ADULT UROLOGY
`CME ARTICLE
`
`ELSEVIER
`
`PROSTATE-SPECIFIC ANTIGEN VERSUS PROSTATE-SPECIFIC
`ANTIGEN DENSITY AS PREDICTOR OF TUMOR VOLUME,
`MARGIN STATUS, PATHOLOGIC STAGE, AND
`BIOCHEMICAL RECURRENCE OF PROSTATE CANCER
`
`STEPHEN A. BRASSELL, TZU-CHEG KAO, LEON SUN, AND JUDD W. MOUL
`
`ABSTRACT
`Objectives. To compare prostate-specific antigen (PSA) and PSA density (PSAD) calculated by transrectal
`ultrasound (TRUS) volume (TRUS PSAD), patqologic volume (Path PSAD), and weight (Weight PSAD) for their
`ability to predict pathologic characteristics and biochemical recurrence of prostate cancer. We also com(cid:173)
`pared all PSAD derivatives to determine consistency.
`Methods. Between 1993 and 2002, 306 patients were retrospectively identified who had had PSAD
`determined preoperatively by TRUS and sub$equently underwent radical prostatectomy with whole mount(cid:173)
`ing and close step sectioning. The determination of stage, margin status, tumor number, individual tumor
`volume, and total tumor volume was obtained from the pathologic evaluation. Clinical follow-up was available for
`265 patients.
`\
`Re~ults. The mean patient age was 62 years, the median Gleason score was 7, the median PSA level was
`5.80 ng/mL, and the median TRUS PSAD was 0. 16. The percentages of concordance for PSA, TRUS PSAD,
`Path PSAD, and Weight PSAD were similar in predicting margin status and extracapsular exte.nsion. Using
`linear regression analysis, PSA was more efficacious than TRUS PSAD, Path PSAD, or Weight PSAD in
`predicting the total tumor volume (R2 0.11, 0.08, 0.04, and 0.06, respectively). A significant positive
`correlation was found among TRUS PSAD, Path PSAD, and Weight PSAD. PSA was significantly better in
`predicting biochemical recurrence than TRUS, Path, or Weight PSAD (concordance 75.5%, 66.6%, 66.5%,
`and 70.4%, respectively).
`Conclusions. PSA and TRUS P~AD are significant and equivalent predictors of margin status and extracap(cid:173)
`sular extension. A marked difference may exist between PSA and TRUS PSAD in predicting the total tumor
`volume and biochemical recurrence. UROLOGY 66: 1229-1233, 2005. © 2005 Elsevier Inc.
`'·
`
`C ontroversy exists concerning the utility of serum
`
`prostate-specific antigen (PSA) as a screening
`tool for prostate cancer, as well as its prognostic value
`in determining tumor burden and posttreatment bio(cid:173)
`chemical recurrence and survival. Recent data sup-
`
`The opinions and assertions contained herein are the private
`views of the authors and are not to be construed as reflective of the
`views of the U.S. Army or the Department of Defense.
`From the Center for Prostate Disease Research, Department of
`Surgery and Division of Epidemiology and Biostatistics, Depart(cid:173)
`ment of Preventive Medicine and Biometrics, Uniformed Services
`University of the Health Sciences, Bethesda, Maryland; and Urol(cid:173)
`ogy Service, Department of Surgery, Walter Reed Army Medical
`Center, Washington, DC
`Reprint requests: Stephen A. Brassell, M.D., 1 Hermann Museum
`Circle, Apt. 2084, Houston, IX 77004. E-mail: stephen.brassell@
`us.army.mil
`Submitted: january 10, 2005, accepted (with revisions): june
`22,2005
`
`port the idea that benign prostatic hyperplasia is the
`major contributor to serum PSA values between 2
`and 10 ng/mL. Moreover, tumor volume and bio(cid:173)
`chemical recurrence might not be predicted by a pre(cid:173)
`treatment PSA level within this range. 1 With the ad(cid:173)
`vent of more rigorous screening efforts and ensuing
`stage migration, most cancers detected presently fall
`within these relatively low PSA values. We are now
`challenged to search for a more reliable prognostic
`marker to assess these tumors accurately and assist in
`preoperative planning and subsequent follow-up.
`The original research concerning PSA density
`(PSAD) by Benson2 focused on its utility in im(cid:173)
`proving the sensitivity and specificity of PSA in
`prostate cancer screening. Less investigation has
`been done into evaluating its role as a predictor of
`tumor characteristics. It has been predicted that,
`on average, 1 g of benign prostatic hyperplasia tis-
`
`© 2005 ELSEVIER INC.
`ALL RIGHTS RESERVED
`
`0090-4 2 9 5/05/$30.00
`doi:l0.1016/j.urology.2005.06.106 1229
`
`WCK1018
`Page 1
`
`

`
`sue increases the serum PSA concentration by
`about 0.3 ng/mL. Furthermore, 1 g of prostate can(cid:173)
`cer tissue increases the serum PSA level by about
`3.5 ng/mL. 3 Thus, the hypothesis could be made
`that the PSAD would be a more accurate marker of
`tumor volume, extracapsular extension, and even(cid:173)
`tual PSA recurrence. Several groups have con(cid:173)
`ducted preliminary investigations into this topic
`with small data sets. One such study directly com(cid:173)
`pared PSA versus PSAD in predicting regional lymph
`node involvement and found that PSAD had a 30%
`greater sensitivity than PSA alone using a value of
`0.15 nglmUcm3 and 10 ng/mL, respectively.4 It has
`also been demonstrated that for a PSAD ofless than
`0.15 nglmUcm2
`, favorable pathologic features (organ(cid:173)
`confined, Gleason score less than 7, and tumor vol(cid:173)
`ume less than 10%) can be predicted with a sensi(cid:173)
`tivity of 74%. 5 More recently, "it has been shown
`that PSAD is a strong predictor of biochemical fail(cid:173)
`ure after prostatectomy. 6 Our goal was to deter(cid:173)
`mine whether PSAD was a more accurate predictor
`of tumor volume, margin status, pathologic stage,
`and biochemical recurrence than PSA using a co(cid:173)
`hort of 1nen enrolled in a large outcomes study and
`who had had unique close-step sectioned patho(cid:173)
`·
`logic assessment.
`
`MATERIAL AND METHODS
`
`A retrospective analysis between 1993 and 2002 revealed
`306 patients treated at 1 of 10 military medical centers in(cid:173)
`cluded in the Center for Prostate Disease Research database
`who had had preoperative PSAD measured by transrectal ul(cid:173)
`trasonography (TRUS) and subsequently proceeded to radical
`prostatectomy with whole mounting at the Armed Forces In(cid:173)
`stitute of Pathology, as previously described.7·8 Clinical data
`were obtained, including clinical stage, biopsy Gleason score,
`race, age, TRUS volume, PSA, and PSAD as determined by
`TRUS volume (TRUS PSAD). Pathologic data were reviewed
`to document prostate size, tumor number, individual and total
`tumor volume, pathologic stage, Gleason score, and margin
`status. The pathologic PSAD (Path PSAD) was then calculated
`using the preoperative PSA level and the prostate volume as
`determined on gross pathologic examination using the solid
`ellipse formula: 0.52 X (length X width X height). PSAD was
`also calculated using the pathologic weight of the prostate
`(Weight PSAD). Clinical follow-up was available for 265 pa(cid:173)
`tients. In the 41 patients for whom a postoperative PSA level
`was not available, 26 were less than 1 year from surgery and 15
`were lost to follow-up. Evidence of biochemical recurrence,
`defined as a solitary PSA value greater than 0.2 nglmL, and
`clinical disease-free survival were noted. All PSA values were
`obtained using the Elecsys· E170/2010 (Roche/Boehringer
`Mannheim, Indianapolis, Ind) "ultrasensitive" PSA assay.
`
`STATISTICAL ANALYSIS
`The association among TRUS PSAD, Path PSAD, and
`Weight PSAD was evaluated using Spearman's correlation.
`Risk factors such as pathologic Gleason score, age, and the
`PSA-related variables (PSA, TRUS PSAD, Path PSAD, and
`Weight PSAD) were treated as continuous variables. For a
`binary outcome, a forward selection procedure was used to
`build the adjusted logistic regression models by including
`
`race, pathologic Gleason score, age, and one PSA-related vari(cid:173)
`able with an entry significance level of 0.15. The area under
`the corresponding receiver operating characteristic curve
`(AUC) was used to compare the logistic regression models for
`a given outcome. Also, to determine which predictor was bet(cid:173)
`ter among the logistic regression models for a given outcome,
`the percentage of concordance (C) was used. To predict the
`time to bioche~ical recurrence after radical prostatectomy,
`Cox proportional hazards models were used. Risks were as(cid:173)
`sessed in hazard ratios. A forward selection procedure with
`similar criteria was applied. To predict the total tumor vol(cid:173)
`ume, R2 of linear regression analysis was used. The 95% con(cid:173)
`fidence interval was used with the corresponding estimate.
`All the models were done in two ways: univariate (or unad(cid:173)
`justed) analysis using only one independent variable in the
`model, and multivariate (or adjusted) analysis using more
`than one independent variable in the model. C, AUC, and R2
`were used in comparing the 1993 to 1997 and 1998 to 2002
`data sets. A statistical software program, Statistical Analysis
`Systems, version 8.2, was used for computations. The signifi(cid:173)
`cance level for a statistical test was set at 5%.
`
`RESULTS
`
`The dervographics of our study group consisted
`of a mean\age of 62 years, with a racial distribution
`of 69% white and 26% black. The clinical parame(cid:173)
`ters consisted of a vean PSA level of 7.34 ng/mL,
`mean TRUS volume'of 40 cm3
`, pathologic volume
`of 32 cm3
`, and median TRUS PSAD, Path PSAD,
`and Wbght PSAD of 0.16, 0.21, and 0.14, respec(cid:173)
`tively.
`First, we examiifled the association among TRUS
`PSAD, Path PSAD, and Weight PSAD. A strong cor(cid:173)
`relation was seen among all the measurements. This
`was indicated by a Spearman correlation coeffic}ent
`ofQ.792 between TRUS PSAD and Path PSAD, 0.837
`between TRUS PSAD and Weight PSAD, and 0.928
`between Path PSAD and Weight PSAD.
`PSA, TRUS PSAD, Path PSAD, and Weight PSAD
`were then analyzed individually and found to have
`a significant association in predicting margin sta(cid:173)
`tus (C = 60.3%, 58.3%, 60.8%, and 60.2%, respec(cid:173)
`tively)t Similar results were obtained in relation to
`extracapsular extension ( C = 65.1% for PSA,
`62.9% for TRUS PSAD, 62.6% for Path PSAD, and
`65.5% for Weight PSAD). Next, we determined
`which factor would be more predictive of the total
`tumor volume. Using linear regression analysis,
`PSA was somewhat superior to TRUS PSAD, Path
`PSAD, and Weight PSAD (R2 = 0.11, 0.08, 0.05,
`and 0.06, respectively).
`Finally, each risk factor was evaluated for its abil(cid:173)
`ity to predict the time to biochemical recurrence.
`With a mean follow-up of 43 months, 49 patients
`( 16%) had biochemical recurrence. When pro(cid:173)
`ceeding with multivariate analysis using forward
`selection, Gleason score, PSA level, TRUS PSAD,
`Path PSAD, and Weight PSAD were all significant
`predictors of the time to biochemical recurrence.
`To negate the impact of scale, a direct comparison
`using the percentage of concordance revealed PSA
`
`1230
`
`UROLOGY 66 (6), 2005
`
`WCK1018
`Page 2
`
`

`
`TABLE I. Percentage of concordance and areas under operating
`characteristic curve for logistic regression models during 1993 to
`2002
`1993-1997
`c (%)
`AUC
`
`1998-2002
`c (%)
`AUC
`
`1993-2002
`c (%)
`AUC
`
`Variable
`Surgical margin
`PSA
`TRUS PSAD
`Path PSAD
`Weight PSAD
`Extracapsular extension
`PSA
`TRUS PSAD
`Path PSAD
`Weight PSAD
`Biochemical recurrence
`PSA
`TRUS PSAD
`Path PSAD
`Weight PSAD
`
`62.5
`60.9
`63.6
`63.7
`
`66.3
`64.5
`63.2
`65.8
`
`75.0
`5p.6
`65.6
`67.3
`
`0.632
`0.617
`0.641
`0.642
`
`0.666
`0.653
`0.637
`0.661
`
`0.753
`0.605
`0.658
`0.673
`
`58.6
`56.3
`58.5
`57.4
`
`64.0
`61.7
`61.8
`65.3
`
`75.7
`71.1
`67.1
`73.1
`
`0.591
`0.573
`0.590
`0.580
`
`0.644
`0.626
`0.624
`0.656
`
`0.763
`0.724
`0.679
`0.736
`
`60.3
`58.3
`60.8
`60.2
`
`65.1
`62.9
`62.6
`65.5
`
`75.5
`66.6
`66.5
`70.4
`
`0.610
`0.592
`0.613
`0.607
`
`0.655
`0.638
`0.631
`0.658
`
`0.760
`0.678
`0.673
`0.709
`
`KEY: C = concordance; AUC = area under receiver operating characteristic curve; PSA = prostate-specific antigen; TRUS
`PSAD = PSA density detennined from travsrectal ultrasound volume; Path PSAD = PSAD detennined from pathologic
`volume; Weight PSAD = PSAD detenninedfrom prostate weight.
`
`as a better predictor than TRUS, Path, or ~~ight
`PSAD (C = 75.5%, 66.6%, 66.5%, and 70.4°/6, re(cid:173)
`spectively; Table I).
`Next, consideration was given to the impact of
`lower tumor volumes on the predictive ability of
`PSAD. Specifically, we wonde~ed whether there(cid:173)
`cent stage migration and downsizing of tumor bur(cid:173)
`den made PSAD more or less relevant currently
`than in the past. To address this issue, we divided
`our analysis into two year-groups: 1993 to 1997
`and 1998 to 2002. Both PSA 'and the PSAD deriva(cid:173)
`tives had a 4% to 6% greater percentage of concor(cid:173)
`dance and AUC in the 1993 to 1997 su~group than
`in the 1998 to 2002 subgroup in determining mar(cid:173)
`gin status (P = 0.0022 for C and P = @.0018 for
`AUC; Table I). Additionally, the earlier subgroup
`exhibited greater predictive capacity for extracap(cid:173)
`sular extension (1% to 3%; P = 0.0410 for C and P
`= 0.0413 for AUC). No difference between PSA
`and the PSAD derivatives was discernible in pre(cid:173)
`dicting these outcomes.
`No significant difference was noted between the
`two year-groups in predicting biochemical recur(cid:173)
`rence (P = 0.1202 for C andP = 0.1436 for AUC).
`However, the PSA level was consistently greater
`than the PSAD derivatives in both year groups for
`determining biochemical recurrence.
`To investigate this difference further, the preop(cid:173)
`erative PSA level was categorized into the follow(cid:173)
`ing subgroups: group 1, PSA less than 4; group 2,
`PSA of 4.0 to 10; and group 3, PSA greater than 10
`ng/mL. The same was done with TRUS PSAD:
`group 1, TRUS PSAD less than 0.15; group 2, TRUS
`PSAD 0.15 to 0.26; and group 3, TRUS PSAD
`
`greater than 0.26. Stratified PSA values were in(cid:173)
`creasingly predictive of the time to biochemical
`recurrence, with a hazard ratio of 8.56 and 19.91
`comparing groups 2 and 1 and groups 3 and 1,
`respectively. This was not the case with TRUS
`PSAD (hazard ratio 1. 77 and ·2.45 comparing
`groups 2 and 1 and groups 3 and 1, respectively).
`Furthermore, these individual groups were evalu(cid:173)
`ated for their ability to predict the time to biochem(cid:173)
`ical recurrence expressed in Kaplan-Meier curves
`(Fig. 1). The separation was not as apparent in the
`TRUS PSAD subgroups as it was in the PSA sub(cid:173)
`groups, with crossover occurring at the 8-year in(cid:173)
`terval for groups 2 and 3.
`
`COMMENT
`
`Few studies have focused on PSAD's prognostic
`value as a determinant of tumor burden and bio(cid:173)
`chemical recurrence. Zentner et al. 9 found that the
`biochemical disease-free survival rate of those pa(cid:173)
`tients treated with external beam radiotherapy was
`100% for those with a PSAD less than 0.3 and 62%
`for those with a PSAD greater than 0.3 at a mean
`follow-up of 13 months. A surgical series pub(cid:173)
`lished in 1994 found that patients with a PSAD less
`than 0.3 had an 80% chance of operative success
`compared with 46% for patients with a PSAD
`greater than 0.3. 10
`Additional analysis by Zlotta et al. 11 determined
`that PSAD of the transition zone was the most sig(cid:173)
`nificant predictor of extracapsular disease by mul(cid:173)
`tivariate and receiver operating characteristic anal-
`
`UROLOGY 66 (6), 2005
`
`1231
`
`WCK1018
`Page 3
`
`

`
`1 .1 . . . . - - - - - - - - - - - - - - - - - - - - - ,
`
`.S:l 1.0 ~ ,..----+--tt+-+IHH--1+11-H---IHIIHHI--1-~
`~
`~
`·~
`:::J
`CJl
`Q)
`Q)
`
`.9
`
`PSA<4
`
`PSA4·10
`
`~ .8
`()
`c:
`~ a .7
`
`~
`<(
`(J)
`0..
`
`.6
`
`PSA>IO
`
`. 5+ - - - , . . - - - - - - . - - - - - , - - - - , - - - - , . - - - - - - l
`40
`20
`100
`120
`80
`60
`
`Months post surgery
`
`A
`
`.S:l
`~
`~ 1.0
`·~
`:::J
`CJl
`~ .9
`d;
`()
`c:
`~ .8
`5
`~
`0
`~
`0..
`~ .6
`0:::
`1-
`
`.7
`
`B
`
`20
`
`40
`
`/
`
`60
`
`80
`
`100
`
`120
`
`Months post surgery
`
`(A} Kaplan-Meier curve for PSA subgroup
`FIGURE 1.
`analysis. P = 0.0001 (log-rank test}. (B) Kaplan-Meier
`curve for TRUS PSAD subgroup analysis. P = 0.0390
`(log-rank test}.
`
`ysis, superseding PSA and Gleason score. A more
`recent study by Freedland et al. 6 indicated that
`PSAD, when using the weight of the prostate as
`measured on pathologic examination as a surro(cid:173)
`gate for volu1ne, was the strongest predictor of ex(cid:173)
`tracapsular extension. They also showed that only
`PSAD and Gleason score were predictors of bio(cid:173)
`chemical recurrence. Furthermore, PSAD was the
`only clinical variable that was a significant inde(cid:173)
`pendent predictor of margin status, extracapsular
`extension, and seminal vesicle involvement, and
`PSA alone was not an independent predictor of
`these pathologic parameters in multivariate analy(cid:173)
`sis.6 The same group then went on to compare PSAD
`determined by TRUS against PSA in predicting bio(cid:173)
`chemical recurrence, finding a slight benefit for
`PSAD in determining biochemical recurrence. 12
`Our data reflect different results, indicating that
`PSA has a greater likelihood of predicting recur-
`
`1232
`
`renee and tumor volume than the PSAD deriva(cid:173)
`tives. Additional advantages to using PSA are the
`ease of acquisition, universal use, and the ability to
`obtain it preoperatively. This is not always the case
`with PSAD.
`Some may argue that the use of TRUS PSAD has
`inherent inaccuracy owing to the reliance on volu(cid:173)
`metric measurements in its determination, which
`may bias its comparison with PSA. Kimura et al. 13
`evaluated ellipse volumetric measurements, the
`mechanism by which TRUS determines the vol(cid:173)
`ume, and multislice planimetric volume calcula(cid:173)
`tions. The error was only 5% to 10% in compara(cid:173)
`tive measurements. We set out to confirm the
`reliability of TRUS PSAD by comparing all known
`modalities of calculating PSAD, thereby substanti(cid:173)
`ating its consistency. Our data confirm their work,
`with significant correlation between the TRUS
`measurements and the pathologic volume and
`weight measurem~nts. It is our belief that it is the
`inherent value of th' PSA test, not human error in
`determining PSAD, \that led to this difference. In
`addition, controversy exists concerning making
`the determination of bioche~ical recurrence on
`the basis of a PSA value of greater than 0.2 ng/mL.
`Using this definition is more acceptable when us(cid:173)
`ing the "ultrasensitive" PSA assay, as was done in
`this study.
`It has been postulated that~prostate size alone can
`affect outcomes, because the apical dissection of
`smaller prostates is more difficult, leading to an
`increased rate of positive margins. In our analysis,
`we had 62 prvstates of less than 20 g. Of these, 23
`had positive margins (37%). For the 244 prostates
`greater than 20 g, 7 4 had positive margins (30%).
`Although interesting, statistical significance was
`not achieved (P ~ 0.594), arguing against techni(cid:173)
`cal issues being a contributory factor.
`When evaluating whether stage migration played
`a role in limiting fSAD's utility, we found that both
`PSA and the PSAD derivatives had increased pre(cid:173)
`dictive value in earlier series for margin status and
`extracapsular extension; no significant difference
`was noted for biochemical recurrence over time.
`The percentage of concordance and AUC of PSA
`remained greater than the PSAD derivatives in pre(cid:173)
`dicting biochemical recurrence for each subgroup.
`Because it is known that the total tumor volume
`in the radical prostatectomy specimen is an inde(cid:173)
`pendent predictor of tumor stage and disease pro(cid:173)
`gression, we sought to determine whether PSAD
`could be used as a surrogate for the tumor volume
`preoperatively. 14,ls Our data indicated that TRUS
`PSAD provides no additional benefit in predicting
`tumor volume than PSA alone (R2 0.085 versus
`0.11). Neither value is highly predictive of tumor
`volume; however, the purpose of this analysis was
`not to emphasize the predictive ability but to pro-
`
`UROLOGY 66 (6), 2005
`
`WCK1018
`Page 4
`
`

`
`vide a comparison of these two values. Further(cid:173)
`more, PSA and TRUS PSAD are similar in predict(cid:173)
`ing margin status and extracapsular extension by
`the percentage of concordance analysis. It appears
`that PSA is more efficacious as a predictor of bio(cid:173)
`chemical recurrence (C = 75.5% compared with
`C = 66.6% for TRUS PSAD). Also, when evaluating
`the predictive capacity of TRUS PSAD subgroups
`(PSAD less than 0.15, 0.15 to 0.26, and greater
`than 0.26), as distinctive a relationship was not
`found between the time to biochemical recurrence
`and the PSAD. This was in contrast to the PSA
`subgroup analysis (PSA less than 4, 4.0 to 10.0, and
`greater than 10 ng/mL) in which a definitive sepa(cid:173)
`ration was apparent.
`Our study had a number of limitations. First, the
`PSA and PSAD subdivisions might have been
`somewhat broad for present day comparisons. Sec(cid:173)
`ond, even though we had the strength df whole
`mounting and close step sectioning at the Armed
`Forces Institute of Pathology, a larger cohort with
`longer follow-up might have provided more defin(cid:173)
`itive results. In particular, our mean follow-up of
`43 months was short, and only 16% of these pa(cid:173)
`tients had developed recurrence. Longer follow-up
`with increased outcome events may alter the 'ton(cid:173)
`clbsions about the value of PSA.
`
`CONCLUSIONS
`""
`The results of our study have indicated that
`TRUS PSAD is a reliable preoperative measure(cid:173)
`ment and has a consistent association with both
`Path PSAD and Weight PSAD. Despite similar re(cid:173)
`sults for PSA and PSAD de\ivatives in predicting
`pathologic characteristics, a marked difference may
`exist between these two variables in their;- association
`with total tumor volume and bioche~ical recur(cid:173)
`rence. PSA was the strongest predictor of biochem(cid:173)
`ical recurrence of all the variables tested. ~lthough
`not without limitations, PSA still remains a valid
`preoperative parameter from which clinical predic(cid:173)
`tions can be made. This holds true despite there(cid:173)
`cent stage migration.
`
`REFERENCES
`1. Stamey TA, Johnstone IM, McNeal JE, et al: Preopera(cid:173)
`tive serum prostate specific antigen levels between 2 and 22
`
`ng/mL correlate poorly with post-radical prostatectomy
`cancer morphology.] Urol167: 103-111, 2002.
`2. Benson MC: Prostate specific antigen density: a means
`of distinguishing benign prostate hypertrophy and prostate
`cancer.] Urol147: 815-816, 1992.
`3. Stamey TA, Yang N, Hay A, et al: Prostate-specific anti(cid:173)
`gen as a serum marker for adenocarcinoma of the prostate.
`N Engl] Med 317: 909-916, 1987.
`4. Wolff]M, Boeckmann W, Effertt P], et al: Evaluation of
`patients with diseases of the prostate using prostate-specific
`antigen density. Br] Ural 76: 41-46, 1995.
`5. Catalona WJ, Southwick PC, Slawin KM, et al: Compar(cid:173)
`ison of percent free PSA, PSA density, and age-specific PSA
`cutoffs for prostate cancer detection and staging. Urology 56:
`255-260, 2000.
`6. Freedland Sj, Wieder ]A, jack GS, et al: Improved risk
`stratification for biochemical recurrence after radical prosta(cid:173)
`tectomy using a novel risk group system based on prostate
`specific antigen density and biopsy Gleason score.] Urol168:
`110-115, 2002.
`7. Maul JW, Wu H, Sun L, et al: Epidemiology of radical
`prostatectomy for localized prostate cancer in the era of pros(cid:173)
`tate specific antigen: an overview of the Department of De(cid:173)
`fense Center for Prostate Disease Research national database.
`Surgery 132: 213-219, 2002.
`8. Desai A, Wu H, Sun L, et al: Complete embedding and
`close step sectioning of radical prostatectomy specimens both
`increase detection of extra-prostatic extension, and correlate
`with increased disease-free survival by stage of prostate cancer
`patients. Prostate Cancer Prostate Dis 5: 212-218, 2002.
`9. Zentner PG, Pao LK, Benson MC, et al: Prostate specific
`antigen density: a new prognostic indicator for prostate cancer.
`Int] Radiat Oneal Biol Phys 27: 47-58, 1993:
`10. Benson MC, and Olsson CA: Prostate specific antigen
`and prostate specific antigen density. Cancer 74: 1667-1673,
`1994.
`11. Zlotta AR, Djavan B, Petein M, et al: Prostate specific
`antigen density of the transition zone for predicting patho(cid:173)
`logic stage of localized prostate cancer in patients with serum
`prostate specific antigen less than 10 ng!ml.j Urol160: 2089-
`2095, 1998.
`12. Freedland S], Kane C], Presti]C, et al: Comparison of
`preoperative prostate specific antigen density and prostate
`specific antigen for predicting recurrence after radical prosta(cid:173)
`tectomy: results from the SEARCH data base.] Urol169: 969-
`973, 2003.
`13. Kimura A, Nakamura S, Niizuma M, et al: Quantitative
`analysis of ultrasonogram of the prostate. ] Clin Ultrasound 14:
`501-507, 1986.
`14. McNeal]E, Bostwick DG, Kindrachuk RA, et al: Patterns
`of progression in prostate cancer. Lancet 1: 60-63, 1986.
`15. D'Amico AV, Chang H, Holupka E, et al: Calculated
`prostate cancer volume: the optimal predictor of actual cancer
`volume and pathologic stage. Urology 49: 385-391, 1997.
`
`UROLOGY 66 (6), 2005
`
`1233
`
`WCK1018
`Page 5