OUTCOME BASED STAGING FOR CLINICALLY LOCALIZED ADENOCARCINOMA OF THE PROSTATE

0022-5347/97/1584-1422$03.00/0

THE JOURNAL OF

UROLOGY AMERICAN

Copyright 8 1997 by

UROLOGICALASSOCIATION,

Val. 158,1422-1426,October 1997 Printed I n U.S.A.

INC.

OUTCOME BASED STAGING FOR CLINICALLY LOCALIZED ADENOCARCINOMA OF THE PROSTATE ANTHONY V. DAMICO, RICHARD WHITTINGTON, DELRAY SCHULTZ, S. BRUCE MALKOWICZ, JOHN E. TOMASZEWSKI AND ALAN WEIN From the Joint Center for Radiation Therapy, Harvard Medical School, Boston, Massachusetts, Departments of Radiation Orlrology, Urology and Pathology, Hospital of the University of Pennsyluania, Philadelphia and Department of Mathematics, University of Millersville, Millersville, Pennsylvania

ABSTRACT

Purpose: Some patients with clinically localized prostate cancer are not cured after radical prostatectomy because of the presence of occult systemic disease. The American Joint Commission on Cancer staging classification for prostate cancer does not reliably distinguish between clinically localized patients who are likely or unlikely to be cured after local therapy. This project was undertaken to develop a staging system capable of predicting long-term outcome after radical prostatectomy on the basis of the clinical parameters obtained routinely during the standard workup for patients with adenocarcinoma of the prostate. Materials and Methods: A total of 688 clinically localized prostate cancer patients managed with a radical retropubic prostatectomy for adenocarcinoma of the prostate between 1989 and 1996 was evaluated for clinical features predictive of time to prostate specific antigen (PSA) failure using a Cox regression multivariate analysis. A recently defined clinical factor called the calculated prostate cancer volume and its ability to predict time to PSA failure in conjunction with PSA, biopsy Gleason score and clinical stage were evaluated. Results: The calculated prostate cancer volume (p <0.0001) and the pretreatment PSA ( p <0.001) provided the optimal staging system for predicting freedom from PSA failure after radical prostatectomy. Conclusions: The calculated prostate cancer volume a n d PSA may provide clinically useful information regarding outcome after radical prostatectomy, enabling the selection of a therapeutic approach for a n individual patient with clinically localized disease. Validation of this staging system is needed. KEY WORDS:prostatic neoplasms, outcome assessment (health care), prostate-specific antigen

In the prostate specific antigen (PSA) based screening era, the current American Joint Commission on Cancer (AJCC) staging for prostate cancer correctly identifies at most 70% of clinically localized patients with pathologic organ-confined disease.' Historically, 10 to 26% of patients with pathologic organ-confined disease sustain clinical failure during a 15year followup period.2-5 Therefore, at best only 63% [70%-0.1 (70%)]of patients with clinically organ-confined prostate cancer as per the current AJCC staging system will remain clinically free of disease 15 years after definitive local therapy. In 1969 McNeal proposed that intracapsular prostate cancer initially grows with a slow and constant doubling time, acquiring cellular heterogeneity and eventually the capacity to metastasize as a function of increasing cancer volume.6 Since 1969, McNeal6 and others7-'3 have confirmed that pathologic tumor volume is correlated with pathological stage, pathological Gleason grade, the margin status and disease progression after radical prostatectomy. Therefore, an accurate assessment of prostate cancer volume preoperatively could serve as the basis for a clinical staging system that would enable the selection of patients likely to be cured with radical prostatectomy. In 1987, Stamey et a1 reported a correlation of 0.7 between preoperative PSA and pathologic tumor volume.14 However, more recently Noldus and Stamey reported that with larger numbers this correlation did not hold for an individual paAccepted for publication February 7, 1997.

tient and at best the correlation was 0.382, 0.438 and 0.363 for patients with pathologically organ-confined disease, nonorgan-confined without lymph node metastases and with lymph node metastases, respectively.'" As a result, the ability to estimate the tumor volume prior to definitive management had remained a n unsolved problem. In a previous report, a methodology was described that provided a calculated estimate of the prostate cancer volume using clinical parameters obtained during a standard workup for prostate cancer.l6 In this study, a staging system is presented for clinically localized patients based on the pretreatment clinical parameters (PSA, calculated volume of cancer) that were found t o be the significant predictors of time to PSA failure after radical prostatectomy using a Cox regression multivariate analysis. The hypothesis tested in this study is that the staging system based on the preoperative PSA and calculated volume of prostate cancer provides improved predictability of postoperative PSA failurefree survival when compared to the current AJCC staging system.

MATERIALS AND hlETHODS

Patient selection. A total of 688 clinically localized prostate cancer patients treated with a radical retropubic prostatectomy at the Hospital of the University of Pennsylvania between January 1989 and May 1996 comprises the patient

1422

STAGING CLINICALLY LOCALIZED PROSTATE CANCER

1423

population in the time to PSA failure study. In all cases, Thus, cancer-specific PSA = PSA - [PSA from benign epistaging evaluation included a history and physical exam, thelial tissue], where the PSA from benign epithelial tissue serum PSA, bone scan, and a transrectal ultrasound-guided was calculated as defined by Lepor et a1 as: PSA from benign needle biopsy of the prostate with Gleason score histologic epithelial tissue = [(epithelialfraction) x (PSNcm." of epigrading.I7 A sextant biopsy was performed using a 17 gauge thelial tissue) x transrectal ultrasound prostate volume in Tru-Cut needle* under ultrasound guidance via a transrectal cm.")].22The values used in the above formula were epithelial approach. The serum PSA was measured using the Hybritech fraction = 0.2 as defined by Marks et a P and the PSA leak assayi.18 and was obtained on an ambulatory basis prior to into serumkm.:' benign prostatic epithelial tissue = 0.33 as the biopsy. Clinical stage was determined by digital rectal defined by Lepor et a1.2' examination and recorded using the 1992 AJCC staging.'" The prostate cancer volume (Vca) was calculated as VC;-= The pathologic biopsy data were not used to determine clincancer-specific PSA/(PSA leak into serum per cm.3 of cancer) ical stage. Pathologic analysis. The diagnostic biopsy and radical where the values in the denominator are as defined by prostatectomy specimens for all patients were reviewed by Aihara et alZ4and are a function of the Gleason grade (that a single genitourinary pathologist (J. E. T.). Prior to pros- is, Gleason grades 1 , 2 , 3 , 4 and 5 correspond to values of 20, tatectomy, all patients underwent bilateral open or lapa- 10, 4, 2 and 1 ng./ml./cm.", respectively). Example of cancer-specific PSA and calculated prostate roscopic pelvic lymph nodal sampling. If the frozen sections of any sampled lymph node were positive for cancer volume calculation. Given a patient with PSA 7.2 carcinoma, then the radical prostatectomy was aborted. At ng./ml. (Hybritech assay) and transrectal ultrasound pros~ , sum 3 + 2 = 5 in 313 cores least 6 lymph nodes were examined pathologically from tate volume 60 ~ m . Gleason each side using the laparoscopic approach. Prostatectomy on the right, Gleason sum 3 + 3 = 6 in 2/3 cores on the specimens were weighed, measured, inked over the entire left, weighted Gleason score ((5 X 3) + (6 x 2)]/5 = 5.4 or a surface and fixed in 10% buffered formalin. Both the apical weighted Gleason grade 2.7 and PSA leak into serum per and basal margins were amputated to a thickness of 5 mm. ~ mof.cancer ~ 5.8 for a weighted Gleason grade of 2.7, thereand sectioned parasagittally in a direction perpendicular fore, cancer-specific PSA = 7.2 ng./ml. - (0.2)(0.33)(60)ng./ to the initial transverse incision at 3 mm. intervals. The ml. = 3.2 ng/ml and the calculated volume of prostate canbase of the seminal vesicles was amputated, and the basal cer = 3.2 ng./m1./(5.8 ng./ml./~m.~) = 0.55 ~ m . ~ cross section was submitted for microscopic analysis. The Followup. The median followup for 688 patients was 30 prostate was then sectioned perpendicular to the long axis months (6 to 84 months). All patients were followed every (apical-basal) of the gland along the posterior/rectal sur- 3 months for the first 2 years, every 6 months for the subseface a t 5 mm. intervals, with most specimens requiring 4 to quent 3 years and annually thereafter. No patient received 7 cross sections to be entirely sectioned. For each cross neoadjuvant hormonal therapy preoperatively or adjuvant section, a single section each of the right and left posterior hormonal or radiation therapy postoperatively and prior to regions was submitted. Finally, a single section from the PSA failure. All patients were censored at the time of biomid anterior prostate was submitted for microscopic evaluation. Evidence of extraprostatic disease including semi- chemical failure. No patient was lost to followup. nal vesicle invasion, extracapsular extension, and/or positive surgical margins was recorded. Disease extending into but not through the prostatic capsule was considered negRESULTS ative for extracapsular extension. Patient characteristics. Table 1lists the clinical and pathoStatistical methods. A Cox regression multivariate analylogical characteristics and 3-year PSA failure-free survival sisZoexamining the predictive value of the PSA (equal to or less than 4 ngJml., greater than 4 and equal to or less than for the 688 surgically managed prostate cancer patients. 10 ng./ml., greater than 10 and equal to or less than 20 Pathologic organ-confined disease was noted in 69% (477/ ng./ml. and greater than 20 ng./ml.), calculated prostate can- 688) of all clinically organ-confined patients. Rates of extracer volume (less than 0.5 equal to or greater than 0.5 prostatic disease and positive surgical margins were noted to increase with increasing PSA, biopsy Gleason score and calequal to or less than 4.0 ~ mand . greater ~ than 4.0 em.?, biopsy Gleason score (2-4,5-7,8-10), and the clinical stage culated tumor volume. Time to PSA failure analyses. The results (p values and (Tla,b, Tlc, T2a, T2b, T2c) on the time to PSA failure performed. PSA failure was defined by 2 consecutive nonzero relative risks) of the Cox regression multivariate analysis PSAs (equal to or greater than 0.2 ng./ml., Hybritech assay) examining the predictive value of the pretreatment PSA, obtained postoperatively after a nondetectable meas- calculated prostate cancer volume, biopsy Gleason score and urement. The time of failure was defined as the time of the clinical stage on the time to PSA failure are listed in table 2 &st detectable measurement. If the PSA never became un- for the 688 study patients. The relative risk of PSA failure detectable postoperatively, then the time of PSA failure was was largest for both calculated volume of cancer categories scored as equal to 0 months. For descriptive purposes, actu- being 6.5 and 14.5 for patients with a calculated volume of arial results were displayed using Kaplan-Meier analysis.21 cancer of at least 1.5 ~ mand . ~ less than 4.0 ~ m or. greater ~ The log rank statistic was used to test for the significance of than 4.0 ~ m . respectively. ~, The serum PSA added independifferences between Kaplan-Meier curves. Step-down regres- dent predictive information regarding postoperative PSA sion methods were used to build parsimonious statistical failure-free survival with a cutoff being above or below 10 models for the association of the clinically significant prog- ng./ml. in the intermediate volume group. Using these 2 nostic factors and time to PSA failure. pretreatment clinical predictors created a staging system Calculated prostate cancer volume, In a previous report the calculated prostate cancer volume has been defined16 and is with 4 categories. The 4 stages were defined as: stage I-cal~ , II-calcurepresented by the quotient of the cancer-specific PSA and culated volume of cancer less than 0.5 ~ m .stage the PSA measured in serum per ~ mof .prostate ~ cancer of a lated volume of cancer equal to or greater than 0.5 and equal to or less than 4.0 cm3 and PSA less than 10 nglml., stage given Gleason score. The cancer-specific PSA was defined as the PSA corrected 111-calculated volume of cancer equal to or greater than 0.5 . PSA ~ equal to or greater for the PSA contributed by benign prostatic epithelial cells. and equal to or less than 4.0 ~ mand than 10 ng./ml., and stage IV-calculated volume of cancer * Travenol Laboratories, Deefield, Illinois. greater than 4.0 ~ m(fig. . ~ 1). t Hybritech, Inc., San Diego, California.

1424

STAGING CLINICALLY LOCALIZED PROSTATE CANCER

TABLE1. Clinical, pathological characteristics arid PSA failure-free survival of the 688 clinically localized prostate cancer patients in the time to PSA failure study _ _ ~ ~~~~

~~~

I r i I Pathologic

Clinical Characteristics

At Risk

( % IPatholopjc

('7 13-Yr. PSA Failure-Free Survival

for 3 Yrs.

Organ Confined

0 3 4 18 23 1 9 28 1 5 9 33 1 6

77 81 79 65 42 83 68 41 91 84 55 23 93 60

8 7 88 17 18 60 79 4 28 82 26 6 60 72

83 73 63 45 79 69 48 81 79 55 37 85 71

15 32

6

75 53

SO

6

22

77 60

45

34

36

6

34

No.

Extracapsular Extension

(7) Pas. Surgical Margins

15 162 311 80 120 155 373 60 67 378 168 75 165 423

27 24 27 36 58 21 33 57 17 23 48 63 15 32

13 15 17 24 42 18 22 30 12 14 33 45 11 21

283 140

26 44

100

66

('7)Pathologic Seminal Vesicle Invasion

~~

Clinical stage T l a b Clinical stage T l c Clinical stage TZa Clinical stage T2b

PSA less than 4 ng./nil. PSA greater than 4-10 ng./ml. PSA greater than 10-20 ng./nil. PSA greater than 20 ng./ml. Calculatcd vol. prostate Ca less than 0.5 cin.' Calculated vol. prostate Ca equal to or greater than 0.5 and less than 4.0 cm:' Calculated vol. prostate Ca PSA less than 10 ng./ml. Calculated vol. prostate Ca + PSA equal to or greater than 10 ng./ml. Calculated vol. prostate Ca equal to or greater than 4.0 cm.:' -.

-

TABLE2 . Relatioe risk and p calues from the Cox regression multivariate analysis evaluating time to PSA failure in the 688 clinical1.y localized prostate cancer patients Clinical Characteristics

D

Value

Relative Risk'

Baseline Grouu ~~

Clinical stage 0.27 TIC Biopsy Gleason 8-10 0.28 Biopsy Gleason 2-4 Biopsy Gleason S-7 0.86 Biopsy Gleason 2-4 PSA greater than 20 ngJml. <0.0001 3.87 PSA less than 4 ng./ml. 0.0009 2.04 PSA less than 4 ng./ml. PSA greater than 10 and equal to or less than 20 ng./ml. PSA less than 4 ng./ml. 0.07 PSA greater than 4 and equal to or less than 10 ng./ml. <0.0001 14.5 Calculated vol. prostate Ca less than 0.5 cm.:' Calculated vol. prostate Ca greater than 4.0 cm.:' 6.5 Calculated vol. prostate Ca less than 0.5 cm." Calculated vol. prostate Ca greater than 0.5 and equal to or less than 4.0 cm." <0.0001 * Represents the increased chance of PSA failure for a patient with the stated clinical characteristic compared to a patient with the baseline group characteristic holding all other clinical factors constant. 100In this study of 688 clinically localized patients, clinical T stage was not a statistically independent predictor of time to PSA failure when compared with the calculated volume of 80. cancer and the pretreatment PSA using a multivariate analysis. Comparing the PSA failure-free survival from the staging system derived in this study to the current AJCC staging system depicted in figure 2 further supports this observation. 60. Specifically, using the combined calculated volume of cancer and PSA staging system, a group of 165 of the 688 (24%) Yo bNED patients were defined who had a n 85% pathologic organ 40.5 confinement rate and a 90% 5-year PSA failure-free survival rate. In these stage I patients, radical prostatectomy is likely to be curative. Conversely, this staging system also identified 20 100 of 688 (15%)patients with a very low 5-year PSA failurefree survival rate. In these stage IV patients radical prostatectomy is likely not to be curative. The 2 intermediate stages had 5-year PSA failure-free survival rates of 63 and 24% 0 12 24 36 48 60 respectively. Time (months) With the possible exception of the 200 (29% of 688) clinical stage T2b and T2c patients, the current AJCC staging system FIG. 1. Biochemical failure-free survival (bNED) using proposed caldepicted in figure 2, did not adequately select patients likely culated volume of prostate cancer and PSA based system for 688 clinor unlikely to be cured after radical prostatectomy based on ically localized prostate cancer patients managed with radical prostaAll possible comparisons among stages yields p (0.0001. Stage the 5-year PSA failure-free survival results. A total of 568 tectomy. I-calculated volume of cancer less than 0.5 cm.3; stage 11-calculated (83%of the 688) patients using the current AJCC staging volume of cancer equal to or greater than 0.5 and equal to or less than system had 5-year PSA failure-free survival rates that were 4.0 emd and PSA less than 10 nglml., stage 111-calculated volume of all in the intermediate range varying from 41 to 81%.More- cancer equal to or greater than 0.5 and equal to or less than 4.0 cm.:' and PSA equal to or greater than 10 nylml., and stage N-calculated over, the p values (>0.05) between the actuarial PSA failure- volume of cancer greater than 4.0 cm: . (Numbersa t risk are in order of free survival for clinical stages Tla,b, Tlc, and T2a are not stage and are shown along time axis.) significantly different. It is noteworthy that only 3 patients (2%)of 165 in the calculated volume of prostate cancer group less than 0.5 cm." (2%))patients of 100 with calculated volume of cancer greater (stage I ) had a serum PSA of 10 ng./ml. or more (that is, 10.8, than 4.0 cm:" (stage IV) had a PSA less than 10 ng./ml. (that 10.1 and 10.5)and all 3 of these patients remain without PSA is, 7.8,9.4) and both of these patients failed biochemically at1 failure at 31, 43 and 53 months, respectively. Similarly, 2 and 18 months, respectively. These cases support the relative

STAGING CLINICALLY LOCALIZED PROSTATE CANCER

1425

the presence or absence of benign prostatic hypertrophy are all considered in defining this calculated volume of prostate cancer16 for a given patient. I---..

STAGE T1 a.b

CONCLUSIONS 60.

O h bNED 40..

20.

I

0

12

24

36

48

60

Time (months)

FIG. 2. Biochemical failure-free survival (bNED) using current AJCC staging system for 688 clinically localized prostate cancer patients managed with radical prostatectomy. p Values for comparisons among stages Tla,b, T l c and T2a are not significant (p >0.05). p Values for possible comparisons among stages T2a, T2b and T2c are S0.013. (Numbers at risk are shown along time axis in order of increasing T stage from top to bottom.)

risk data in table 2. In particular, the calculated volume of prostate cancer was more reliable than the PSA in predicting the posttreatment outcome in these cases. DISCUSSION

A Cox regression analysis performed in this study confirmed that a calculated volume of cancer less than 0.5 cm.3 allowed for the determination of stage I patients whose biochemical failure-free survival at 5 years after radical prostatectomy was 90%. The pathologic organ confinement rate from this surgical series in this subgroup was also high at 859. Therefore, this staging system unlike the current AJCC system for clinically localized patients provided a highly accurate method for predicting those patients who are optimal candidates for radical prostatectomy. Likewise, this staging system also identified stage IV patients who were at high risk for extraprostatic disease and subsequent PSA failure within 5 years after local therapy. In particular patients with a calculated prostate cancer volume greater than 4.0 ~ m had only a 34% pathologic organ confinement rate and a very low 5-year PSA failure-free survival rate. Stages I1 and I11 patients had 5-year PSA failure-free survival rates of 63 and 248,respectively. Clearly improved control rates are needed in stages I11 and Iv patients. These are the ideal groups in which phase I1 studies investigating the efficacy of adding investigational systemic agents (for example, chemo-hormonal therapy) to definitive local therapy could be employed. Stage I patients are best served by radical local therapy alone. Stage I1 patients may represent those who may benefit the addition of a traditional systemic therapy (that is, androgen ablation) to definitive local therapy. This could be best assessed in the setting of a phase I11 trial. Both the high degree of correlation of the calculated prostate cancer volume with the actual prostate cancer volume as Previously noted,'" and the significant predictive power of this parameter on the time to PSA failure noted in the current study suggest its utility as a clinical tool for preoperative staging. A possible reason why the calculated volume of prostate cancer provides the optimal stratification of PSA failurefree survival PostoperatiVelY may be because it incorporates all of the previously recognized parameters that are known to be correlated with PSA failure-free survival. Namely, the PSA, biopsy Gleason score, number of positive biopsies, and

A staging system for clinically localized prostate cancer is presented which is based on the PSA and the calculated volume of prostate cancer as follows: stage I-calculated volume of cancer less than 0.5 cm:', stage 11-calculated volume of cancer equal to or greater than 0.5 and equal to or less than 4.0cm.3 and PSA less than 10 ng./ml., stage IIIcalculated volume of cancer equal to or greater than 0.5 and equal to or less than 4.0 cm.3 and PSA equal to or greater than 10 ng./ml., and stage IV-calculated volume of cancer greater than 4.0 ~ mThis . ~staging system may provide clinically useful information regarding outcome after radical prostatectomy enabling the selection of a therapeutic approach for a n individual patient with clinically localized disease. Validation of this staging system is needed. REFERENCES

1. Smith, D. S. and Catalona, W. J.: The nature of prostate cancer detected through prostate specific antigen based screening. J. Urol., 152: 1732, 1994. 2. Culp, 0. S.: Radical perineal prostatectomy: its past, present, and possible future. J. Urol., 9 8 618, 1968. 3. Jewett, H. J., Eggleston, J. C. and Yawn, D. H.: Radical prostatectomy in the management of carcinoma of the prostate: probable causes of some therapeutic failures. J . Urol., 107: 1034, 1972. 4. Benson, R. C., Jr., Tomera, K. M., Zincke, H., Fleming, T. R. and Utz, D. C.: Bilateral pelvic lympadenectomy and radical retropubic prostatectomy for adenocarcinoma confined to the prostate. J. Urol., 131: 1103, 1984. 5. Blute, M. L., Nativ, O., Zincke, H., Farrow, G. M., Therneau, T. and Lieber, M. M.: Pattern of failure after radical retropubic prostatectomy for clinically and pathologically localized adenocarcinoma of the prostate: influence of tumor deoxyribonucleic acid ploidy. J. Urol., 142: 1262, 1989. 6. McNeal, J. E.: Origin and development of carcinoma in the prostate. Cancer, 2 3 24, 1969. 7. Blackwell, K. L., Bostwick, D. G . , Myers, R. P., Zincke, H. and Oesterling, J. E.: Combining prostate specific antigen with cancer and gland volume to predict more reliably pathological stage: the influence of prostate specific antigen cancer density. J. Urol., 151: 1565, 1994. . 8.~ McNeal, J. E.: Cancer volume and site of origin of adenocarcinoma in the prostate: relationship to local and distant spread. Hum. Path., 2 3 258, 1992. 9. McNeal, J. E., Bostwick, D. G., Kindrachuk, R. A.. Redwine. E. A,, Freiha, F. S. and Stamey, T. A.: Patterns of progression in prostate cancer. Lancet, 1: 60, 1986. 10. McNeal, J. E., Villers. A. A., Redwine, E. A., Freiha, F. S. and Stamey. T. A,: Capsular penetration in prostate cancer. Significance for natural history and treatment. Amer. J. Surg. Path., 1 4 240, 1990. 11. McNeal. J. E., Villers. A. A., Redwine, E. A.. Freiha, F. S. and Stamey, T. A.: Histologic differentiation. cancer volume, and pelvic lymph node metastasis in adenocarcinoma of the prostate. Cancer, 66.1225, 1990. 12. Stamey, T. A.. McNeal, J. E., Freiha, F. S. and Redwine, E.: Morphometric and clinical studies on 68 consecutive radical prostatectomies. J. Urol.. 139 1235, 1988. 13. Villers. A. A.. McNeal. J. E., Redwine, E. A., Freiha. F. S. and Stamey, T. A.: Pathogenesis and biological significance of seminal vesicle invasion in prostatic adenocarcinoma. J . Urol.. 143 1183, 1990. 14. Stamey, T. A.. Yang. N.. Hay, A. R.. McNeal. J. E.. Freiha. F. S. and Redwine. E.: Prostate-specific antigen a s a serum marker for adenocarcinoma of the prostate. N. Engl. J. Med., 317: 909, 1987. 15. Noldus, J. and Stamey, T. A.: Limitations of serum prostate specific antigen in predicting peripheral and transition zone cancer volumes as measured by correlation coefficients. J. Urol., 155 232, 1996.

1426

STAGING CLINICALLY LOCALIZED PROSTATE CANCER

21. &plan, E. L. and Meier, P.: Non-parametric estimation from incomplete observations. J. Amer. Stat. Assoc.. 53: 457, Desjarden, A,, Chen, M., Loughlin, K. and Richie, J.: Calcu1958. lated prostate cancer volume: the optimal predictor of the 22. k p o r , H., Wang, B. and Shapiro. E.: Relationship between prosactual prostate cancer volume. Urology. 4 9 385, 1997. tatic epithelial volume and serum prostate-specific antigen 17. Gleason. D. F.: Histologic grading and staging of prostatic carlevels. Urology, 44: 199, 1994. cinoma. In: Urologic Pathology: The Prostate. Edited by M. 23. Marks, L. S., Treiger, B., Dorey, F. J., Fu, Y. S.and deKernion, Tannenbaum. Philadelphia: Lea & Febiger. pp. 171-187,1977. J. B.: Morphometry of the prostate. I. Distribution of tissue 18. Myrtle. J. F.. Klimley. P. G., Ivor, L. P. and Brun, J. F.: Clinical components in hyperplastic glands. Urology, 44: 486, utility of prostate-specific antigen (PSA, in the management of 1994. prostate cancer. Adv. Cancer Diag. Hybritech Inc., 1986. 19. American Joint Committee on Cancer: Manual for Staging Can- 24. Aihara, M., Lebovitz, R. M., Wheeler, T. M., Kinner. B. M., Ohori, M. and Scardino, P. T.:Prostate specific antigen and cer, 4th ed. Philadelphia: J. P. Lippincott Co., 1992. Gleason grade: a n immunohistochemical study of prostate 20. Cox, D. R.: Regression models and life-tables. J. Roy. Stat. SOC. cancer. J . Urol., 151: 1558, 1994. B.. 34: 187, 1972. 16. DAmico, A. V., Chang, H., Holoupka, E.. Renshaw, A.,