound unit and the transrectal transducer No
8531. We included all measurements performed on these
patients during the 3-y period. Some patients underwent
prostate surgery, and follow-up was discontinued for
some patients, thereby reducing the number of study
patients from 41 at the start of the study to eight by
week 144.
The transducer, surrrounded by a small water filled
balloon, was mounted on a fixing sledge, and scanning of
the prostate in the transverse plane was performed while
moving the transducer in the cephalocaudal direction of
the prostate. The height and width of the prostate section
with the greatest surface area was recorded. Planimetry
was performed by serial scanning at 5 mm steps. At each
section the circumference of the prostate was marked and
the prostate volume was calculated by the ultrasound
unit computer. Planimetry was performed twice in each
session, first from apex to base (first planimetry), and then
in the opposite direction (second planimetry). The equipment was not designed for scanning of the prostate in the
sagital plane. Therefore, prostate length was measured
indirectly by the stepping device by counting the number
of 5 mm steps from apex to base (N1), and then repeating
the procedure in the opposite direction (N2). Prostate
length in cm was calculated by the formula (N1 þ N2)/4.
All examinations were done by the same physician
(LME), within a tight time schedule, as in a routine
clinical situation. With the exception of the second of
the duplicate planimetries, all measurements were made
without knowledge of the previous result.
At each time point, six prostate volume estimates were
made in the following way: (1) the first planimetry, (2) the
second planimetry, (3) the average of the two planimetries, (4) the prolate ellipsoid formula H6W6L6p/6, (5)
the prolate spheroid formula W6W6H6p/6, and (6) the
spherical formula W6W6W6p/6.
The mean of all planimetry volume estimates for an
individual patient was assumed to be the best estimate of
the true prostate volume, and was defined as the prostate
reference volume. However, before accepting this assumption, we ascertained that the average prostate volume in
Table 1 Prostate volume of study patients
Week
Week
Week
Week
Week
Week
Week
Week
Week
Week
Prostate volume (ml) (mean s.e.m.)
No. of patients
58.1 4.4
57.5 5.7
57.5 4.7
57.5 4.8
58.6 4.7
56.7 5.9
56.2 4.5
56.2 9.9
60.3 13
62.5 18
41
26
37
36
32
30
20
18
14
8
0
8
16
24
36
48
72
96
120
144
Volumes were measured by duplicate planimetry.
Prostate Cancer and Prostatic Diseases
fact was stable during the study period for the patient
population (Table 1).
Agreement with the prostate reference volume for a
specific volume determination method was visualized by
scatter plots and by calculating Pearson’s coefficient of
correlation, r, which measures the strength of the relation
between the two series of measurements (Figure 1). A
value of r close to one indicates a strong linear relationship between two methods of volume determination, but
r will also be high if two measurements are equivalent but
are based on a different scale. Agreement was initially
calculated separately for each of the 10 time points, and
was similar throughout the study period. Therefore, for
ease of presentation, we combined the results of all time
points.
Repeatability of a specific method was evaluated by
comparing pairs of volume determinations performed at
successive time points; week 0 with week 8, week 16 with
week 24, etc. We calculated the mean and the standard
deviation of the differences between two successive measurements. For all parameters, results were similar over
the five time intervals, and therefore, for ease of presentation, data for all time points were combined.
The data generally had a symmetrical distribution, and
the two sample t-test was used for the statistics.
Results
Mean age of the 41 patients was 70.0 y (s.d. 7.2). On
average the three prostate diameters were measured at 6.5
time points (s.d. 2.3, range 1 – 10). The mean of all
measurements of prostate width was 5.3 cm (range 3.9 –
7.6), of prostate height was 3.5 cm (range 1.8 – 6.2) and of
prostate length was 4.9 cm (range 3.5 – 7.5). Planimetry
was performed an average of 12.6 times (s.d. 4.7, range
4 – 20) on each patient. Mean prostate volume by planimetry for the patients was 58.0 ml (s.d. 28.9 ml, median
47.9 ml, range 26.6 – 164.8 ml), being unchanged throughout the study period (Table 1).
Agreement with reference volume
Agreement between prostate reference volumes and the
volumes by planimetry or based on the formulas
H6W6L6p/6, W6W6H6p/6 or W6W6W6p/6
are illustrated in Figure 1. Average deviation from the
prostate reference volume was 7 1.1 ml for the first
planimetry, 1.1 ml for the second planimetry and
71.4 ml for the formula W6W6H6p/6. On average,
the formula H6W6L6p/6 underestimated prostate
volume by 5.7 ml (P < 0.0001) and the formula
W6W6W6p/6 overestimated it by 27 ml (Table 2, first
column). Deviations from the reference volume can also
be assessed by evaluating standard deviations of the
differences in Table 2. Standard deviation is based on
the squared differences, thereby eliminating the effect of
negative values. Results are presented separately for the
volume categories less than 40 ml, 40 – 80 ml and above
80 ml (Table 2, second to fourth columns). Part of the
agreement between single planimetries and the reference
volume is due to the fact that the reference volume is
Accuracy of prostate volume measurements
LM Eri et al
275
Figure 1 Agreement with prostate reference volume. Scattergrams of four methods of prostate volume determination vs an estimate of true prostate volume
(prostate reference volume, see text). Prostate volume is measured by one single planimetry (a), by the formula H6W6L6p/6 (b), by the formula
W6W6H6p/6 (c), and by the formula W6W6W6p/6 (d). On each figure, the Pearson’s coefficient of correlation (r), and the equation of the
regression line are presented. The line on the figures represent the line of equality (X ¼ Y).
Table 2 Agreement between five methods of prostate volume determination and prostate reference volume
Vol. category
Method of prostate volume determination
First planimetry
Second planimetry
H6W6L6p/6
W6W6H6p/6
W6W6W6p/6
Total
Average diff. s.d.(ml)
< 40 ml
Average diff. s.d.(ml)
40-80 ml
Average diff. s.d.(ml)
> 80 ml
Average diff. s.d.(ml)
7 1.1 7.0
1.1 6.8
7 5.7 6.9
7 1.4 8.9
27.1 19.5
7 0.9 3.4
1.0 3.4
7 3.7 3.4
7 1.1 4.8
20.1 11.6
7 1.7 5.8
1.7 5.0
7 5.5 6.1
7 2.0 7.3
27.4 18.9
7 0.4 12.7
0.3 12.7
7 9.9 10.8
7 0.3 15.8
37.9 25.5
Prostate reference volume was defined as the average of multiple planimetries. Average prostate reference volume ( s.d.) was 58.0 28.9 ml for all 41 prostates,
31.7 3.4 ml for 12 prostates less than 40 ml, 55.9 12.6 ml for 21 prostates 40 – 80 ml and 102.9 28.2 ml for eight prostates greater than 80 ml.
calculated as the average of multiple planimetries. Therefore, we abstained from evaluating statistically possible
differences in agreement with prostate reference volume
for the different volume determination methods.
Reproducibility
The relationships between pairs of successive measurements of prostate height, width and length are shown in
Figure 2 a – c. Standard deviation (s.d.) for the differences
was the same for measurements of prostate height and
width (0.32 cm), and was statistically significantly larger
for prostate length (s.d. 0.48, P ¼ 0.0003).
Reproducibility of prostate volume determinations by
planimetry is illustrated in Figure 3. Standard deviation
of the differences between pairs of measurements, performed at successive time points, was 8.71 ml for the first
planimetry and 7.83 ml for the second planimetry
(P ¼ 0.21). Using the average of two repeat planimetries
performed during the same session as the volume estimate, s.d. for the differences between volume determinations performed at succesive sessions was 6.45,
statistically significantly lower than based on one single
planimetry (P ¼ 0.024).
Reproducibility of the three formula derived volume
estimates, based on measurement of three, two or one
diameters, are illustrated in a similar way (Figure 4).
Standard deviation for the differences between pairs of
volume measurements for H6W6L6p/6 and
W6W6H6p/6 were 8.54 ml and 8.84 ml, respectively.
Reproducibility of estimates based on H6W6L6p/6
or W6W6H6p/6 was not statistically significantly
different from prostate volume determinations based
on one single (the first) planimetry, with P-values of
0.72 and 0.88, respectively (Figures 3 and 4). Volume
Prostate Cancer and Prostatic Diseases
Accuracy of prostate volume measurements
LM Eri et al
276
Reproducibility of prostate volume determinations
based on one single planimetry was not different from
those based on the formulas H6W6L6p/6 (P ¼ 0.72) or
W6W6H6p/6 (P ¼ 0.88). Reproducibility based on the
average of two planimetries was somewhat better than
based on the formula W6W6H6p/6 (P ¼ 0.048), but not
statistically significantly better than using the formula
H6W6L6p/6 (P ¼ 0.072).
Reproducibility of two repeat planimetries performed
during the same session was similar to the reproducibility
when performed at two separate sessions 8 – 24 weeks
apart (Table 3), suggesting that over a time span of this
magnitude fluctuation in prostate volume is small.
Discussion
Figure 2 Reproducibility of prostate diameters. Scattergrams of pairs of
measurements of prostate diameters by transrectal ultrasound, performed at
two successive examinations 8 – 24 weeks apart. Anterioposterior diameters
(a), transverse diameters (b) and cephalocaudal diameters (c) are measured.
The line on the figures represent the line of equality (X ¼ Y).
determinations based on the formula W6W6W6p/6
had significantly poorer reproducibility, with a s.d. of
15.73 ml (P < 0.001).
The most widely used formula derived method of prostate volume determination, H6W6L6p/6, is reported to
underestimate prostate volume by 7 – 27%.4,7,8 Among
our patients this method underestimated prostate
volume by an average of 5.7 ml (9.9%), whereas
W6W6H6p/6 underestimated the volume by 1.4 ml
(2.4%) and W6W6W6p/6 overestimated prostate
volume by 27.1 ml (47%) (Figure 1, Table 2).
Direct measurement of prostate volume can be performed on radical prostatectomy specimens. However,
this method does not necessarily determine the true in
vivo prostate volume, due to shrinkage because of loss of
blood supply,5 and the volume might also be overestimated because of adjacent periprostatic tissue.13 Several
methodological studies, like measurement of water filled
balloons,14 and cadaver studies13,15 have demonstrated
that under ideal conditions, planimetry measures a
volume very close to the true volume, and can be used
as the gold standard of prostate volume measurement.3
However, problems with planimetry measurements
include poor visibility of the prostate border and patient
movement. Ultrasound equipment that provides better
resolution and a more accurate delineation of the borders
of the gland will probably increase the accuracy of all
types of prostate volume measurement.
We registered a stable prostate volume during the
study period (Table 1), which was a precondition for
using the average of all measurements as the prostate
reference volume without correcting for change over time.
Figure 3 Reproducibility of prostate volume determinations by planimetry. Scattergrams of the difference between two successive volume estimates 8 – 24
weeks apart (Vol 2 – Vol 1) vs prostate reference volume. Prostate volumes are determined by the first of two repeat planimetries (a), the second of two repeat
planimetries (b) and the average of the two (c).
Prostate Cancer and Prostatic Diseases
Accuracy of prostate volume measurements
LM Eri et al
277
Figure 4 Reproducibility of formula derived prostate volume determinations. Scattergrams of the difference between two successive volume determinations
8 – 24 weeks apart (Vol 2 – Vol 1) vs prostate reference volume.
Table 3 Intraindividual
determinations
reproducibility
of
prostate
volume
Separate sessions
Same session
First
Second
Two
planimetry
planimetry
planimetries at two sessions at two sessions
Mean of differences (ml)
Standard deviation (ml)
No of observations
1.2
8.1
252
7 1.9
8.7
116
0.6
7.8
102
Mean and standard deviation of differences between pairs of planimetries on
the same patient performed during the same session or at two separate
sessions 8 – 24 (mean 12.6) weeks apart.
However, the prostates of these BPH patients most likely
increased by an average of 1 – 3% over the 3-y study
period.9,12,16 A change of this magnitude was less than
what we could expect to detect in this relatively small
patient population.
Prostate width and height were measured directly on
the ultrasound monitor and had similar reproducibility
(Figure 2a – b). Prostate length was measured by the
stepping device. This method has been suggested to be
more accurate than direct measurement of prostate length
in the sagital plane, because of better visualization of the
point of juncture between the prostatic apex and distal
urethra and better possibilities to distinguish the base of
the prostate from seminal vesicles and the bladder neck.5
However, in spite of this, reproducibility of prostate
length was poorer (P < 0.001) than prostate height and
width (Figure 2), confirming that prostate length is difficult to measure, irrespective of method.
Several investigators have reported an excellent reproducibility of prostate volume measurements by planimetry,2,4,17 and a better reproducibility of planimetry than
for volume determinations based on measurement of
prostate diameters.4,17,18 Terris et al, however, concluded
that volume determinations based on the formulas
H6W6L6p/6 and W6W6H6p/6 were comparable
to planimetry.4
Our study suggests that the simpler formula based
methods of prostate volume determination provide
results that are only marginally inferior to planimetry.
Of the formulae based on one dimensional measurements, W6W6H6p/6 might be preferable to
H6W6L6p/6, because only two measurements are
necessary. This formula does not contain prostate length
which is the least accurate of the three prostate diameters.
W6W6H6p/6 in this patient population resulted in an
average volume only 1.4 ml less than the prostate reference volume.
Using the average of two planimetries generally
resulted in improved reproducibility compared to methods based on one single planimetry or based on measurement of prostate diameters, and might be valuable in a
research setting. However, repeat measurements result in
increased time consumption and patient discomfort.
Average prostate volume (58.0 ml) for the patients in
the present study was larger than in an average population because a prostate volume above 30 ml was required
for inclusion. Because of possible differences of prostate
configuration and measurement quality depending on
prostate size, the usefulness of the various measurement
methods might vary depending on prostate size.5,8 However, our study does not suggest that any particular
method of volume determination was preferable in prostates of a specific size range (Table 2).
We found the same reproducibility irrespective of
whether planimetries were performed as two successive
prostate volume measurements during the same session
or with 8 – 24 weeks’ interval (Table 3). This finding
suggests that a possible biological fluctuation of prostate
volume over time, as proposed by some authors,11 is due
to measurement error.
Conclusions
Our study suggests that the simple formula based methods of prostate volume determination, based on prostate
diameters, provide results that are only marginally inferior to planimetry, and are preferable in the clinic because
they are simpler to perform and are associated with less
patient discomfort. Of the one dimensional measurement
methods, H6W6L6p/6 (0.524) and W6W6H6p/6
are almost equivalent. Our study does not support the
assumption of significant spontaneous fluctuations in
prostate volume over time.
Prostate Cancer and Prostatic Diseases
Accuracy of prostate volume measurements
LM Eri et al
278
References
1 Aarnink RG, De La Rosette JJMCH, Debruyne FMJ, Wijkstra H.
Reproducibility of prostate volume measurements from transrectal ultrasonography by an automated and a manual technique. Br
J Urol 1996; 78: 219 – 223.
2 Hendrikx AJM et al. Audex medical, a new system for digital
processing and analysis of ultrasonographic images of the prostate. Scand J Urol Nephrol 1991; 137: 95 – 100.
3 Kimura A, Kurooka Y, Kitamura T, Kawabe K. Biplane planimetry as a new method for prostatic volume calculation in transrectal ultrasonography. Int J Urol 1997; 4: 152 – 156.
4 Nathan MS et al. Transrectal ultrasonography: why are estimates
of prostate volume and dimension so inaccurate? Br J Urol 1996;
77: 401 – 407.
5 Terris MK, Stamey TA. Determination of prostate volume by
transrectal ultrasound. J Urol 1991; 145: 984 – 987.
6 Bates TS, Reynard JM, Peters TJ, Gingell JC. Determination of
prostatic volume with transrectal ultrasound: a study of intraobserver and interobserver variation. J Urol 1996; 155: 1299 –
1300.
7 Myschetzky PS et al. Determination of prostate gland volume by
transrectal ultrasound: correlation with radical prostatectomy
specimens. Scand J Urol Nephrol 1991; 137(Suppl), 107 – 111.
8 Matthews GJ, Motta J, Fracchia JA. The accuracy of transrectal
ultrasound prostate volume estimation: clinical correlations.
J Clin Ultrasound 1996; 24: 501 – 505.
9 Bonilla J et al. Intra- and interobserver variability of MRI prostate
volume measurements. Prostate 1997; 31: 98 – 102.
Prostate Cancer and Prostatic Diseases
10 Andersen JT et al, and the Scandinavian BPH Study Group. Can
finasteride reverse the progress of benign prostatic hyperplasia?
A two-year placebo-controlled study. Urology 1995; 46: 631 – 637.
11 Jepsen JV, Leverson G, Bruskewitz C. Variability in urinary flow
rate and prostate volume: an investigation using the placebo arm
of a drug trial. J Urol 1998; 160: 1689 – 1694.
12 Rhodes T et al. High variability change in prostate volume by
transrectal ultrasound over time within untreated men. J Urol
1996; 155: 461A (suppl), (abstract 600).
13 Jones DR et al. Assessment of volume measurement of the
prostate using per-rectal ultrasonography. Br J Urol 1989; 64:
493 – 495.
14 Hastak SM, Gammelgaard J, Holm HH. Transrectal ultrasonic
volume determination of the prostate — a preoperative and postoperative study. J Urol 1982; 127: 1115 – 1118.
15 Elliot TL et al. Accuracy of prostate volume measurements in
vitro using three-dimensional ultrasound. Academ Radiol 1996; 3:
401 – 406.
16 Bonilla J et al. Patterns of prostate growth observed in placebo
treated patients in the PLESS trial over four years. J Urol 1998;
159: 333 (suppl), (abstract 1281).
17 Ray PS et al. Methodologic variations in medical treatment of
BPH. J Urol 1989; 149: 534A (suppl), (abstract 1478).
18 Tong S et al. Intra- and inter-observer variability and reliability of
prostate volume measurement via two-dimensional and threedimensional ultrasound imaging. Ultrason Med Biol 1998; 5: 673 –
681.