Genitourinary H. Ballentine Jonathon
Carter,
MD
I. Epstein,
MD
#{149}Rachel
Patrick
#{149}
Nonpalpable with
MR
for
terms:
a sensitivity
Prostate,
MR
plasms,
I
MD
#{149}
M. Tempany,
Elias
MD
A. Zerhouni,
Cancer:
From
1991;
the
(H.B.C.,
of 58%.
screening. 844. 1214
P.C.W.),
C.M.T., A.Y., Johns Hopkins more,
MD
sion
requested
The
844.1214 #{149} Prostate,
neo-
is the
cancer
second
deaths
of Urology, Radiology
Received July
23;
403
Mar-
States in the
(1). There detection
cause
in the
of
United
is increasing of prostate
interest cancers at
an organ-confined stage the (a) high prevalence
because of prostate
of
cancer
50 years
(2);
in men
over
age
(b) poor disease-free interval after therapy in patients with advanced disease compared to those with ongan-confined disease (3); and (c) impnovements in surgical therapy that
have
decreased
rectal
morbidity
ultrasound
imaging and aids histologic lesions digital
ever,
(4). Trans-
(US)
unsuspected in obtaining
is capable
of
prostate specimens
examination are not palpable rectal examination
in a previous
Figure
cancer for
when these by means of (5). How-
evaluation
of pa-
tients undergoing radical metnopubic prostatectomy (RRP) for clinically organ-confined disease, we have shown that tnansrectal US has a high false-positive mate in the detection of nonpalpable, posteriorly located tumoms (6). In the past decade theme
have
been
vances
rapid
technologic
in magnetic
ad-
resonance
cluding study, imaging
the
prostate evaluated to depict
we
the
(Fig i). In the ability nonpalpable
iminthis of MR pros-
June revision
1, 1990; received
reviSep-
1991
prostate
METHODS
one
lobe
and
that
the
palpable
into
another
tate mm the
specimens intervals long axis
bedded and histologically
cancer
pathologic region
did
of the
confined not prostate.
is present
in
tense
relative
were
measured
mensions.
peripheral
in In
step the
the
measured
cephalocaudal plane by
zone
millimeters
extend Pros-
in
transverse in
the
is hypoin-
zone.
three
plane,
anterior
di-
lesions
to posteri-
(up and down) and the dimension (left to right). In adlesions were measured in a dimension reconstructing
sections. location
in the
For purposes of the lesions,
the sagittal transverse
of identifying the transverse
into four quadrants and posterior), was drawn.
MR imaging was performed RRP with a i.5-T superconducting
prior
net
Systems,
imager
planes. to
[nepeti-
and
to peripheral
(Signa;
Milwaukee). were obtained
confirmation
cancer
(2,500/80
sections were divided (left and right, anterior and a map of the lesion
The records of 53 patients who underwent pelvic MR imaging prior to RRP were reviewed. All patients had a clinipalpable
T2-weighted
or dimension transverse dition, the
tate cancer in patients undergoing RRP for clinically localized prostate cancer. AND
1.
tion time msec/echo time msec]) image of prostate in axial plane obtained at 1.5 1. Note relatively hypointense central zone (single black arrowhead) and more intense peripheral zone (double black arrowheads). Rectum is shown by double white arrowheads. Prostate cancer confirmed at pathologic examination (single white arrowhead)
were
(MR)
imaging that have improved age quality of pelvic structures,
cally
(R.F.B.,
tember 6; accepted September 7. Supported by grants Ca 15416 and Ca 45202 from the National Cancer Institute. Address reprint requests to H.B.C. RSNA,
MD
prostate
leading
of men
PATIENTS
E.A.Z.), and Pathology (J.I.E.), The Hospital, 600 N Wolfe St. Balti-
21205.
Yang,
Detection
of the
DENOCARCINOMA
178:523-525
Departments
#{149} Andrew
MD
844.32
Radiology
burg
Cancer studies,
#{149} Clare
Imaging’
sensitivity of MR imaging in the detection of nonpalpable, posteriorly located tumors was greater than for those located anteriorly (85% vs 15%). MR imaging was false-positive for nonpalpable tumor in 17 of 30 patients for a specificity of 43%. On the basis of these data, MR imaging has greater sensitivity in the depiction of posteriorly located cancer and is limited by a high falsepositive rate in the depiction of nonpalpable tumors. Index
C. Walsh,
MD
Prostate
The pathologic specimens and magnetic resonance (MR) images of 53 patients with clinically palpable prostate cancer confined to one lobe were studied to evaluate the ability of MR imaging to depict clinically nonpalpable prostate cancer. All patients had undergone imaging with a 1.5-T imager with Ti- and T2weighted sequences in both axial and sagittal planes before undergoing radical retropubic prostatectomy. At pathologic examination, only the palpable tumor was present in 30 of the 53 patients (57%), and 33 unsuspected tumors were present in an area distinct from the palpable tumor in 23 of the patients (43%). MR imaging successfully depicted 51 palpable tumors for a sensitivity of 96% and 19 of the 33 unsuspected tumors
F. Brem,
Radiology
For
GE
Medical
Ti- and T2-weighted in both axial and Ti-weighted
to mag-
images sagittal
imaging,
the
se-
quence parameters were 600/20, and for T2-weighted imaging, the sequence parameters were 2,500/20-80. A retrospective analysis of the MR images was per-
were step sectioned at 3in a plane perpendicular to of the prostate and were emexamined identified
histologically. prostate
The cancers
Abbreviation:
RRP
radical
retropubic
pros-
tatectomy.
523
IT LT
ST
b. Figure
2.
analysis.
Location
of 53 palpable anterior, POST
ANT
of patients
with
palpable
(a) and posterior,
tumors
(in
33 nonpabpabbe RT right,
b, those
patients
(b) tumors LT
also
in transverse
left. Striped had nonpalpable
plane
determined
ovals = number tumors).
by means
of nonpalpable
Figure
zone
on T2-weighted
images
were
in
con-
sidered suggestive of cancer. sion was made by a committee,
This deciwithout
knowledge
location
of the pathologic
tumor
in the right
of
posterior
re-
regions
of the
the palpable or
prostate
tumor
or anterior
distinct
from
(contralateral
regions).
In
posteni-
31 of the
53 pa-
tients (58%), the palpable tumor demonstrated microscopic capsular penetration and/or seminal vesicle invasion at pathologic examination. The largest dimension of each tumor
was used
to calculate
the average
dimension of palpable and tumors. The average largest for palpable and nonpalpable
compared
by means
ney U (Wilcoxon) nificance.
largest
nonpalpable dimension tumors
was
of the Mann-Whittest
for statistical
sig-
RESULTS Clinically
sualized peripheral for
palpable
cancers
were
vi-
as a hypointense area in the zone in 51 of the patients
a sensitivity
of 96%.
The
Table
sum-
manizes the MR findings with respect to the presence or absence of unsuspected (nonpalpable) tumors. As 524
Radiology
#{149}
sensitivity
imaging detumors for
of 58% (sensitiv-
[n i91/true-positive [n = 19] + false-negative [n 14]). The sensitivity of MR imaging in depicting nonpalpable posteriorly located tumors was greater than that for nonpalpable anteriorly located tumors (85% vs 15%). There was no statistically significant difference in Gleason scores ity
gion of the prostate and 19 (36%) had a palpable tumor in the left posterior region (Fig 2a). There were 33 unsuspected tumors in 23 of the patients (43%). These tumors were located in regions distinct from the palpable tumor (Fig 2b). The remaining 30 patients (57%) had no cancer in
in the Table, MR 19 of 33 nonpalpable
an overall
the prostate cancer. The location of all besions suspected of being carcinoma were noted in both transverse and longitudinal planes and recorded as right or left and anterior or posterior. Thirty-four of the 53 patients (64%) had
a palpable
shown picted
true-positive
between mors
palpable
and
or between
detected
and
those
ing
not
shown).
MR imaging
was
(data
nonpalpable
nonpalpable
false-positive
in 17
of 30 patients (four false-positive findings in anterior region, 13 false-positive findings in posterior region) and correctly excluded cancer in 13 of 30 patients for an overall specificity of 43% (specificity = true-negative [n 13]/tnue-negative [n 13] + false-positive [n = 17]). In three of the 17 false-
positive
cases,
a lesion
was detected
pathologically in the false-positive area (dysplasia, benign prostatic hypentrophy, and prostatitis). The average largest pathologic dimension and standard error of the 53
palpable compared
tumors with
was 21.2 mm ± 1.4 an average largest di-
Average
size
of
the
on
prostate
gan-confined
stage,
pation
at
on
the
largest
indicator
of
survival To
modality
to
is being cancer
an
or-
surgical
can
extir-
increase
with
dis-
low associated the ability of a
evaluate
depict
placed at
since
stage
morbidity.
cancer,
unsuspected
one must have accurate documentation of the presence or absence of the cancer, since it is impossible to determine sensitivity
and
information. for clinically high in
of
opposite
lobe
are
an ideal
the
ability
pict
group
step
of
of
ing,
the
weighted proved prostatic on rately
the
These
of
patients
to evaluate
of
a nonpalpa-
reports were
to de-
since pathoentire prostate documents
absence
initial
this
RRP have a tumors
techniques
of the accurately
prostate use
(7).
tumors
or
the
nonpalpable
imaging
examination sections
Although ing
without
undergoing tumors
in which
nonpalpable
logic in
specificity
Patients palpable
prevalence the
intensity
dependent
this
ease-free
MR imaging
not
as an
emphasis
detecting
presence
was
average
used
DISCUSSION Increasing
tumor.
sensitivity
of
size.
ble
imaging
dimension
when
was
the
MR
largest
tumor
dimension
mension of 14.6 mm ± 1.5 for the 33 nonpalpable tumors. There was a statistically significant difference between the average largest pathologic dimension of palpable and nonpalpable tumoms (P < .01). In Figure 3, the palpable and nonpalpable tumors are subdivided on the basis of largest pathologic dimension and plotted as a function of sensitivity.
number
=
7:-:3c
.-2:
tu-
at MR imag-
ovals
palpable and nonpalpable tumors as a function of MR imaging sensitivity. n number of tumors in each size category, checked bars palpable tumors, striped bars nonpalpable tumors.
tumors
missed
3.
step-section
checked
.
i
formed; presence of hypointense areas the relatively hyperintense peripheral
of pathologic
tumors,
of MR not
imag-
encourag-
high-resolution,
T2-
images has resulted in imvisualization of internal characteristics (8). Depending technique
used,
peripheral delineated
the zone
from
high-signalcan
the
be
accu-
lower-sig-
February
1991
nal-intensity
central
and
zones. Early prostate a low-signal-intensity atively
transition
cancer appears as defect in the mel-
high-signal-intensity
zone (9-11). tate cancers (12),
one
cancers
peripheral
Since the majority of prosarise in the peripheral zone
would
anticipate
could
that
be accurately
detected
with MR imaging. Camnol et al (9) evaluated with
palpable
prostate
early
i3 patients
tumors
(Bi
and
B2 nodules) who underwent in vivo MR imaging of the prostate prior to RRP. Step sectioning of the pnostatectomy
specimens
RRP.
was
performed
MR imaging
tumors;
area
depicted
all were
seen
failed ability
to depict of MR
67% of the
as a hypointense
in the peripheral
ing The
after
zone.
MR imag-
cancer did not appear to depend Gleason grade or tumor volume.
on the In a
similar
evalu-
ated
study,
Gevenois
1 1 patients
with
et al (10)
use of MR imag-
ing prior to RRP for palpable prostate cancer. MR imaging depicted 82% of the tumors as a hypointense area in the peripheral zone on T2-weighted images. Schiebler et al (ii) used MR imaging to evaluate 24 radical prostatectomy specimens ex vivo prior to pathologic examination; however, the pnostates were not serially step sectioned. MR imaging depicted all macroscopic cancers (n = 14) as an area of low signal intensity. There was no macroscopic evidence of prostate cancer in 10 patients, although biopsy specimens obtained prior to RRP revealed cancer. MR imaging did not depict cancer in these patients. To our knowledge, our study is the first to evaluate the ability of MR imaging to depict nonpalpable prostate can-
cen in carefully
step-sectioned
radical
prostatectomy specimens. Since presence or absence of unsuspected cancers was accurately determined,
allowed sitivity
determination and
the this
of the true
specificity
of MR
sen-
imaging
in detecting nonpalpable prostate cancen. Our results in detecting 96% of palpable prostate cancers by means of MR imaging are better than those previously
reported
and
of different
may
be
imaging
finding is an accurate ability of MR imaging as a hypointense area,
MR imaging
may
a result
are
be similar
ity of MR
pected this
similar.
imaging
cancer group
of
if the size nonpalpable the
in depicting
patients
was
178
Number
#{149}
2
and
sensitiv-
unsus-
posteriorly
Conversely, the sensitivity ing in depicting anteriorly suspected cancers was only
Volume
in detect-
Indeed,
located
use
If this
assessment of the to depict cancer the sensitivity of
ing nonpalpable tumors location of palpable and tumors
of
techniques.
high
in (85%).
of MR imaglocated un15%. This
been
used
in this
as an indicator
of tumor
a significant
difference
study,
Gleason
scores
between
nonpalpable
tumors
demonstrated. The study
groups
size in
palpable may
and
have
been
on which
our
MR
imaging and transrectal US data are based were known to have palpable prostate cancer, and, thus, the high
prevalence this
of unsuspected
high-risk
enced
group
cancer
may
the results.
have
However,
in
influ-
on the ba-
cancer with use of MR imaging is limited by a low signal-to-noise ratio on T2weighted images. Thirteen of the 17
sis of these data and our previous experience with transrectal US, MR imaging has a higher sensitivity for detection of unsuspected prostate cancer than transrectal US when only posteriorly located tumors are considered. The specificities of tnansrectal US and MR imaging are roughly comparable. Further advances
false-positive
in
fact that
33% of the cancers. imaging in depicting
ble tumors in this study, given the significant difference in largest pathologic dimension between palpable and nonpalpable tumors. If tumor volume had
discrepancy may be due to the higher signal intensity of the posterior region compared with the anterior region since a low-signal-intensity tumor would be seen best against a background of high signal intensity. In addition, it may be difficult to detect antemionly located tumors because of pantial volume averaging between the peripheral and central zones, which masks the cancer in the anterior region of the peripheral zone. MR imaging demonstrated a high false-positive rate and low specificity (48%) in the detection of nonpalpable cancer, suggesting that other normal on pathologic processes can result in hypointensity in the peripheral zone. In addition, the
most
not associated implies that
false-positive with depiction
cases
were
a pathologic lesion of unsuspected
lesions
(76%)
ed in the posterior
were
region;
locat-
thus,
the
We have
previously
in
a similar
evaluated
the
US to depict in the posterior
group
of
59
palpable
prostate
dergoing
RRP (6). Transnectal
1. 2.
with
cancer
un-
US dem-
onstrated a sensitivity of 52% in depicting unsuspected cancers located posteriorly in 25 patients and a specificity of
68% in correctly
excluding
cancer
ities
to
depict
prostate
deserves
further
cancers.
measured
was
5.
6.
the cancer seen at MR plane. This measurement
ed to reflect
tumor
as
with
there scores
may be a difference between palpable
Gleason
1987; 137:613-619.
Walsh PC. prostatectomy
therapy
for prostatic
Techniques of radical with preservation an anatomic
approach.
phia: Hodge sound
JI. Walsh
retropubic of sexual In: Skinner
Diagnosis cancer.
Saunders, 1988; 753-778. KK, McNeal JE, Stamey guided transrectal core
palpably abnormal 142:66-70. Carter HB, Hamper
cancer.
TA. Ultrabiopsies of the
prostate. UM, PC.
and manPhiladel-
J Urol
Sheth
1989;
S. Sanders
Evaluation
RC,
of transrec-
8.
9.
11.
12.
FS,
nance imaging of the abdomen and pelvis. JAMA 1989; 261:420-433. Carrol CL, Sornrner FG, McNeal JE, Starney TA. The abnormal prostate: MR imaging at 1.5 1 with histopathologic correlation. Radiology 1987; 163:521-525. Gevenois PA, Stallenberg B, Van Sinoy ML, Salmon I, Van Regemorter C, Struyven J. Place diagnostique de l’irm dans le cancer lo-
calise
However,
and,
definitive
Urol
Stamey TA. Stage A versus stage B adenocarcinoma of the prostate: morphological cornpanison and biological significance. J Urol 1988; 139:61-65. Council on Scientific Affairs. Magnetic reso-
imaging in one is not intend-
scone,
of
im-
7.
diameter
the largest diameter of the tumor may better reflect “true” MR image size as opposed to tumor volume. Tumor volume has been shown to correlate dinectly
use
may
tal ultrasound in early detection of prostate cancer. J Urol 1989; 142:1008-1010. McNeal JE, Price HM, Redwine EA, Freiha
a
of the size of
volume.
after
Epstein
examination
as an indicator
the
Silverberg E, Lubera JA. Cancer statistics. CA 1988; 38:14-15. Feldman AR, Kessler L, Myers MH, Naughton MD. The prevalence of cancer: estimates based on the Connecticut Tumor Registry. N EngI J Med 1986; 315:1394-1397. Robey EL, Schellhammer PF. Local failure
function:
10.
the largest
at pathologic
used
4.
This
investigation,
as
imaging,
DC. Lieskovsky C, eds. agement of genitourinary
careful correlation between pathologic and imaging studies may help detect the critical factors that play a role in the ability to image early prostate can-
cer. In this study,
3.
in 34
patients without nonpalpable tumors. The ability of tmansnectal US to depict nonpalpable cancer was independent of the largest dimension of the tumor measured at pathologic examination. Camrol et al (9) also noted, as we did, the lack of correlation between cancer detection and lesion size, which suggests that factors other than size play role in the ability of imaging modal-
area
such
in MR
References
unzone
patients
clinically
imaging,
coils
prove the detection of early prostate cancer and allow treatment of this disease at an earlier stage. U
higher MR imaging sensitivity in detecting unsuspected cancer in the postenor zone is associated with a higher false-positive rate. ability of transrectal suspected cancers
prostatic
surface
de Ia prostate.
32:163-167. Schiebler ML,
Ann
Tomaszewski
Prostatic
carcinoma
perplasia:
correlation
and
Radiol
(Paris)
JE, Bezzi
benign
1989; M, et al.
prostatic
hy-
of high-resolution
MR
and histopathologic findings. Radiology 1989; 172:131-137. McNeal JE, Redwine EA, Freiha FS, Stamey TA. Zonal distribution of prostatic adenocarcinoma:
and
correlation
direction
with
of spread.
histologic
pattern
Am J Surg
Pathol
1988; 12:897-906.
thus,
in Gleason and nonpalpa-
Radiology
525
#{149}