Open Journal of Urology, 2011, 1, 28-36
doi:10.4236/oju.2011.13008 Published Online August 2011 (http://www.SciRP.org/journal/oju)
Correlation of PSA Density to Prostate Cancer Based on
Pr ostate Volume by 3.0 T MRI
Rulon L. Hardman1*, Yuanyuan Liang2,3, Steve Ware1, Adam J. Jung, Qi Peng1,
Fadi El-Merhi1, Yumin Chen2, Ian M. Thompson3
1Department of Radi ology, Univ ersity of Texas Health Science
Center at San Anto ni o, San Antoni o, USA
2Department of Epi demiology and Biostat i st i c s , University of Texas Health Science
Center at San Anto ni o, San Antoni o, USA
3Department of Ur ol o gy, University of Texas Health Science Center at San Antonio, San A nt o nio, USA
Received March 1, 2011; revised May 13, 2011; accepted June 20, 2011
Abstract
Purpose: Prostate specific antigen levels can be normalized by the prostate volume to give a prostate specific
antigen density (PSAd). Magnetic resonance imaging (MRI) can more accurately determine prostate zonal
anatomy and prostate volumes compared to transrectal ultrasound, and hence may lead to more accurate
PSAd measurements. Methods: Imaging and pathology of men undergoing prostate MRI from April 2007 to
May 2009 were reviewed in this retrospective study. 73 patients were included for analysis, of which 45 had
prostate cancer and 28 did not have cancer. Total, transitional zone, and peripheral zone values were deter-
mined by ultrasound prolate ellipse, MRI prolate ellipse, and MRI segmentation methods. Results: The study
population showed an average PSA of 6.3 ng/mL, with the control mean PSA (8.8 ng/mL) being greater than
the cancer group (5.3 ng/mL). Transrectal ultrasound underestimated the prostate volume (mean 27.7 mL
versus MRI volume of 38.3 mL, p ≤ 0.001). No difference was seen between cancer and control populations
using PSAd. PSAd correctly categorized low (Gleason < 7) and high-grade cancers (Gleason ≥ 7) in patients
with malignancy. Conclusion: Transrectal ultrasound underestimates prostate volumes and hence is inaccu-
rate in calculating PSAd. MRI more accurately depicts PSAd, however PSAd is unable to differentiate be-
tween patients with cancer and benign disease such as BPH or prostatitis.
Keywords: Magnetic Resonance Imaging, Prostatic Neoplasm, Prostate-Specific Antigen
1. Introduction
Prostate specific antigen (PSA) is a commonly used tu-
mor marker in screening patients for prostate cancer risk
[1,2]. A typical cut off value of 4.0 ng/mL indicates an
abnormal result [1]. However, the test is not specific as
other pathologies including inflammation and benign
prostatic hyperplasia (BPH) can increase PSA [1]. Cur-
rent PSA recommendations lead to significant over-
treatment and associated morbidity from unnecessary
biopsies, as well as psychological stress in the patient.
For these reasons, many have advocated normalizing the
PSA by the volume of the prostate gland, yielding a PSA
density (PSAd) [3-5].
The use of PSAd for cancer diagnosis is controversial
with studies both confirming and refuting the use of
PSAd [6-10]. In an effort to further improve the specific-
ity, PSAd can be based on the transitional zone of the
prostate, the region of the prostate most affected by BPH
changes, however, these result remain controversial [6-8,
11]. A possible explanation for the lack of added speci-
ficity of PSAd may be that the volume of the prostate is
typically determined by transrectal ultrasound, which is
inaccurate in determining prostate volume [12,13]. Pros-
tate MRI can accurately depict prostate zonal anatomy.
MRI has been shown to be useful in differentiating
high-grade and low-grade cancer, but groups have not
shown their results for sensitivity for detecting cancer
[14-17]. We undertook the study to determine the accu-
racy of PSAd, calculated using MRI and transrectal ul-
trasound, in characterizing benign from malignant dis-
ease in a population with persistently elevated PSA.

R. L. HARDMAN ET AL.
29
2. Methods
Investigational Review Board (IRB) consent was ob-
tained and Health Insurance Portability and Accountabil-
ity Act (HIPAA) and confidentiality practices were fol-
lowed.
Patients were referred for prostate MRI for staging,
cancer follow-up, or being high risk for prostate cancer
(based on elevated PSA or palpable nodule). Patients
were excluded from analysis if they had any prostate
volume altering therapies, including 5-alpha inhibitor
treatment, brachytherapy or cryoablation. Patients were
also excluded if pathologic biopsy or prostatectomy re-
sults could not be obtained.
All patients underwent a prostate MRI on a 3.0 Tesla
platform (Achieva, Philips Healthcare, Best, Netherlands)
using combined cardiac phase array coil (Philips Health-
care, Best, Netherlands), and endorectal coil (BPX-30,
Medrad, Pittsburgh, PA). High resolution T2W protocol
was performed by a fast spin-echo sequence covering the
prostate and seminal vesicles. The scan parameters were:
TR/TE (3000 - 4000)/110 ms, echo train length = 20,
slice thickness = 3 mm, interslice gap = 0 mm, field of
view = 14 × 14 cm2, matrix 500 × 550, fold-over direc-
tion left to right, 3 excitations, and in-plane resolution of
0.3 × 0.3 mm2. High-resolution T2W sequences were
used for volume measurement as they depict the prostate
in the greatest detail.
Axial and sagittal images were analyzed to determine
the prostate volume by prolate ellipse and segmentation
methods (Figure 1). The anterior-posterior, cranial-
caudal, and transverse dimensions were taken of the total
prostate and transitional zone at the widest portion of the
prostate. The prostate volume was then determined using
Equation (1). The volume of an object may also be ap-
proached by taking thin segments though the object and
adding the areas of each segment, also known as seg-
mentation. Segmentation was performed by drawing a
region of interest around the total prostate and transi-
tional zone on each axial image slice. The sum of the
areas was then multiplied by the slice thickness (3 mm)
and the segmented volumes were summed to determine
the total prostate volume.
Prostate Volume = AP × CC × Trans × 0.52 (1)
Patient records were reviewed to determine pathology
from pre- and post- MRI biopsies and prostatectomy.
Prostate volumes based on digital rectal exam and trans-
rectal ultrasound was also recorded when available and
recorded. Data was only available on 52 out of the 92
men. The maximum time from ultrasound to MRI meas-
urement was 6 months. Three urologists (26 years, 13
years, and 6 years experience) at our institution per-
formed all biopsies and ultrasound volume calculations
using a standard prolate ellipse methodology. Transrectal
ultrasound measurements were performed using the
Hawk 2102 XDI Ultrasound Scanner (B-K Medical,
Herlev, Denmark). A single radiologist (4 years experi-
ence) performed all MRI calculations. The most recent
PSA prior to the prostate MRI was used for PSAd calcu-
lations. The PSA density was determined by dividing the
PSA by the ultrasound volume, prolate ellipse MRI vo-
lume, and segmentation MRI volume. Peripheral zone
volume was determined by subtracting the transitional
zone volume from the total prostate volume.
Statistical Methods
Seven different PSA densities were calculated by di-
viding the PSA by various prostate volumes (total pros-
tate volume, transitional zone volume, or peripheral zone
volume) using different methods (ultrasound, prolate
ellipse or segmentation). Areas underneath the receiver-
(a) (b) (c)
Figure 1. Prostate volume calculation. Image (a) and (b) show the measurements for the prolate ellipse method. The black
line demarcates the total prostate. The white line shows the central gland. Image (c) shows the segmentation curves placed
round the prostate margins. The volume is calculated by adding each slice area. a

R. L. HARDMAN ET AL.
30
operating characteristic (ROC) curve (AUCs) were cal-
culated for PSA and various PSAds using the Wilcoxon
statistic, and nonparametric U-statistic method was used
performing the differences between AUCs [18] with ap-
propriate Bonferroni adjustment. Comparison was con-
ducted between the patients with aggressive cancer and
the patients with no cancer or low-grade cancer using a
Mann-Whitney U test.
All statistical tests were performed at a significance
level of 0.05 (two sided) and conducted using SAS (Ver-
sion 9.2, Cary, NC: SAS Institute Inc.).
3. Results
Patient demographics are summarized in Table 1. A total
of 92 patients underwent prostate MRI from April 2007
to May 2009. 73 patients were included for analysis us-
ing our inclusion criteria. 16 patients underwent prosta-
tectomy with 57 men only having pathology based on
TRUS biopsy. 45 patients had pathologic evidence of
prostate cancer. The average PSA was 8.3 ng/mL ± 8.3
(mean ± standard deviation) (range 0.3 - 54.3). The mean
Gleason score was 6.67 ± 0.83. The average number of
prior biopsies was 2.26 ± 1.06 (range 1 - 6) and the av-
erage number of prior biopsy cores per patient was 24.7
± 12.1 (range 6 - 77). Median PSA was significantly
higher in the non-cancer group compared to the cancer
group (8.8 ng/mL vs. 5.3 ng/mL, p = 0.03). Median tran-
sitional zone volume was also higher in the non-cancer
group but was not statistically significant (p = 0.09 by
prolate ellipse method; p = 0.06 by segmentation me-
thod). There was no significant difference between the
cancer and non-cancer groups in terms of age, race or
other types of prostate volumes.
3.1. Volume Measurements
Transrectal ultrasound data was available on 52 patients.
The comparison of total prostate volume measured using
transrectal ultrasound and MRI (prolate ellipse or seg-
mentation) are summarized in Table 2 . Mean ultrasound
volumes were significantly smaller than MRI volumes,
except for the subset of non-cancer patients with ultra-
sound volumes greater than 30 mL3. The comparison of
MRI prolate ellipse and segmentation methods are sum-
marized in Table 2 as well. The prolate ellipse method
significantly under-estimates the transitional zone vol-
ume and total prostate volume (transitional zone volume:
Table 1. Patient characteristics.
Characteristic Cancer (n = 45) Non-cancer (n = 28)All Subjects (n = 73) P-value
Age (year) Mean (SD) 69.4 (10) 69.3 (8.3) 69.3 (9.3) 0.961
Race White 35 (77.8) 25 (89.3) 60 (82.2) 0.482
N (%) Black 8 (17.8) 2 (7.1) 10 (13.7)
Hispanic 2 (4.4) 1 (3.6) 3 (4.1)
PSA (ng/mL) Median (Q1, Q3) 5.3 (2.7, 8.1) 8.8 (5.2, 11.2) 6.3 (3.6, 9.8) 0.031
<2 3 (6.7) 2 (7.1) 5 (6.8) 0.12
N (%) 2 - 4 13 (28.9) 2 (7.1) 15 (20.5)
4 - 10 21 (46.7) 15 (53.6) 36 (49.3)
≥10 8 (17.8) 9 (32.1) 17 (23.3)
Prostate Volumes (mL3) Total
3
U
Vol 27.7 (22.2, 40.8) 26.4 (24.1, 59.5) 27.1 (22.4, 42.5) 0.371
Transitional VolE 16.6 (11.3, 29.6) 24.4 (15.9, 39) 20.2 (11.8, 35.2) 0.091
Peripheral VolE 16.7 (12.9, 20.2) 16.7 (13, 23.4) 16.7 (12.9, 23) 0.751
Median (Q1, Q3) Total VolE 35.3 (25.7, 49.7) 42 (31.5, 65) 37.5 (26, 57.1) 0.121
Transitional VolS 18.5 (13.5, 31.3) 24.4 (17.2, 46.4) 20.1 (14.5, 40.5) 0.061
Peripheral VolS 17 (15.2, 20.7) 18.7 (14.1, 23.6) 17.1 (14.9, 22.5) 0.51
Total VolS 38.3 (28.1, 47.4) 44.3 (32.9, 71.4) 39.8 (30.3, 60.2) 0.121
Gleason grade <7 24 (53.3) NA
N (%) ≥7 21 (46.7)

R. L. HARDMAN ET AL.
31
Table 2. Comparison of total prostate volume measured by different approaches and comparison of transitional zone volume,
peripheral zone volume and total prostate volume measured by prolate ellipse method vs. segmentation method.
All subjects
Statistics Ultrasound Prolate ellipse Segmentation P-value
ALL N 52 73 73
Median 27.05 37.5 39.8 <0.001
(Q1, Q3) (22.4, 42.5) (26, 57.1) (30.3, 60.2)
VolU < 30 N 30 30 30
Median 23.15 26.2 29.85 <0.001
(Q1, Q3) (20.5, 25.6) (23.6, 33.2) (26.7, 35.1)
VolU ≥ 30 N 22 22 22
Median 44.05 57 57.5 0.02
(Q1, Q3) (39.7, 60.4) (43.8, 81.9) (45.5, 83.9)
All subjects (N = 73)
Statistics Prolate ellipse Segmentation P-value
Transitional Median 20.2 20.1 0.002
(Q1, Q3) (11.8, 35.2) (14.5, 40.5)
Peripheral Median 16.7 17.1 0.18
(Q1, Q3) (12.9, 23) (14.9, 22.5)
Total Median 37.5 39.8 <0.001
p = 0.002 for all subjects; total prostate volume: p < 0.001
for all subjects). Prolate ellipse and segmentation meth-
ods were similar for peripheral zone volume calculations
p = 0.18 for all subjects.
3.2. PSA Density
The average PSA in this population was lower in the
cancer group than in the benign prostatic hypertrophy
group (p = 0.03). The descriptive and diagnostic statistics
of PSA and PSA density based on transitional zone
volume, total prostate volume, and peripheral zone vol-
umes using MRI prolate ellipse or segmentation methods
or ultrasound are summarized in Table 3. Only the pe-
ripheral zone PSAd was significantly different between
the BPH and cancer groups, based prolate ellipse or
segmentation calculations (p = 0.03 for both calcula-
tions). The medians of the rest of the biomarkers are not
significantly different between cancer and non-cancer
groups. All AUCs were below 0.5 indicating all the
markers were under-expressed in cancer in this sample.
PSA and PSAds are also compared between cancer
groups, high-grade cancer (Gleason ≥ 7) and low-grade/
non-cancer (Gleason score < 7). As shown in Table 4, all
markers were significantly higher in the high-grade can-
cer group compared to low-grade cancer/non-cancer
group. The median values of ultrasound PSAd and seg-
mentation transitional zone PSAd were significantly
higher in high-grade cancer group compared to non-
cancer or low-grade cancer groups. The AUCs for ultra-
sound PSAd and segmentation transitional zone PSAd
were 0.68 (95% CI: 0.52 - 0.83) and 0.67 (95% CI: 0.54 -
0.8), respectively, which were higher, although not sig-
nificantly different than the AUC for PSA of 0.57 (95%
CI: 0.43 - 0.71) (Figure 2).
4. Discussion
PSA is a commonly used tool to screen men for possible
prostate cancer. While a cutoff of 4.0 ng/mL is fre-
quently used to indicate greater cancer risk, the value
often over-diagnoses many men, leading to significant
morbidity from biopsies and surgeries [19]. PSA alone
has no effect on long-term cancer survival [20,21]. One
major hindrance to PSA testing is its lack of specificity.
Benign prostatic hypertrophy is known to increase PSA,
accounting for up to 85% of cases with a PSA between
4.1 - 10.0 ng/mL [22]. Infl mmation can also increase a

32 R. L. HARDMAN ET AL.
Table 3. Biomarker characteristics in cancer and non-cancer groups.
Non-Cancer Cancer P-value AUC (95%CI)
PSA 0.03 0.35 (0.22, 0.48)
N 28 45
Mean (SD) 9.19 (6.64) 7.75 (9.23)
Median (Q1, Q3) 8.75 (5.2, 11.2) 5.3 (2.7, 8.1)
PSAdUS 0.33 0.41 (0.25, 0.57)
N 15 37
Mean (SD) 0.23 (0.18) 0.21 (0.21)
Median (Q1, Q3) 0.22 (0.12, 0.27) 0.14 (0.1, 0.28)
PSAdET 0.17 0.4 (0.27, 0.53)
N 28 45
Mean (SD) 0.21 (0.2) 0.25 (0.4)
Median (Q1, Q3) 0.15 (0.11, 0.23) 0.12 (0.07, 0.23)
PSAdST 0.19 0.41 (0.28,0.54)
N 28 45
Mean (SD) 0.2 (0.18) 0.21 (0.27)
Median (Q1, Q3) 0.15 (0.1, 0.25) 0.11 (0.07, 0.24)
PSAdEC 0.59 0.46 (0.33, 0.59)
N 28 45
Mean (SD) 0.4 (0.47) 0.48 (0.79)
Median (Q1, Q3) 0.27 (0.18, 0.47) 0.25 (0.13, 0.55)
PSAdEP 0.03 0.35 (0.22, 0.48)
N 28 45
Mean (SD) 0.53 (0.4) 0.55 (0.83)
Median (Q1, Q3) 0.42 (0.28, 0.67) 0.25 (0.17, 0.47)
PSAdSC 0.65 0.47 (0.34, 0.6)
N 28 45
Mean (SD) 0.37 (0.39) 0.41 (0.54)
Median (Q1, Q3) 0.24 (0.14, 0.48) 0.22 (0.11, 0.5)
PSAdSP 0.03 0.34 (0.21, 0.47)
N 28 45
Mean (SD) 0.51 (0.39) 0.45 (0.54)
Median (Q1, Q3) 0.49 (0.28, 0.58) 0.25 (0.16, 0.42)
PSA levels. For these reasons, normalizing PSA to
prostate volume has been advocated. Kalish first sug-
gested in 1994 to normalize PSA to the transitional zone,
as this would normalize PSA values for BPH changes [8].
Kalish showed that PSAd based on transitional zone was
the most correlated with the presence of cancer. However,
transitional zone PSAd has been refuted by another study
PSAd based on transitional zne volume has since been o

R. L. HARDMAN ET AL.
33
Table 4. Biomarker characteristics in high-grade cancer and non-cancer or low-grade cancer groups.
Non-Cancer or low-grade cancerHigh-grade cancer(Gleason ≥ 7) P-value AUC (95% CI)
PSA 0.34 0.57 (0.43, 0.71)
N 52 21
Mean (SD) 7.74 (6.94) 9.71 (11.09)
Median (Q1, Q3) 5.8 (3.5, 9.8) 7.3 (5, 8.8)
PSAdUS 0.05 0.68 (0.52, 0.83)
N 37 15
Mean (SD) 0.2 (0.2) 0.27 (0.18)
Median (Q1, Q3) 0.14 (0.1, 0.24) 0.24 (0.13, 0.33)
PSAdET 0.18 0.6 (0.46, 0.74)
N 52 21
Mean (SD) 0.2 (0.26) 0.31 (0.48)
Median (Q1, Q3) 0.12 (0.08, 0.22) 0.18 (0.09, 0.33)
PSAdST 0.1 0.62 (0.49, 0.76)
N 52 21
Mean (SD) 0.19 (0.21) 0.25 (0.28)
Median (Q1, Q3) 0.11 (0.07, 0.22) 0.17 (0.1, 0.27)
PSAdEC 0.11 0.62 (0.48, 0.76)
N 52 21
Mean (SD) 0.39 (0.5) 0.61 (1.01)
Median (Q1, Q3) 0.24 (0.13, 0.37) 0.35 (0.19, 0.62)
PSAdEP 0.39 0.56 (0.42, 0.71)
N 52 21
Mean (SD) 0.5 (0.59) 0.66 (0.92)
Median (Q1, Q3) 0.3 (0.17, 0.57) 0.38 (0.21, 0.66)
PSAdSC 0.02 0.67 (0.54, 0.8)
N 52 21
Mean (SD) 0.35 (0.44) 0.51 (0.59)
Median (Q1, Q3) 0.19 (0.12, 0.39) 0.33 (0.2, 0.61)
PSAdSP 0.49 0.55 (0.41, 0.69)
N 52 21
Mean (SD) 0.45 (0.45) 0.54 (0.57)
Median (Q1, Q3) 0.32 (0.2, 0.53) 0.34 (0.21, 0.56)
refuted in another study [9]. There is great variability in
determining the transitional zone volumes by transrectal
ultrasound [12,13]. MRI provides an accurate depiction
of prostate zonal anatomy, and has been used to deter-
mine PSAd in a small number of studies [14-17]. The
technique used for the calculation is varied and each
sample has been small. While transitional zone PSAd has
been shown to be the most accurate parameter at pre-

34 R. L. HARDMAN ET AL.
(a) (b)
Figure 2. ROC curves: (a) Detection of prostate cancer. (b) Detection of aggressive prostate cancer. In this sample, all mark-
ers were lower in cancer group, therefore, sensitivity was defined as the proportion of cancer cases with biomarker value less
than a cut point and specificity as the proportion of controls with biomarker value exceeding a cut point.
dicting Gleason score and presence of cancer, the sensi-
tivity and specificity of such reports is less encouraging.
We attempted to analyze the accuracy of transrectal
ultrasound in determining prostate volumes. Similar to
prior reports [12], we demonstrated that transrectal ul-
trasound significantly underestimates prostate volume.
The gross underestimation of prostate volume by trans-
rectal ultrasound limits the possible use of transrectal
ultrasound based PSAd.
We also attempted to determine if the added volume
accuracy of MRI could improve PSAd accuracy. This
population was unique in that the average PSA for the
cancer patients was actually lower than that of the “con-
trol” group. The average PSA in the group (6.3 ng/ml) is
higher than expected for the general public. This would
be the ideal population where PSAd could be useful, as
PSA alone is not specific in accessing cancer risk. Un-
fortunately, the density normalization was not helpful in
determining cancer risk. The only statistically significant
PSAd measurements were PSA and peripheral zone
PSAd; however, these were inverted from the hypothe-
sized values with the cancer group demonstrating lower
averages. The PSAd calculation was useful in determin-
ing high-grade malignancy versus low-grade cancer. The
majority of prior reports on the usefulness of MRI in
PSAd have focused on the showing that MRI PSAD can
determine high versus low grade malignancy, but fail to
demonstrate their results for cancer sensitivity.
The lack of added benefit for PSAd based on MRI vo-
lumes may be due to selection bias. The patient popula-
tion generally referred to MRI in this retrospective study
tended to have high PSA values and no cancer to low
grade Gleason scores after multiple biopsies. The referral
group may reflect a subset of patients that are at the high
end of the normal curve of PSA values. The use of MRI
based PSAd may be more useful in a screening population;
however, this is yet to be determined.
The study is limited by the lack of prostatectomy on
most patients. Even in men with prostatectomy, only one
individual had prostate volume recorded. Correlation to
prostate volume by prostatectomy sample should be per-
formed on future studies. Another limitation is the un-
known affect of the presence of an endorectal coil in
volume determinations. The prostate is known to deform
with the presence of the endorectal coil. Additional trials
should be performed in a screening population to deter-
mine of PSAd may be helpful in this more standard pa-
tient population. PSAd may be helpful in prediction al-
gorithm to be combined with other parameters to assess
prostate cancer risk, as suggested by Kubota [16]. This
study is useful in that is reports results from a moderate
sized sample, tests the added resolution of 3.0 T MRI,
and shows cancer sensitivity in addition to high grade
cancer differentiation.
The cost of MRI calculated volume is much greater
than transrectal ultrasound calculated volume. While the
volume measurement is not the sole reason for perform-
ing prostate MRI or transrectal ultrasound, the technical
fee for a prostate MRI is $1900, while the transrectal
ultrasound technical fee is $180 at our institution.

R. L. HARDMAN ET AL.
35
In summary, transitional PSAd by MRI segmentation
method and transrectal ultrasound PSAd positively cor-
related with aggressive cancer. However, PSAd was un-
able to demonstrate a statistically significant difference
from PSA for detection of cancer risk in our patient pop-
ulation.
Conflict of interest: Dr. Ian Thompson is a paid con-
sultant for Mission Pharmaceutical, which provides
compensation to the University of Texas Health Science
Center at San Antonio.
5. References
[1] J. Hernandez and I. Thompson, “Prostate-Specific Anti-
gen: A Review of the Validation of the Most Commonly
Used Cancer Biomarker,” A Cancer Journal for Clini-
cians, Vol. 101, No. 5, 2004, pp. 894-904.
[2] S. Wilson and E. Crawford, “Screening for Prostate Can-
cer,” Clinical Genitourinary Cancer, Vol. 3, No. 1, 2004,
pp. 21-25. doi:10.3816/CGC.2004.n.008
[3] M. Bazinet, A. Meshref, C. Trudel, S. Aronson, F. PÈlo-
quin, M. Nachabe, L. BÈgin and M. Elhilali, “Prospective
Evaluation of Prostate-Specificantigen Density and Sys-
tematic Biopsies for Early Detection of Prosttic Carci-
noma,” Urology, Vol. 43, No. 1, 1994, pp. 44-51.
doi:10.1016/S0090-4295(94)80260-2
[4] M. Benson, I. Whang, A. Pantuck, K. Ring, S. Kaplan, C.
Olsson and W. Cooner, “Prostate Specific Antigen Den-
sity: A Means of Distinguishing Benign Prostatic Hyper-
trophy and Prostate Cancer,” The Journal of Urology,
Vol. 147, No. 3, 1992, pp. 815-816.
[5] E. Seaman, M. Whang, C. Olsson, A. Katz, W. Cooner
and M Benson, “Psa Density (Psad). Role in Patient
Evaluation and Management,” The Urologic Clinics of
North America, Vol. 20, No. 4, 1993, pp. 653-663.
[6] B. Djavan, M. Remzi, A. Zlotta, V. Ravery, P. Hammerer,
A. Reissigl, P. Dobronski, A. Kaisary and M. Marberger,
“Complexed Prostate-Specific Antigen, Complexed Pros-
tate-Specific Antigen Density of Total and Transition
Zone, Complexed/Total Prostate-Specific Antigen Ratio,
Free-to-Total Prostate-Specific Antigen Ratio, Density of
Total and Transition Zone Prostate-Specific Antigen:
Results of the Prospective Multicenter European Trial,”
Urology, Vol. 60, No. 4, 2002, pp. 4-9.
doi:10.1016/S0090-4295(02)01896-4
[7] B. Djavan, A. Zlotta, C. Kratzik, M. Remzi, C. Seitz, C.
Schulman and M. Marberger, “Psa, Psa Density, Psa
Density of Transition Zone, Free/Total Psa Ratio, and Psa
Velocity for Early Detection of Prostate Cancer in Men
with Serum Psa 2.5 to 4.0 ng/ml,” Urology, Vol. 54, No.
3, 1999, pp. 517-522.
[8] J. Kalish, W. Cooner and S. Graham Jr., “Serum Psa Ad-
justed for Volume of Transition Zone (Psat) Is More Ac-
curate than Psa Adjusted for Total Gland Volume (Psad)
in Detecting Adenocarcinoma of the Prostate,” Urology,
Vol. 43, No. 5, 1994, pp. 601-606.
doi:10.1016/0090-4295(94)90170-8
[9] D. Lin, M. Gold, S. Ransom, W. Ellis and M. Brawer,
“Transition Zone Prostate Specific Antigen Density: Lack
of Use in Prediction of Prostatic Carcinoma,” The Jour-
nal of Urology, Vol. 160, No. 1, 1998, pp. 77-82.
doi:10.1016/S0022-5347(01)63036-2
[10] M. Remzi, B. Djavan, R. Wammack, M. Momeni, C.
Seitz, B. Erne, M. Dobrovits, S. Alavi and M. Marberger,
“Can Total and Transition Zone Volume of the Prostate
Determine whether to Perform a Repeat Biopsy?” Urol-
ogy, Vol. 61, No. 1, 2003, pp. 161-166.
doi:10.1016/S0090-4295(02)02099-X
[11] H. Maeda, Y. Arai, S. Ishitoya, K. Okubo, Y. Aoki and T.
Okada, “Prostate Specific Antigen Adjusted for the Tran-
sition Zone Volume as an Indicator of Prostate Cancer,”
The Journal of Urology, Vol. 158, No. 6, 1997, pp. 2193-
2196. doi:10.1016/S0022-5347(01)68193-X
[12] M. Nathan, K. Seenivasagam, Q. Mei, J. Wickham and R.
Miller, “Transrectal Ultrasonography: Why Are Esti-
mates of Prostate Volume and Dimension So Inaccurate,”
British Journal of Urology, Vol. 77, No. 3, 1996, pp. 401-
407.
[13] M. Terris and T. Stamey, “Determination of Prostate
Volume by Transrectal Ultrasound,” The Journal of
urology, Vol. 145, No. 5, 1991, p. 984.
[14] A. Graser, A. Heuck, B. Sommer, J. Massmann, M. Reis-
er and U. Mueller-Lisse, “Mri-Based Psa Density and Psa
Density of the Transitional Zone Compared with Psa
Alone: Correlation with Prostate Cancer Gleason Score,”
Journal of Computer Assisted Tomography, Vol. 30, No.
6, 2006, pp. 891-895.
doi:10.1097/01.rct.0000221947.68016.d1
[15] T. Hoshii, T. Nishiyama, S. Toyabe, K. Akazawa, S.
Komatsu, M. Kaneko, N. Hara and K. Takahashi, “Eval-
uation of Magnetic Resonance Imaging-Based Pros-
tate-Specific Antigen Density of the Prostate in the Di-
agnosis of Prostate Cancer,” International Journal of
Urology, Vol. 14, No. 4, 2007, pp. 305-310.
doi:10.1111/j.1442-2042.2007.01686.x
[16] Y. Kubota, S. Kamei, M. Nakano, H. Ehara, T. Deguchi
and O. Tanaka, “The Potential Role of Prebiopsy Mag-
netic Resonance Imaging Combined with Prostate-Spe-
cific Antigen Density in the Detection of Prostate Can-
cer,” International Journal of Urology, Vol. 15, No. 4,
2008, pp. 322-326.
doi:10.1111/j.1442-2042.2008.01991.x
[17] U. Mueller-Lisse, U. Mueller-Lisse, S. Haller, P. Schn-
eede, J. Scheidler, N. Schmeller, A. Hofstetter and M.
Reiser, “Likelihood of Prostate Cancer Based on Pros-
tate-Specific Antigen Density by Mri: Retrospective
Analysis,” Journal of Computer Assisted Tomography,
Vol. 26, No. 3, 2002, pp. 432-437.
doi:10.1097/00004728-200205000-00020
[18] E. DeLong, D. DeLong and D. Clarke-Pearson, “Com-
paring the Areas under Two or More Correlated Receiver
Operating Characteristic Curves: A Nonparametric Ap-
proach,” Biometrics, Vol. 44, No. 3, 1988, pp. 837-845.
doi:10.2307/2531595
[19] I. Thompson, D. Pauler, P. Goodman, C. Tangen, M.

R. L. HARDMAN ET AL.
36
Lucia, H. Parnes, L. Minasian, L. Ford, S. Lippman and E.
Crawford, “Prevalence of Prostate Cancer among Men
with a Prostate-Specific Antigen Level <= 4.0 ng per
Milliliter,” New England Journal of Medicine, Vol. 350,
No. 22, 2004, pp. 2239-2321.
doi:10.1056/NEJMoa031918
[20] G. Andriole, E. Crawford, R. Grubb III, S. Buys, D. Chia,
T. Church, M. Fouad, E. Gelmann, P. Kvale and D. Red-
ing, “Mortality Results from a Randomized Prostate-Can-
cer Screening Trial,” The New England Journal of Medi-
cine, Vol. 360, No. 13, 2009, pp. 1310-1319.
doi:10.1056/NEJMoa0810696
[21] F. Schroder, J. Hugosson, M. Roobol, T. Tammela, S.
Ciatto, V. Nelen, M. Kwiatkowski, M. Lujan, H. Lilja
and M. Zappa, “Screening and Prostate-Cancer Mortality
in a Randomized European Study,” New England Journal
of Medicine, Vol. 360, No. 13, 2009, pp. 1320-1328.
doi:10.1056/NEJMoa0810084
[22] W. Catalona, J. Richie, F. Ahmann, M. Hudson, P. Scar-
dino, R. Flanigan, J. DeKernion, T. Ratliff, L. Kavoussi
and B. Dalkin, “Comparison of Digital Rectal Examina-
tion and Serum Prostate Specific Antigen in the Early
Detection of Prostate Cancer: Results of a Multicenter
Clinical Trial of 6,630 Men,” The Journal of Urology,
Vol. 151, No. 5, 1994, pp. 1283-1290.