Combined role of transrectal ultrasonography, gleason score, and prostate-specific antigen in predicting organ-confined prostate cancer

Combined role of transrectal ultrasonography, gleason score, and prostate-specific antigen in predicting organ-confined prostate cancer

suggests that patients with “minimal disease” fare better with combination therapy than those with extensive disease and correctly concludes that a pr...

892KB Sizes 0 Downloads 1 Views

suggests that patients with “minimal disease” fare better with combination therapy than those with extensive disease and correctly concludes that a prospective trial involving these patients with minimal disease is clearly warranted. Overall, this study adds additional credence to the concept that complete androgen deprivation results in a statistically significant increased overall survival and perhaps a better quality of life for those patients with disseminated prostate cancer. An ongoing U.S. Intergroup trial investigating or-

130

chiectomy plus flutamide versus orchiectomy plus placebo is nearing completion. The results of this large Intergroup trial which lnvolves orchiectomy in both arms of the randomization will substantially complete the initial phase of investigation of complete androgen deprivation. Robert C. Flanigan, M.D. Loyola University Medical Center Maywood, Illinois 60153

UROLOGY

I AUGUST1993 / VOLUME42, NUMBER2

COMBINED ROLE OF TRANSRECTAL ULTRASONOCRAPHY, GLEASON SCORE, AND PROSTATE-SPECIFIC ANTIGEN IN PREDICTING ORGAN-CONFINED PROSTATE CANCER J. STUART WOLF, JR., M.D. KATSUTO SHINOHARA, M.D. PETER R. CARROLL, M.D. PERINCHERY NARAYAN, M.D. From the Departments of Urology, University of California School of Medicine, and Department of Veterans Affairs Medical Center, San Francisco, California

cancer staging is frequently inaccurate. By combining transrectal ultrasonography (TRUS) with a retrospectively derived grade-stratified prostate-specific antigen (PSA-GS) scale, we demonstrated 77 percent staging accuracy in 155 men with clinically localized prostate cancer undergoing radical prostatectomy. When used as the first step in a staging algorithm, PSA-GS (Score 2 7: PSA 2 4.0 ng/mL, uncontained; Score = 5 or 6: PSA 2 8.0 ng/mL, uncontained; Score 5 4: PSA 2 16.0 ng/mL, uncontained) had a sensitivity of 75 percent and a specificity of 72 percent. The addition of TRUS to the staging algorithm, necessary only in patients with negative predictions by PSA-GS (46%), increased the sensitivity to 86 percent and the negative predictive value to 79 percent, while positive predictive value was unchanged at 77 percent. The combination of TRUS with PSA c 4 ng/mL or 2 16 ng/mL identified subsets of patients with 85 percent and 88 percent likelihood of contained and uncontained disease, respectively. Our algorithm minimizes operator dependency by requiring TRUS in less than half of the patients. It produced improved staging, but the overall results were inaccurate in 23 percent of patients. Further refinements in prostate cancer staging are still necessary.

ABSTRACT-Prostate

Despite the proliferation of imaging and biochemical tests, the clinical staging of prostate cancer remains inadequate. Digital rectal examination (DRE) and computed tomography lack sensitivity.’ Magnetic resonance imaging is expensive, time-consuming, and has only a 69 percent overall accuracy in staging local disease.2 Transrectal ultrasonography (TRUS), although initially very promising, has had a disappointing performance as reported in recent studies.2-4 Prostate-specific antigen (PSA) is not sufficient for use as an independent predictor of cancer stage.5-8 PSA density (PSAD), the PSA divided by the prostate volume, has been applied to the detection and staging of prostate cancer.9,‘0 So far the utility of PSAD for staging has not proved superior to using PSA Submitted: February 28, 1993

UROLOGY

/ Aumr

22, 1993, accepted (with revisions): April

I993

i VOLL~ME 42, NUMBER 2

alone. The general weakness of all current staging modalities is insufficient sensitivity for locally advanced disease. Cooner and associatesl’ have demonstrated that the combination of TRUS, PSA, and DRE is better than any single method for the detection of prostate cancer. We postulated that combining methods also might be more effective for prostate cancer staging. Intrigued by reports describing the relationship of PSA to tumor grade,8l12 we stratified PSA values by grade to produce more accurate staging. The goal of this retrospective study was to devise a way of combining modalities to maximize staging accuracy in patients with localized prostate cancer. MATERIAL AND METHODS The study group consisted of 155 men, ranging in age from forty-five to seventy-five years (mean,

131

65.4 years). They were selected from 208 patients under oing radical prostatectomy for clinically localize 8 prostate cancer (Stage A or B) at our institutions, in whom TRUS and PSA measurements were part of the preoperative evaluation. Fortythree patients were not included because TRUS did not include estimation of prostate volume (24), PSA was obtained less than three weeks after biopsy (6), or patients had received neoadjuvant hormonal therapy (23). TRUS was performed using a 7.5 MHz biplaner (Teknar, Inc., St. Louis, MO or Siemans Quantum, Inc., Issaquah, WA) or multiplaner probe (Diasonics, Inc., Milpitas, CA). Sonography was performed by one of us (K.S.) in all but 9 patients. It was the impression of the ultrasonographer that image quality did not differ significantly between the three machines. The glands were sized by multiplying the product of the greatest anteroposterior, transverse, and superior-inferior dimensions by 0.52 (prolated ellipse formula). For staging, criteria published previously were used.’ Carcinoma was suggested by hypoechoic regions in the prostate. Bulging, irregularity, or obscuring of the prostatic capsular margin in the hypoechoic area indicated extracapsular extension. When the hypoechoic lesion abutted and adhered to the seminal vesicles, or if the seminal vesicles displayed posterior convexity, tumor involvement was predicted. Isolated asymmetry of either the prostatic margin or the seminal vesicles, in the absence of any of the aforementioned signs of extraprostatic tumor, was not indicative of uncontained disease. If either extracapsular extension or seminal vesicle involvement was noted on TRUS, uncontained disease was predicted. PSA was measured in blood samples drawn prior to TRUS with the Tandem-R monoclonal immunoradiometric assay (Hybritech, Inc., San Diego, CA). Biopsy specimens were obtained by both digital and ultrasonographic guidance. A TRUS-guided biopsy was not always performed if TRUS findings were consistent with the digital rectal examination and a biopsy positive for cancer already had been obtained. Systematic or extraprostatic biopsies were not performed routinely; if specimens were obtained, localization data that might have been gleaned from the results were not used in this analysis. In 8 patients only transurethral prostatectomy specimens were available preoperatively. Biopsy scores were determined by adding the grades of the predominant 2 Gleason patterns. In 5 patients, only fine-needle aspiration was performed. Well-differentiated cytologic findings were considered equivalent to a 132

Gleason score I 4, moderately differentiated was equivalent to a score of 5 or 6, and poorly differentiated was equivalent to a score 2 7.13 The prostate specimens after radical prostatectomy were step-sectioned into 3-mm to 5-mm slices after inking the surgical margin. No attempts were made to quantify the tumor mass exactly, Since our intent was to determine the ability to stage patients preoperatively, the Gleason score of the biopsy specimen was used in our analysis rather than the Gleason score of the tumor in the prostatectomy specimen. Pathologic stages were assigned. Uncontained tumors were those that extended into the periprostatic adipose tissue (extracapsular extension), involved the surgical margins, invaded a seminal vesicle, or were accompanied by lymph node metastases. All others were considered to be contained tumors, including those with invasion of the capsule without periprostatic adipose tissue involvement. “Positive” in the following discussion indicates the prediction of uncontained disease; “negative” indicates the prediction of contained disease. All predictions could, therefore, be classified as true positive, false positive, true negative, or false negative. Sensitivity is the percentage of patients correctly predicted to have uncontained tumors. Specificity is the percentage correctly predicted to be contained. Positive predictive value (PPV) and negative predictive value (NPV) indicate the likelihood of correct staging when the prediction is uncontained or contained, respectively. Accuracy is the percentage of correct predictions in the whole group. Association of extraprostatic tumor with biopsy score and PSA categories was assessed with the chi-square test. l4 The Mann-Whitney Utest was used to compare PSA and PSA density values between the patients with contained and uncontained tumors.14 Youden’s index, which reveals that the point of least error occurs when the sum of sensitivity and specificity is highest,15 was used to determine the best cutoff values for PSA and PSA density, both in the group as a whole and in subgroups stratified by biopsy score. RESULTS Of the 155 patients, 68 (44%) had contained tumors and 87 (56Oh) had uncontained disease (Table I). There were 10 patients (6%) with microscopic lymph node metastases; among them, 4 also had seminal vesicle involvement, 5 had extracapsular extension or positive margins without seminal vesicle involvement, and I had disease that was otherwise confined within the prostate. UROLOGY

/ AUGUST1993

I VOLUME42, NUMBER2

TABLE I.

Pathologic results in 155 patients with clinically localized prostate cancer undergoing radical prostatectomy Seminal Vesicles

Lymph Nodes

Surgical Margin

Lymph node or seminal vesicle involvement + + + + + + + + + + + + + + + + Surgical margin involvement -

f +

contained

3 0 0

+ + _

_ + + _

+ -

+ +

f _

-

+ _

No. of Pts. (Percent)

4 0 1

+ -

7 3 3 2 25(16)

extension

+ + -

_ -

Pathologically

or extracapsular

Extracapsular Extension

+ _

32 (21) 16 (10)

I

14

-

(9)

68 (44)

+Pathologicinvolvementby tumm by tumor -No Pathologicinvolvement

Seminal vesicle involvement was noted in 19 patients (12%), all but 2 of whom also had positive margins, extracapsular extension, or lymph node involvement. Sixty-two patients (40%) had either extracapsular extension or surgical margin involvement, without seminal vesicle or lymph node involvement. Of this group, 32 had both extracapsular extension and positive surgical margins, 14 had extracapsular disease with surgical confinement, and 16 had positive surgical margins without extension beyond an identifiable capsule. Among this last group, surgical margin involvement was at the bladder neck in 1 patient, at the urethral margin in 3, at the apex of the prostate in 4, and posterior or posterolateral in 8. The pathologic results are segregated by biopsy score and PSA in Table II. Both the biopsy score and the PSA

TABLE’IL Score

PSAL

16*

57 5-6 <4

719 11114 5/6 23129

(78)t (79) (83) (79)

TOTALS

groupings were significantly associated with the pathologic stage of the tumor (p < 0.001, chisquare test).

TRUS staging of disease consisted of prediction of extracapsular extension and evaluation for seminal vesicle involvement. If disease was predicted at any extraprostatic site that was confirmed pathologically, the prediction was considered true positive. There were 6 patients with uncontained disease in whom the TRUS predicted extraprostatic disease only at incorrect sites. These six predictions were considered to be false positive, even though the patients had uncontained disease. The results of TRUS staging in 18 patients with the sonographically undetectable

kathologic results segregated by biopsy score and PSA PSA 8.0-l 8112 22129 o/4 30145

5.9 (67j (76) (0) (67)

PSA 4.0-7.9 12114 8118 4113 24145

(86) (44) (31) [53)

PSA<

4.0

l/4 8120 1112 10136

(25) (40) (8) (28)

Total 28139 49181 10/35 871155

(72) (60) (29) (56)

*Ng/mL. fPatients with pathologically uncontained disease/number ofpatients in the subgroup(percent).

UROLOGY

I AUGUST1993 / VOLUME 42, NUMBER 2

133

TABLE III.

Staging by transrectul ultrasonography

Predictions Site

FP

TP

Seminal veSicles 7’ 5 Extracapsular extension 16t 40 Overall 42 17*

Descriptions

TN

FN

Sensitivity

Specificity

PPV

NPV

Accuracy

130

13

26%

96%

42%

91%

87%

58

41

48%

81 %

71%

59%

63%

57

39

48%

84%

71 %

59%

64%

*Includes 1 patient with seminal vesicle invasion only contralateral to the side predicted. tlncludes 2 patients with extracapsular extension only contralateral to the side predicted. ?lncludes, in addition to the 3 patients noted above, 3 more patients with extraprostatic disease only at

ANTIGEN

The pre-TRUS PSA values in this group ranged from 0.3 ng/mL to 120 ng/mL. The mean value for patients with pathologically contained disease was 8.6 + 15.1 (s,d.) ng/mL. For patients with uncontained disease, the mean was 15.2 + 16.4 (s.d.) ng/mL (p < 0.001, Mann-Whitney U-test), for an overall mean of 12.3 k 16.1 (s.d.) ng/mL. When

TABLE IV.

difJmnt~mn

those

predicted.

applied to the entire patient group, a PSA cutoff value of 6.0 ng/mL most accurately differentiated between patients with contained and uncontained prostate cancer (Table IV). The PSA data were then stratified by biopsy score and the optimal cutoff value (as determined by Youden’s index15) was assigned to each group. We found that multiples of 4 ng/mL (the upper limit of normal) were the best cutoff values in the grade-stratified groups! Score 17: PSA 2 4.0 ng/mL, uncontained Score = 5 or 6: PSA 2 8.0 ng/mL, uncontained Score 5 4: PSA 2 16.0 ng/mL, uncontained. This staging method, called “PSA, grade-stratified” (PSA-GS), had a sensitivity of 75 percent and a specificity of 72 percent (Table IV). PSA density, the PSA value divided by the sonographic prostate volume, was calculated. The mean PSAD for patients with contained disease was 0.22 +_0.28 (s.d.), with uncontained disease 0.47 f 0.58 (s.d.) (p < 0.001, Mann-Whitney Utest), and for the whole group 0.36 + 0.49 (s.d.).

tumors (of whom 6 had uncontained disease pathologically) were considefed predictive of contained disease. Overall, TRUS had a sensitivity of 48 percent and a specificity of 84 percent (Table III). Besides the 18 patients with sonographically undetectable tumors, there were 16 patients in whom uncontained disease was pr’esent contralatera1 to the sonographically hypoechoic lesion. Seven of the 16 had a TRUS predictive of contained disease. There were 27 patients in whom the preoperative biopsy specimens demonstrated cancer contralateral to the hypoechoic lesion (including 7 of the aforementioned 16 patients). PROSTATE-SPECIFIC

sites

Staging methods Descriptions

Predictions Method

TP

FP

TN

FN

Sensitivity

Specificity

PPV

NPV

Accuracy

PSA 2 6.0’ PSA-GS PSAD 2 0.25 PSAD-GS TRUS Algorithm

67 65 51 56 42 75

29 19 15 15 17t 23

39 49 53 53 57 45

20 22 36 31 39 12

77% 75% 59% 64% 48% 86%

57% 72% 78% 78% 84% 66%

70% 77% 77% 79% 71% 77%

66% 69% 60% 63% 59% 79%

68% 74% 67% 70% 64% 77%

Either TRUS + and PSA 2 4

83

45

23

4

95%

34%

65%

85%

6896

Both TRUS + andPSAr 16

14

2

66

73

16%

97%

88%

47%

52%

Key’TRW+ = TRUSpredictive 01 uncontained disease. *PSA in ng/mL. fSix patients with extraprostatic disease predicted by TRUS had disease only at other sites pathologically. The test was consideredJalse positive in these 6 patients WUI though the tumcw was uncontained

134

UROLOGY

/ AUGUST1993

/ VOLUME 42, NUMBER 2

Clinically

Localized

Prostete Cancer

I _ Biopsy Score

TRUS

I

Contained

I

1 I

Uncontained

_A + FIGURE 1. Algorithm for staging of clinically localized prostate cancer. PSA units are ng/mL (monoclonal assay); “+ ” indicates that uncontained disease is predicted; “-” indicates that contained disease is predicted.

A value of 0.25 was the best cutoff value between contained and uncontained cancer for the entire group (Table IV). The PSAD data were also stratified by biopsy score, with optimal cutoffs determined with Youden’s index. l5 Score L 7: PSAD L 0.15, uncontained Score = 5 or 6: PSAD 2 0.30, uncontained Score 54: PSAD L 0.45, uncontained. This method, called “PSAD, grade-stratified” (PSAD-GS), had a sensitivity of 64 percent and a specificity of 78 percent (Table IV). STAGINGALGORITHM We first found the combination of tests that provided the best overall staging results. Once this was established, subgroups were identified in which predictive values could be maximized. Finally, the combinations were used to construct an algorithm that minimized testing while maintaining staging ability The algorithm is depicted in Figure 1. The first step is to separate the patients into three groups based on the biopsy score (I 4, 5-6, or 2 7). The

UROLOGY

I AUGUST 1993 / VOLUME 42.

NL~MBER2

PSA-GS scale is then applied by comparing the patient’s PSA to the cutoff value for his biopsy group. If the PSA is above the cutoff, the tumor is predicted to be uncontained and evaluation can stop. If the PSA level is below the cutoff value, the patient moves on to TRUS evaluation in order to determine the preoperative stage. For example, if a patient’s biopsy score is 4 and his PSA is 19 ng/mL, he is predicted to have uncontained disease solely on the basis of a PSA 2 16 ng/mL (Fig. 1, one arm from left in center section). If the same patient’s PSA is 14 ng/mL, however, TRUS is necessary to determine the preoperative stage because the PSA is < 16 ng/mL (Fig. 1, left-most arm in center section). In our series, the biopsy score combined with PSA would have predicted 84 patients (54%) as having uncontained disease. TRUS would have been necessary only in the remaining 71 patients (46%). Of these patients, 14 (20%) had TRUS indicative of uncontained disease (10 true positive, 4 false positive). The result of this staging algorithm among our patients would have been: sensitivity 86 percent, specificity 66 percent, positive predictive value 77 percent, and negative predictive value 79 percent (Table IV). The subset of patients with PSA < 4 ng/mL and TRUS predictive of contained disease (27 patients) had an 85 percent incidence of contained disease (Table IV). Among the 29 patients with a PSA 2 16 ng/mL, those with a TRUS indicating uncontained disease (16 patients) had an 88 percent likelihood of uncontained disease (Table IV). OTHER METHODS

We have found sonographic tumor volume to be an excellent indicator of the risk of lymph node metastases,‘” but it was not effective for local staging in this study We previously have described a “PSA formula,” a method of comparing patients’ actual PSA values to predicted PSA values for prostate cancer staging. l7 Our preliminary results with this technique were excellent in a small series that included some patients with clinical Stage C tumors and the few who had received neoadjuvant hormonal therapy In this larger series, when clinically advanced or hormonally treated patients were excluded, the PSA formula in combination with TRUS yielded results equivalent to the PSA-GS/TRUS algorithm. Because of its complexity and lack of substantial improvement over other (simpler) methods, the PSA formula was not incorporated into the staging algorithm. Besides the aforementioned algorithm, we evaluated several other schemes involving absolute PSA

135

cutoffs, biopsy score cutoffs, PSAD, PSAD-GS, and PSAD applied to groups stratified by PSA. None were as effective as the one chosen. PATHOLOGICALLYUNCONTAINED TUMORS

The sensitivity for detecting uncontained disease with our algorithm was 86 percent in the entire group. Among the 10 patients with lymph node metastases, the sensitivity was 100 percent. The sensitivity was 84 percent in the 19 patients with seminal vesicle involvement. As mentioned, the 62 patients with extraprostatic disease solely on the basis of extracapsular extension and/or surgical margin involvement could be divided into three groups: 32 with both extracapsular extension and positive surgical margins, 14 with surgically confined extracapsular disease, and 16 with positive surgical margins but without extension beyond an identifiable capsule. The sensitivities for detecting uncontained disease in these groups were 88 percent, 93 percent, and 75 percent, respectively Although some of the involved surgical margins in the latter group may have represented technical error rather than the true stage of the tumor, our staging algorithm could not reliably differentiate these patients from those with pathologic extracapsular extension of the tumor. COMMENT The results of the prospective analysis of TRUS together with a retrospective analysis of PSA data indicate that the combination of TRUS with a PSA scale stratified by biopsy score yielded predictive values in excess of 7.5 percent. Using the tests in an algorithm format, TRUS would have been necessary in less than one half of our patients to complete the staging. Those patients with low PSA (c 4 ng/mL) and a TRUS indicating contained tumor had the highest likelihood of contained disease (85%). The highest incidence of uncontained disease (88%) occurred in those patients with high PSA (2 16 ng/mL) and TRUS indicating uncontained tumor. These figures, 85 percent and 88 percent, are an improvement over the predictive values of 72 percent and 79 percent of a PSA level < 4 ng/mL and 2 16 ng/mL, respectively, without TRUS. Despite the potential usefulness of PSAD for the detection of prostate cancer: we and otherslo have not found it to be more effective than PSA for staging. Although PSAD alone was slightly better than PSA alone in our analysis, PSA was superior when combined with biopsy score. Stratifying by biopsy score may have been effective because PSA production by a tumor is determined, in part, by

136

grade. It has been shown that PSA is inversely correlated to tumor grade when corrected for volume.8 Since higher grade tumors also tend to be larger, however, PSA correlates positively with grade if not corrected for volume. Consistent with previous observations, the pathologic stage was more likely to be advanced in our patients with higher biopsy scores (Table II). A lower PSA cutoff for tumors of higher grade may have been effective for this reason as well. By whatever mechanism, stratification by biopsy score decreased the overlap between the PSA distributions of patients with contained or uncontained prostate cancer and allowed better differentiation between the groups. Other investigators also have found the combination of PSA and Gleason grade to be predictive of pathologic findings.‘*J8 Of the many combinations and stratifications we evaluated, those involving both TRUS and PSA were superior. Our results of TRUS staging (sensitivity 48%, specificity 84%) are similar to those recently reported by other investigators (sensitivity 29-66%; specificity 46-90°h).2-4 We agree with these investigators that TRUS alone, with current technology, is not sufficient for the staging of prostate cancer in an individual patient. Our patients may be different from those included in other studies because many of the patients in this series already were known to have cancer at the time of TRUS at our institution. Biopsy artifacts were noted in approximately 10 percent of such cases. Such artifacts made the assessment of the extent of tumor involvement difficult in only a small number of patients. Transurethral resection defects were not problematic in this regard. TRUS is highly operator-dependent. A large experience is required to achieve the results reported by academic centers with dedicated urologic ultrasonographers. One advantage of our algorithm is that TRUS would be required in less than one half of patients to obtain most of the staging information. In the 46 percent of patients who would have required TRUS, 20 percent were predicted to have contained tumors based on TRUS alone. The positive predictive value was 71 percent in these patients. Such selective use of TRUS is especially applicable at institutions where TRUS is not readily available or reliable. At our cancer referral centers, with an experienced urologist ultrasonographer, we obtain TRUS in all patients being considered for local treatment of prostate cancer. The information gained from TRUS, even in a patient who is likely to have uncontained disease based on PSA data, is helpful to us in planning treatment.

UROLOGY

I AUGUST1993 I VOLUME42,NUMBER2

TRUS characterizes the prostatic tumor anatomically, and PSA characterizes it biochemically The combination of both attributes best predicts local tumor stage. One or the other will detect most cases of pathologically uncontained tumor. The usefulness of TRUS and PSA is enhanced by an algorithm that combines the two. At the extremes of PSA (< 4 ng/mL and 1 16 ng/mL), TRUS and PSA together identify pathologic stage with predictive values approaching 90 percent. Although we were able to achieve these high predictive values in slightly more than one quarter of patients, the overall positive and negative predictive values for the whole group (77% and 79%, respectively) were disappointing. Further advances are required before the urologist can confidently stage prostate cancer in the majority of patients. Only 16 percent of the patients in our series had uncontained tumor on the basis of seminal vesicle or lymph node involvement. The chance for cure is reduced in these patients. For patients with only extracapsular extension and/or surgical margin involvement, however, the risk for cancer progression after treatment is less certain. The true role of staging in these patients, which comprised the majority of patients with uncontained disease in our study, is therefore unclear. Despite these caveats, we believe that staging information derived with an algorithm such as ours might be used while weighing the possible benefits of various therapeutic options for the management of prostate cancer. While we currently would not deny a potentially curative radical prostatectomy from a patient who was thought to have uncontained tumor by our algorithm, we would counsel the patient that the tumor is likely to be uncontained and that the chance for cure might therefore be reduced-but that the exact significance of some types of locally uncontained pathology is unknown. Additionally, the specific surgical approach might be altered by the preoperative prediction of extraprostatic tumor. As the PSA stratifications and the algorithm format were derived retrospectively, further prospective or blinded retrospective studies will be necessary to confirm their utility Perinchery Department

Narayan,

M.D.

ofUrology,

U-575

University of California San Francisco, CaliJornia 94143-0738 REFERENCES 1. Scardino PT, Shinohara K, Wheeler TM, and Carter SS: Staging of prostate cancer: value of ultrasonography. Urol Clin North Am 16: 713-734, 1989. 2. Rifkin MD, Zerhouni EA, Gatsonis CA, Quint LE,

UROLOGY

/ Avcusr

1993 / VOLUME42, NUMBER2

Paushter DM, Epstein Jl, Hamper U, Walsh PC, and McNeil BJ: Comparison of magnetic resonance imaging and ultrasonography in staging early prostate cancer: results of a multi-institutional cooperative trial. N Engl J Med 323: 621-626,199O. 3. McSherry SA, Levy E Schiebler ML, Keefe B, Dent GA, and Mohler JL: Preoperative prediction of tumor volume and stage in clinically localized prostate cancer: comparison of digital rectal exam, transrectal ultrasonography and magnetic resonance imaging. J Urol 146: 85-89, 1991. 4. Andriole GL, Coplen DE, Mikkelsen DJ, and Catalona WJ: Sonographic and pathological staging of patients with clinically localized prostate cancer. J Urol 142: 1259-1261. 1989. 5. Lange PH, Ercole CJ, Lightner DJ, Fraley EE, and Vessella R: The value of serum prostate specific antigen determinations before and after radical prostatectomy. J Ural 141: 873-879,1989. 6. Brawer MK, and Lange PH: Prostate-specific antigen in management of prostatic carcinoma. Urology 33: 1 l-16, 1989. 7. Hudson MA, Bahnson RR, and Catalona WJ: Clinical use of prostate-specific antigen in patients with prostate cancer J Urol 142: 1011-1017, 1989. 8. Partin AW, Carter HB, Chan DW, Epstein Jl, Oesterhng JE, Rock RC, Weber JP, and Walsh PC: Prostate-specific antigen in the staging of localized prostate cancer: influence of tumor differentiation, tumor volume and benign hyperplasia. J Ural 143: 747-752,199O. 9. Benson MC, Whang IS, Pantuck A, Ring K, Kaplan SA, Olsson CA, and Cooner WH: Prostate specific antigen density: a means of distinguishing benign prostatic hypertrophy and prostate cancer. J Urol 147: 815-816, 1992. 10. Andriole GL, Coplen DE, Telle WB, and Catalona WJ: Serum PSA index (PSAI) as a staging tool for patients with otherwise clinically localized prostate cancer (Cap) (Abstr. 317). J Ural 147: 292A, 1992. 11. Cooner WH, Mosley BR, Rutherford CL Jr, Berad JH, Pond HS, Terry WJ, lgel TC, and Kidd DD: Prostate cancer detection in a clinical urological practice by ultrasonography, digital rectal examination and prostate specific antigen. J Urol 143: 1146-1156, 1990. 12. Partin AW, Lee BR, Chan DW, Epstein JI, and Walsh PC: The use of prostate-specific antigen (PSA) and Gleason score to predict pathologic stage in men with localized prostate cancer (Abstr. 701). J Urol 147: 387A, 1992. 13. Narayan P Jajodia P Stein R, and Tanagho EA: A comparison of fine needle aspiration and core biopsy in diagnosis and preoperative grading of prostate cancer. J Urol 141: 560-563,1989. 14. Armitage P, and Berry G: Statistical methods in medical research, 2nd edition. Oxford, Blackwell Scientific Publications, 1987. 15. Youden WJ: Index for rating diagnostic tests. Cancer 3: 32-35, 1950. 16. Wolf JS Jr, Shinohara K, Narayan P and Carroll PR: Selection of patients for laparoscopic pelvic lymph node dissection (LPLND) prior to radical prostatectomy: decision analysis (Abstr. 817). J Ural 149: 417A, 1993. 17. Wolf JS Jr, Shinohara KS, and Narayan P: Prostate cancer staging: accuracy of transrectal ultrasound enhanced by prostate-specific antigen. Br J Urol 70: 534-541, 1992. 18. Kleer E, Larson-Keeler JJ, Zincke H, and Oesterling JE: Ability of preoperative serum prostate-specific antigen value to predict pathologic stage and DNA ploidy. Influence of clinical stage and tumor grade. Urology 41: 207-216, 1993.

137