1187
Transrectal Sonography Benign and Malignant Prostatic Lesions
L Deland
D.
2
Leo F. Drolshagen1 Arthur
C. FleiScher1 Hal T. Liddell3 w. Scott McDougal3 Edward M. Karl4 A. Everette James, Jr.1
.
of
..
,.
Using linear-array transrectal prostate sonography, malignant lesions in 43 patients and benign lesions in 74 patients were evaluated. Prostate sonography was sensitive to textural changes produced by both benign and malignant diseases. However, there was considerable overlap in the sonographic appearances of benign and malignant lesions, and there were no sonographic features that reliably predicted malignancy. Hyperechoic areas were present in 58% of the carcinomas, while 19% were purely hypoechoic. Lesions containing hypoechoic foci that were posterior or posterolateral had a high incidence of cancer. The resufts confirm the need to biopsy all suspicious palpable lesions of the prostate.
Transrectal prostatic sonography has been shown to be sensitive to textural changes produced by both benign and malignant disease, but no sonographic feature or features within the gland has been found to be accurate in predicting cancer [1 -7]. Most reports characterizing prostate lesions by transrectal sonography are based on transverse images of the prostate obtained with a radial scanner [1 , 5-7]. As yet, there have been only a few reports of sonographic features using a linear array probe [2-4]. In addition, as previously noted by Rifkin et al. [2], most studies have assumed the sonographically abnormal area to be the histologically abnormal area, without direct evidence that the biopsied area actually corresponded to the abnormal region identified by sonography. Rifkin et al., using a real-time linear array probe, obtained direct sonographichistologic correlation by performing prostate biopsies using sonographic guidance or by correlating images with results of prostatic resection [2]. In their study they noted a moderate degree of overlap between the sonographic features of benign and malignant disease. However, the authors reported certain features to be highly suggestive of benign disease, including (1) purely hypoechoic areas; (2) hyperechoic foci that are more echoic than the capsule and greater than 4 mm thick; and (3) foci with shadowing. Twenty cancers were hyperechoic; the other six had mixed October 22, 1985; accepted vision January 23, 1986. Received
1
Department
ences, Vanderbllt villa, TN 37232. FleiSCher. 2
Present
after re-
of Radiology and Radiological Sdlkiiversity Medical Center, NashAddress
address:
reprint
requests
Materials Department
of Surgery, Center, Nashville, TN 4 Department of Pathology, Medical Center, Nashville, TN Department
Medical
AJR
146:1187-1191,
The purpose of this report who had sonographic-pathlogic
June
vanderbilt
1986
0361 -803X/86/1466--1 187 © American Roentgen Ray Society
and illustrate of prostatic
the findings lesions.
in 1 1 7 patients
and
Methods
of Radiology,
Vanderbilt 37232. 37232.
is to describe correlation
to A. C.
January
Between
Navy Hospital, Corpus Christi, TX 78419. 3
echogenicity.
University University
patients
in whom
guided transperineal the 1 17 patients,
1 984 and
histologic
August
correlation
1 985, was
1 1 7 prostate
subsequently
sonograms obtained,
were
either
performed
on
by sonographically
biopsy (1 01 patients) or transurethral prostatectomy (1 6 patients). Of 43 had a histologic diagnosis of adenocarcinoma. The remaining 74 had
benign disease,
predominantly benign prostatic hyperplasia (BPH). Twelve of the cancer patients subsequently underwent radical prostatectomy. Six additional patients had pelvic lymphadenectomy without prostatectomy, five of whom had positive lymph nodes; one other patient
was
discovered
to have
periprostatic
invasion
at operation.
Overall,
three
patients
1188
BURKS
were found to be stage A; 1 8 patients, stage B; nine patients, stage C; and 1 3 patients, stage 0. Sonography was performed using a Toshiba 5055 5-MHz, longitudinally
oriented,
linear-array
transrectal
probe
designed
Toshiba SAL32A linear-array unit (Tustin, nique was similar to that previously described rectal
examination
was
performed
to identify
the
location
of any
palpable lesions, the probe was inserted and the entire prostate gland imaged by first identifying the anatomic midline, as evidenced by the presence of the prostatic urethra, then rotating the probe clockwise and counterclockwise to image the lateral lobes. Representative freeze-frame images from midline and at 5#{176} to 10#{176} increments laterally were recorded on Polaroid 667 film (Cambridge, MA). We preferred, but did not always
bladder.
require,
One hundred
that
the patient
and one patients
transperineal
biopsy
of the
Trucut biopsy
needle
(Travenol,
be scanned
with
had sonographically
prostate
with
performed
Morrow,
guided
a 16-gauge
GA).
genicity
was
predominant
Isoechoic hypoechoic
diffuse different disease,
or cysts and (6) presence estimated
relative
echogenicity
lesions
could
rim or “halo,”
enlargement,
to normal,
of the peripheral
be identified surrounding
of prostatic which
was
generally
the
gland.
either by a surrounding
or focal asymmetry.
thin
June 1986
Features in the Benign and Malignant
Groups M (N
characteristic
=
125)
(N
No. of sites (%)
=
43)
No. of sites (%)
Echogenicity Hyperechoic Isoechoic
38 (30) [55]
5 (1 2) [58J
40 (32)[56J
2 (5)[51]
Hypoechoic
11 (8) [24J 36 (28) 7 (6)
8 (19) [58] 28 (65) 1 (2) 1 1 (25) 8(19) 8 (19)
Mixed Hyper/Hypo/lso
17 (14)
Hyper/Iso
Hyper/Hypo Hype/Iso
6 (5) 6 (5)
Margin Well defined Poorly defined
81 (65) 43 (35)
9 (22) 34 (78)
89 (71) 36 (29)
1 5(35) 28 (65)
Homogeneity Homogeneous
Inhomogeneous Halo
19(14)
Dilated glands or cysts
11
3(7)
(8)b
4 (9)
24 (32)b 15 (2O)b
16(37) 9 (21)
Calculi foci (4 mm and with shadowing)
Echogenic
calculi. Echo-
zone of the prostate hyper-
1 : Sonographic
TABLE
a full
The area of the biopsy or transurethral resection of the prostate was correlated with the histologic diagnosis of either malignancy or benignancy and the lesion was characterized with regard to the following sonographic features: (1) echogenicity, (2) definition of the margins, (3) textural homogeneity, (4) presence or absence of a thin hypoechoic rim or “halo” around the abnormal focus, (5) presence of
dilated glands
AJR:i46,
AL.
for use with
CA). Scanning tech[2, 3]. After a digital
the
ET
m brackets
a
echogenicity
is obtaned
by adding
the components
of the
lesions. For example, 30% of the benign lesions we essentily
mixed
purely hyper-
echoic whereas 55% contan hyperechoty. b is relative to 74 rather than 125.
or hypo-echogenicity,
One hundred
and three
benign sites were analyzed in the 74 patients with benign and only the malignant site was analyzed in the 43 patients
with malignancy. No attempt was made to prospectively presence or absence of cancer.
predict the
In order to obtain direct sonographic-pathologic correlation, specimen sonograms were performed on six consecutive radical prostatectomy specimens containing cancer. Specimens were immersed in saline and scanned with a 7.5-MHz real-time sector probe (Diasonics 400 DRF, Milpitas, CA), and images were recorded on a multiformat camera. The specimens were imaged in both transverse and longi-
tudinal planes. The prostate gland was systematically sectioned for comparison with the images so as to map out the exact extent of the cancer.
Results This prostate gland in all of the patients had some alteration from normal homogeneous, low-level echogenicity [2]. Table 1 summarizes the sonographic features for the lesions. There was considerable overlap of sonographic charactenstics of benign and malignant lesions. Mixed echogenicity, which could be almost any combination of hyper-, hypo-, and iso-echogenicity, was the most common pattern in cancer (65%), but represented only 28% of the benign conditions. This is indirectly reflected in the greater homogeneity seen in BPH lesions. Hyperechogenicity was present in 55% of benign lesions and 58% of malignant lesions when considering the components of the mixed lesions. However, hypoechogenicity was much less frequent in benign disease (24%) than in malignant disease (58%). There was also a notable difference in the location of the hypoechoic lesions between the benign and malignant groups. In the cancer group, most (58%) were posterior or posterolateral, with the
rest (42%) being anterosupenor. Only 1 8% of the hypoechoic lesions in the benign group were posterior or posterolateral. Overall, lesion margins were definable in 65% of the benign group, whereas only 22% were definable in the malignant group. The margin of the lesion within the gland was analyzed. The best definition of a lesion was obtained in nodules surrounded by a thin hypoechoic rim or “halo”; however, this was seen in only a minority of the lesions. Margins of a lesion, when
definable,
were
usually
delineated
by a change
in ech-
ogenicity. Decisions concerning the extent of a lesion were subjective and varied significantly from one observer to another. To a large extent, the homogeneity of a lesion paralleled the delineation of a lesion; the more inhomogeneous, the more indistinct were the margins. Eighty-nine percent of benign lesions were homogeneous, whereas only 20% of malignant lesions were homogeneous. A thin hypoechoic rim or “halo,” a feature not previously described in prostate nodules, was observed in 1 9 of the benign nodules. To be considered a halo, the rim had to be completely circumferential (Fig. 1). Three malignant nodules were
encountered
with
what
could
be considered
a halo,
although one had a much thicker rim than that typical of those encountered in benign lesions. Dilated prostatic glands or cysts are frequently seen microscopically. On sonography, these were identified in 1 2 of the 74 patients with benign disease and in four of the 43 patients with cancer. Even in the glands with cancer, the dilated glands were
in benign
a malignant Punctate,
areas;
however,
one cyst was
focus and contained bloody fluid. brightly echogenic foci, presumably
definitely
within
correspond-
TRANSRECTAL
AJR:146, June 1986
SONOGRAPHY
OF
PROSTATIC
LESIONS
1189
Fig. 2.-Sonogram showing thick, echogenic focus (open arrows) with shadowing possibly due to calcification. Surrounding hypoechoic area (closed arrows) was cancer. Sagittal plane. SP = symphisis publs, B = bladder.
Fig. 1 .-Sonogram ing slightly hyperechoic =
seminal
showing thin hypoechoic rim or “halo” (arrows) surroundBPH nodule. Sagittal plane. SP = symphisis pubis, SV
vesicle.
ing to corpora amylacea deposits or areas of calcification, were frequently seen in both benign and malignant lesions. We specifically analyzed for the presence or absence of thick (>4 mm), brightly echogenic (equal to or greater than the surrounding periprostatic tissue) foci with and without shadowing. In the cancer group, only those foci within or in close proximity to the cancer were included (Fig. 2). Because of the frequent difficulty in defining the extent of the cancer, it was often difficult to ascertain whether the foci were actually within or just adjacent to the cancer. Sixteen (37%) glands in the cancer group had thick, brightly echogenic foci, nine of which had shadowing. Twenty-four (32%) were identified in the benign group, 1 5 of which had shadowing. The foci may well be due to benign material, but they are certainly not an indication that the adjacent or surrounding glandular tissue is benign.
Benign
Prostatic
Hyperplasia
(BPH)
Typically in hyperplasia there was symmetric enlargement of the gland and the gland assumed a more rounded configuration than normal. In the majority of cases, discrete nodules could be identified, some with a “nodule within a nodule” appearance. Nodule echogenicity was most frequently hyperechoic or isoechoic and was least often hypoechoic. Inhomogeneous nodules were not uncommon. Distinction between the hyperplastic central periurethral glands and the compressed peripheral glands was often possible. The peripheral
zone usually had lower echogenicity relative to the centrally located BPH, but some also have greater echogenicity (Fig. 3). Prostatic calculi, identified as bright echogenic foci that were either solitary or clustered, were frequent. These were located anywhere within the gland, but were uncommon in the posterior peripheral gland. They were usually located in two areas: (1) arranged in an arc or ring around BPH nodules, usually between the central and peripheral gland, and (2) periurethral. Prostate margins were generally smooth.
Carcinoma BPH frequently coexisted with cancer; thus the size of the gland ranged from small to markedly enlarged. Most cancers (72%) were located posteriorly or posterolaterally. Diffuse cancers (21 %) usually had a significant posterior component. The remaining cancers were anterosuperior. Cancer foci most often had mixed echogenicity, frequently with a significant hypoechoic component (Fig. 4). When the hypoechoic component of a lesion was within the peripheral zone of the gland (posterior or lateral), it was highly suggestive of cancer. Purely hypoechoic lesions were not uncommon and, when posterior or posterolateral, were highly suggestive of cancer. Asymmetry of the gland was usually not apparent with the linear array
probe,
were
as frequent
benign
but
was
helpful
when
in the gland
with
identified.
cancer
Prostatic
as in those
calculi
with
disease.
In Vitro Study
of Prostatic
Carcinomas
Pathologic analysis of the six prostatectomy specimens demonstrated good correlation between the location of the bulk of the tumor in the specimen with the abnormality on the sonogram. Pathologically, the tumor was usually slightly more extensive than predicted by sonography because of micro-
1190
BURKS
showing BPH, all in sagittal plane. SP = symphisis vesicle. A, WelI.defined, fairly homogeneous, hyperechoic nodule (arrows). B, Central BPH (dashed line). Lesion is composed of multiple nodules that are both hyperechoic and have a nodule-within-a-nodule appearance. Dilated glands are evident at the superiormargin(arrowheads). Peripheral Fig. 3.-Sonograms
pubis, sv
=
seminai
ET
AJR:146, June 1986
AL.
arrows) has slightly greater echogenicity than adjacent central BPH. C, Homogeneous central BPH. Peripheral zone is slightly less echogenic. Small echogenic foci are present (arrowheads), presumably caused by corpora
zone (straight
am.
compensation
or recording medium accounted for the hypoof these carcinomas observed in vivo.
echoic appearance Discussion
Our results expand the spectrum of sonographic appearances of carcinoma beyond those described by Rifkin et al., who reported no hypoechoic cancers and indicated thick, brightly echogenic foci and shadowing to be highly suggestive of benignancy [2]. In our series, hypoechogenicity was a frequent
finding
in cancer
(58%).
If a hypoechoic
lesion
was
posterior
or posterolateral in the peripheral zone, the most common site for cancer to develop, it was highly suggestive of cancer [8], but if it was elsewhere, such as central, it was more likely to represent benign disease. Purely hypoechoic
cancers
were
not uncommon.
Others
have
reported
hypoech-
oic cancers, but they used either transabdominal sonography or a radial scanner and they lacked direct sonographic histologic correlation [5, 9]. In addition, we found that thick, brightly echogenic foci with or without shadowing within or adjacent to a lesion did not exclude cancer. Because of the infiltrating nature of prostate carcinoma, a carcinoma may encompass calculi or develop in close proximity to calculi. Prostate carcinoma may also produce corpora amylacea. Fig. 4.-Carcinoma with mixed echogenicity. Lesion is posterior. There is a hyperechoic area (closed arrows) within a large hypoechoic area (open arrows). Exact extent ofcancer is difficult to define. SP = symphisis publs, B = bladder.
scopic infiltration at the periphery. All of the the six specimens had a hypoechoic area on sonogram (5 MHz) and the in vitro sonogram 5). This finding suggests that it is unlikely
cancers within both the in vivo (7.5 MHz) (Fig. that faulty gain
Our results,
using
linear-array
transducers,
indicate
consid-
erable overlap in the sonographic features between benign and malignant disease; however, some investigators, using radial scanners, have reported a high degree of accuracy. Brooman et al. [1 ] reported 96% overall accuracy, and Harada et al. [5] reported 89% overall accuracy. Rifkin, in a comparison of linear array and radial scanners, found the radial scanner to more clearly define asymmetry and capsular ex-
MR:146,
June 1986
TRANSRECTAL
SONOGRAPHY
Fig. 5.-A, Purely hypoechoic cancer. Sagittal plane. Small tumor is in the peripheral zone posteriotiy (arrows). There is central BPH. SP = symphisis pubis, SV = seminal vesicle. B, Specimen sonogram of same prostate as in 5A but in a transverse orientation using a 7.5-MHz sector probe with a water-path offset, scanned from its posterior aspect. Prostate margins are identified by
tension [1 0], a finding that is supported by our experience with specimen sonograms. Asymmetry in our series was easily detected on specimen sonograms performed in the transverse plane, but was difficult to discern on longitudinally oriented images. Although a technique for precise needle placement in the prostate for biopsy has been described for radial scanners [1 1], the longitudinal orientation and real-time capability of the linear-array probe make it more versatile in biopsy guidance. Prostate biopsy guidance is perhaps the most significant role of transrectal sonography at its present stage of development [10, 12].
In summary, although certain sonographic features may be more suggestive of benign or malignant disease, there are no specific features for either. Thus biopsy of all suspicious palpable lesions of the prostate gland is required. ACKNOWLEDGMENTS
The authors thank the Toshiba Corporation for the use of the equipment that made this study possible. We also thank Monica Harper for preparation of the manuscript and John Bobbitt for photographic assistance. We appreciate the efforts of the urology staff and house staff for referral of the patients. Drs. Linza Killion and Randall Falk are especially thanked for their contributions to this study. REFERENCES
1 Brooman PJC, Griffiths GJ, Roberts E, Peeling WB, Evans K. Per rectal ultrasound in the investigation of prostatic disease. .
OF
PROSTATIC
LESIONS
open arrows. Hypoechoic gross specimen obtained
cancer
1191
(closed arrows). C, Corresponding transverse plane and sectioned at approximately the same level as the sonogram. The cancer(straight arrows) is located in the peripheral zone. lkethra (curved arrow) and hyperplastic central tissue is identified
in the
are demonstrated.
Clln Radio!
1981;32:669-676
2. Rifkin MD, Kurtz AB, Choi HY, Goldberg sonic evaluation
spective
of the prostate
evaluation
using
and acoustic
BB.
Endoscopic
a transrectal
characterization.
ultra-
probe: proRadiology
1983;149:265-271
3. Rifkin MD, Kurtz AB, Goldberg BB. Sonographically guided transpenneal prostatic biopsy: preliminary experience with a longitudinal linear array transducer. AJR 1983;140:745-747 4. Sekine H, Oka K, Takehara V. Transrectal longitudinal ultrasonotomography of the prostate by electronic linear scanning. J Uro! 1982;1 27:62-65 5. Harada K, Tanahaski evaluation
Y, Igari
of inside
echo
D, Numata
I, Orikasa
ography. 6. Fritzsche
J Urol 1980;124:216-220 PJ, Axford PD, Ching VC, Rosenquist Correlation of transrectal sonographic findings
suspected and 1983;1 30:272-274
S. Clinical ech-
patterns in gray scale prostatic
unsuspected
prostatic
RW, Moore RJ. in patients with disease. J Uro!
7. Spimak JP, Resnick Ml. Transrectal ultrasonography. Urology 1984;23:461 -467 8. McNeal JE. The prostate gland: morphology and pathobiology. Monogr Urol 1983;4(1): 1-3 9. Greenberg M, Neiman HL, Vogelzang A, Falkowski W. Ultrasonographic features of prostatic carcinoma. J Clin Ultrasound 1982;1 0:307-312 10. Rifkin MD. Transrectal prostatic ultrasonography: comparison of linear array and radial scanners. J Ultrasound Med 1985;4: 1-5 1 1 . Holms HH, Gammelgaard J. Ultrasonically guided precise needle placement in the prostate and seminal vesicles. J Urol 1981;1 25:385-387
12. Fomage
BD, Touche
OH, Deglaire
M, Faroux
MC, Simatos
A.
Real-time ultrasound-guided prostatic biopsy using a new transrectal linear-array probe. Radiology 1983;146:547-548