Sei sulla pagina 1di 6

ADULT UROLOGY

CME ARTICLE

PSA, PSA DENSITY, PSA DENSITY OF TRANSITION ZONE,


FREE/TOTAL PSA RATIO, AND PSA VELOCITY FOR EARLY
DETECTION OF PROSTATE CANCER IN MEN WITH
SERUM PSA 2.5 TO 4.0 ng/mL
BOB DJAVAN, ALEXANDRE ZLOTTA, CHRISTIAN KRATZIK, MESUT REMZI, CHRISTIAN SEITZ,
CLAUDE C. SCHULMAN, AND MICHAEL MARBERGER

ABSTRACT
Objectives. To enhance the specificity of prostate cancer (PCa) detection and reduce unnecessary biopsies
in men with prostate-specific antigen (PSA) levels of 2.5 to 4.0 ng/mL, we prospectively evaluated various
PSA-based diagnostic parameters.
Methods. This study included 273 consecutive men with serum PSA of 2.5 to 4.0 ng/mL referred for early
PCa detection or lower urinary tract symptoms. All men underwent prostate ultrasound and sextant biopsy
with two additional transition zone (TZ) biopsies. If the first biopsies were negative, repeated biopsies were
performed at 6 weeks. Total PSA, PSA density (PSAD), PSA density of the transition zone (PSA-TZ), free/total
PSA ratio (f/t PSA), and PSA velocity (PSAV) were determined, and the sensitivity, specificity, and predictive
values of these various parameters were calculated.
Results. Of 273 patients, 207 had histologically confirmed benign prostatic hyperplasia (BPH) and 66 had
PCa. f/t PSA and PSA-TZ were the most powerful predictors of PCa, followed by PSA, PSAD, and PSAV. Areas
under the receiver operating characteristic curves for f/t PSA and PSA-TZ were 74.9% and 70.1%, respec-
tively. With a 95% sensitivity for PCa detection, an f/t PSA cutoff of 41% and a PSA-TZ cutoff of 0.095 would
result in the lowest number of unnecessary biopsies (29.3% and 17.2% specificity for f/t PSA and PSA-TZ,
respectively) compared with all other PSA-related parameters evaluated.
Conclusions. Compared with standard total PSA assays, f/t PSA and PSA-TZ significantly enhance the
sensitivity and specificity of PCa detection in a referral patient population with a total PSA of 2.5 to 4.0
ng/mL. UROLOGY 54: 517–522, 1999. © 1999, Elsevier Science Inc.

A lthough standard assays of total serum pros-


tate-specific antigen (PSA) have revolution-
ized the monitoring of disease progression and
PCa. More than 20% of men diagnosed with PCa
have PSA levels lower than 4.0 ng/mL. Further-
more, in men whose total PSA falls between 2.5
management of patients with prostate cancer and 4 ng/mL, PCa can be detected in 13% to 20%
(PCa), these measurements have proved insuffi- within 3 to 5 years. Early detection in these pa-
ciently sensitive and specific for early detection or tients could be expected to result in a higher prob-
staging of PCa. Thus, PSA values less than 10 ability of successful and potentially curative treat-
ng/mL are of limited utility in discriminating be- ment. Unfortunately, more intensive diagnostic
tween benign prostatic hyperplasia (BPH) and evaluation of this group based on standard total
PSA assay values of 2.5 to 4 ng/mL would undoubt-
This work was supported by a grant from the Austrian Cancer edly increase the frequency of unnecessary biop-
Society (OKH521). sies. Novel PSA-based diagnostic parameters could
From the Department of Urology, University Hospital of Vi-
enna, Vienna, Austria; and Department of Urology, Erasme Hos- potentially aid in early PCa detection with fewer
pital, University Clinics of Brussels, Brussels, Belgium unnecessary biopsies.1–5
Reprint requests: Bob Djavan, M.D., Ph.D., Department of Little is known about the potential utility of var-
Urology, University of Vienna, Währinger Gürtel 18-20, 1090 ious PSA-based parameters among patients with
Vienna, Austria
Submitted: January 28, 1999, accepted (with revisions): March PSA levels in the 2.5 to 4.0 ng/mL range. In this
22, 1999 prospective investigation of men with this PSA

© 1999, ELSEVIER SCIENCE INC. 0090-4295/99/$20.00


ALL RIGHTS RESERVED PII S0090-4295(99)00153-3 517
TABLE I. Age, total prostate and transition zone volumes,* serum PSA, percent free PSA, PSA
density of total prostate and transition zone, and PSA velocity in patients with benign histologic
findings and prostate cancer
Benign Disease Prostate Cancer
(n ⴝ 207) (n ⴝ 66) P Value
Age (yr) 67.2 (8.8) 66.7 (7.6) 0.6566
Total prostate volume (cc) 35.0 (10.6) 33.1 (14.9) 0.04
Transition zone volume (cc) 20.16 (8.91) 13.03 (8.1) ⬍0.0001
Serum PSA (ng/mL) 3.05 (0.49) 3.15 (0.41) 0.17
PSA velocity (ng/mL/yr) 0.38 (0.278) 0.527 (0.39) 0.0091
PSAD (ng/mL/cc) 0.0864 (0.039) 0.101 (0.04) 0.0074
f/t PSA (%) 37.38 (17.94) 19.59 (12.79) ⬍0.0001
PSA-TZ (ng/mL/cc) 0.159 (0.079) 0.277 (0.172) ⬍0.0001
KEY: PSA ⫽ prostate-specific antigen; PSAD ⫽ PSA density of total prostate; f/t PSA ⫽ free/total PSA ratio; PSA-TZ ⫽ PSA density of the transition zone.
Data presented as mean with SD in parentheses.
* Measured by transrectal ultrasound.

range, the predictive value of free/total PSA ratio (f/t measuring the tri-axial distances at their maximal diame-
PSA), PSA density (PSAD), PSA density of the transi- ter.4
PSAD and PSA-TZ, expressed in nanograms per milliliter per
tion zone (PSA-TZ), and PSA velocity (PSAV) was cubic centimeter, were calculated as the serum PSA divided
assessed. by the total prostate volume or the transition zone volume,
respectively.
MATERIAL AND METHODS PSA velocity (in nanograms per milliliter per year) was deter-
mined in patients who before study entry had had at least
PATIENTS three PSA measurements at 12-month intervals, all using
From January 1997 to October 1998, 273 men, 40 to 78 the same assay.
years old, with a serum PSA of 2.5 to 4 ng/mL were enrolled at Abnormal DRE findings were defined as any DRE judged sus-
the Departments of Urology of the University of Vienna, Aus- picious for cancer by the examining urologist.
tria and of Erasme Hospital, University Clinics of Brussels, Histopathologic findings on the biopsy specimen were classi-
Belgium. The study population consisted of men referred for fied as negative (normal or BPH) or positive (cancer) using
either early cancer detection or for lower urinary tract symp- the Gleason system.
toms.
Patients were excluded if they presented with a previous
history of PCa, acute or chronic prostatis, histologic diagnosis
of prostatic intraepithelial neoplasia of any grade, urinary re-
STATISTICAL ANALYSIS
The significance of between-group differences was deter-
tention, use of an indwelling catheter, or proven urinary tract
mined by Student’s t test for grouped data or the Mann-Whit-
infection.
ney U test. The ability of PSA-based parameters to serve as
significant predictors of PCa was evaluated by univariate and
STUDY DESIGN multivariate logistic regression analysis. Receiver operating
Transrectal ultrasound with systematic sextant needle biop- characteristic (ROC) curves were constructed by plotting sen-
sies and two additional transition zone biopsies were per- sitivity versus the false-positive rate. Between-test differences
formed in all patients. Biopsies were also taken from suspi- in areas under ROC curves were assessed by the Mann-Whit-
cious areas at ultrasound or on digital rectal examination ney U test modified with the Bonferroni-Holm adjustment. P
(DRE). In all cases, if the first biopsies were negative (benign), values less than 0.05 were considered statistically significant.
within 6 weeks an additional set of sextant biopsies (and two Data were analyzed using SAS for Windows, version 6.12 (SAS
transition zone biopsies) was routinely performed to mini- Institute, Cary, NC), and JMP for Windows, version 3 (SAS
mize sampling errors. Before the first biopsy, the following Institute).
data were recorded:
The level of total PSA in the samples was determined in dupli-
cate using the equimolar AxSYM PSA assay (Abbott Labo- RESULTS
ratories, Abbott Park, Ill), according to the recommended
manufacturer protocol. The level of free PSA was deter- Two hundred seven patients had benign histo-
mined using the AxSYM Free PSA assay (Abbott Laborato- logic findings, and 66 had diagnoses of PCa. Over-
ries). The mean intra-assay coefficient of variation with this
method in our study was 1.9%, and the mean interassay all, 23% of patients with an abnormal DRE had
coefficient of variation was 2.9%. Correlation coefficients PCa. Of those diagnosed with PCa, 59% had an
between this commercial assay and other currently available abnormal and 41% a normal DRE. Table I shows
assays (Tandem E, Hybritech) have been demonstrated to the mean and standard deviation of age, serum
be at 0.99. PSA, total and transition zone volumes, f/t PSA,
Transrectal ultrasound scans were obtained using a biplanar
7.5-MHz probe (Bruel & Kjaer, Denmark or Siemens, Ger- PSA velocity, PSAD, and PSA-TZ in each group.
many). Length, height, and width of the total prostate and of Statistically significant differences included larger
the transition zone were measured as previously described, total and transition zone volumes and higher f/t

518 UROLOGY 54 (3), 1999


TABLE II. Sensitivity, specificity, and predictive values of f/t PSA
and PSA-TZ at varying cutoff values (n ⴝ 273)
Cutoff Value Sensitivity (%) Specificity (%) PPV NPV
f/t PSA (%)
10 33.333 94.244 0.647 0.815
15 46.921 89.928 0.596 0.841
20 55.948 80.575 0.486 0.852
30 71.212 64.731 0.392 0.875
40 93.939 39.613 0.331 0.953
PSA-TZ (ng/mL/cc)
0.1 93.181 17.985 0.266 0.867
0.15 70.454 50.359 0.336 0.856
0.2 59.091 74.11 0.458 0.836
0.3 38.636 92.086 0.699 0.827
0.4 15.909 98.561 0.786 0.787
KEY: NPV ⫽ negative predictive value; PPV ⫽ positive predictive value. All other abbreviations as in Table I.

PSA in patients with benign disease (P ⫽ 0.04, P for PSA-TZ and f/t PSA for patients with an abnor-
⬍0.0001, and P ⬍0.0001, respectively). PSA ve- mal and normal DRE clearly showed that the per-
locity, PSAD, and PSA-TZ were significantly higher formance of f/t PSA and PSA-TZ depended on the
in patients with PCa (P ⫽ 0.0091, P ⫽ 0.0074, DRE status. The AUC for f/t PSA for patients with
and P ⬍0.0001, respectively). Age and serum PSA an abnormal DRE was 98.2% versus 78.1% for
were not significantly different (P ⫽ 0.6566 and those with a normal DRE. Similarly, the AUC for
P ⫽ 0.17, respectively). Age and serum PSA were PSA-TZ for patients with an abnormal and normal
not significantly different (P ⫽ 0.6566 and P ⫽ DRE was 93.3% and 85.2%, respectively.
0.17, respectively). Cutoff values corresponding The predictive value of f/t PSA and PSA-TZ was
to the values between P75 in cancer (75% of pa- assessed with respect to prostate size (Fig. 1B). f/t
tients above this cutoff value) and P25 in benign PSA as well as PSA-TZ significantly failed to retain
disease (75% of patients at or below this cutoff their predictive power among patients with a pros-
value) for f/t PSA, PSA-TZ, and PSAD were 31%, tate size less than 30 cc (n ⫽ 81; AUC 61.8% and
0.362 ng/mL/cc, and 0.12 ng/mL/cc, respectively. 53.5%, respectively) compared with whole pros-
A total of 76 patients (27.8%) had an f/t PSA greater tate sizes greater than 30 cc (n ⫽ 192; AUC 69.3%
than 41%, and a total of 46 patients (16.8%) a and 67.6%, respectively).
PSA-TZ less than 0.095 ng/mL/cc.
Table II shows the sensitivity, specificity, and COMMENT
positive and negative predictive values of f/t PSA
and PSA-TZ. With 95% sensitivity, specificity for The prevalence of detectable PCa in men with
all PSA parameters was markedly reduced. How- serum PSA levels in the 2.5 to 4 ng/mL range has
ever, at 95% sensitivity, specificity of f/t PSA was been shown to be 22%.6 In addition, approximately
significantly greater than that of all other parame- 20% of men with serum PSA between 2.5 and 4
ters (P ⬍0.0001). An f/t PSA cutoff value of 41% ng/mL will be diagnosed with PCa within 3 to 5
afforded 95% sensitivity for detecting PCa while years. Furthermore, extracapsular disease is
avoiding unnecessary biopsies in 29.3% of cases. present in 40% of patients with PCa with PSA lev-
On the basis of the ROC curves presented in Fig- els of 4 to 10 ng/mL.7 Accordingly, early detection
ure 1A, f/t PSA was a significantly better PCa pre- (at a lower PSA level) may allow successful treat-
dictor than any other test evaluated. The area un- ment at a curable stage. However, the value of early
der the curve (AUC) values derived from the ROC detection in patients with low PSA values remains
curves for f/t PSA and PSA-TZ were 74.9% and to be established. In any case, lowering the current
70.1%, respectively, significantly exceeding those 4 ng/mL cutoff would lead to a large number of
of other tests investigated. The AUC was signifi- unnecessary biopsies in patients with benign dis-
cantly larger for PSAD and PSA than for PSAV. ease and would increase related costs.6 –10 Another
Logistic regression analysis was applied to PSAD, major concern attending a lower cutoff is the in-
f/t PSA, PSA-TZ, and total PSA with stepwise entry creased detection of “clinically insignificant” can-
of variables into multivariate models, f/t PSA and cers.
PSA-TZ were found to be significant PCa predic- The present study provides evidence that f/t PSA
tors. In multivariate models, f/t PSA outperformed and PSA-TZ could have a useful role in PCa pre-
PSA-TZ as a PCa predictor (P ⫽ 0.001). The AUC diction in patients with a serum PSA between 2.5

UROLOGY 54 (3), 1999 519


FIGURE 1. ROC curves for (A) all evaluated PSA-based parameters and (B) for f/t PSA and PSA-TZ as stratified by
prostate volume.

and 4 ng/mL. Although use of these parameters values remains to be established. Overall, 66 of 273
could potentially facilitate early PCa detection, the patients were diagnosed with PCa. This rather high
value of early detection in patients with low PSA cancer rate can be explained by the fact that pa-

520 UROLOGY 54 (3), 1999


tients included in this study were referred for early consensus on the optimal measurement ap-
PCa diagnosis and/or lower urinary tract symp- proaches is still lacking.
toms and not for screening purposes. In the cur- We found that the combination of PSA-TZ and f/t
rent setting, the large proportion of patients with a PSA significantly increased PCa prediction and
suspicious DRE and the fact that the repeated bi- could further decrease the number of unnecessary
opsies were performed systematically may also biopsies. If both tests were to indicate a very low
have increased the number of PCas diagnosed. PCa probability (eg, less than 5%), it could be con-
With respective areas under the ROC curves of sidered prudent to forego biopsy for a particular
74.9% and 70.1%, f/t PSA and PSA-TZ clearly out- patient. Nevertheless, additional issues regarding
performed all other PSA-based parameters. With a the usefulness of f/t PSA in clinical practice remain
95% sensitivity for PCa detection, an f/t PSA cutoff to be resolved. Thus, the optimal ratio of f/t PSA
of 41% would result in the lowest number of un- remains to be determined, and results may be in-
necessary biopsies (specificity of 29.3%) compared fluenced by the type of assay used, the size of the
with total PSA, PSAD, PSA-TZ, and PSAV. In the prostate, or the PSA range.13,15,16
low PSA range, f/t PSA has previously been shown
to increase sensitivity.11 In contrast to earlier stud- CONCLUSIONS
ies, we did find that f/t PSA was a useful marker for
PCa prediction even when PSA was less than 4 The present prospective study supports the util-
ng/mL.12 We also found a volume-dependent dif- ity of f/t PSA and PSA-TZ for predicting PCa within
ference in the performance of f/t PSA and PSA-TZ, the PSA range 2.5 to 4 ng/mL in a referral popula-
which failed to retain their predictive power tion. Patients with an f/t PSA higher than 41% or a
among patients with a prostate size smaller than PSA-TZ of less than 0.095 ng/mL/cc could safely
30 cc.1,13 have been spared unnecessary biopsies. For small
PSA-TZ proved to be the second most potentially prostates (less than 30 cc), PSA-TZ is less effective
useful parameter in the current study. At a sensi- than in patients with larger prostates, and the use
tivity of 95%, specificity was 17.2% and the cutoff of f/t PSA alone appears preferable in such cases.
value 0.095 ng/mL/cc. The f/t PSA and PSA-TZ cut- Use of PSA-TZ in conjunction with f/t PSA may be
off values associated with the combination of the of value in patients with larger prostate volumes
highest sensitivity (63.2% and 77.3%, respectively) (greater than 30 cc).
and specificity (74.1% and 56.9%, respectively)
were 0.21 ng/mL/cc and 34%, respectively. For a REFERENCES
95% PCa detection rate, the f/t PSA and PSA-TZ 1. Catalona WJ, Smith DS, Wolfert RL, et al: Evaluation of
cutoff values in patients with serum PSA from 2.5 percentage of free serum prostate-specific antigen to improve
to 4 ng/mL were 41% and 0.095 ng/mL/cc, respec- specificity of prostate cancer screening. JAMA 274: 1214 –
tively. In this PSA range, maintaining 95% sensi- 1220, 1995.
2. Partin AW, and Carter HB: The use of prostate-specific
tivity would result in 89.9%, 89.9%, and 88.5% antigen and free/total prostate-specific antigen in the diagno-
unnecessary biopsies when using PSAD, PSAV, sis of localized prostate cancer. Urol Clin North Am 23: 531–
and PSA, respectively, compared with 70.7% and 540, 1996.
82.8% with f/t PSA and PSA-TZ, respectively. The 3. Vashi AR, Wojno KJ, Henricks W, et al: Determination
DRE status (in contrast to patient age) did have an of the “reflex range” and appropriate cutpoints for percent free
prostate-specific antigen in 413 men referred for prostatic
impact on the performance of f/t PSA and PSA-TZ, evaluation using the AxSYM system. Urology 49: 19 –26,
with enhanced predictive values (AUC 98.2% and 1997.
93.3%, respectively) in patients with a suspicious 4. Zlotta AR, Djavan B, Marberger M, et al: Prostate spe-
DRE. cific antigen density of the transition zone: a new effective
A notable limitation of PSA-TZ was that small parameter for prostate cancer prediction. J Urol 157: 1315–
1321, 1997.
prostate (and transition zone) size adversely af- 5. Djavan B, Zlotta AR, Byttebier G, et al: Prostate specific
fected the diagnostic utility of this parameter. With antigen density of the transition zone for early detection of
entire prostate sizes less than 30 cc, in which the prostate cancer. J Urol 160: 411– 419, 1998.
transition zone is not markedly enlarged and there- 6. Catalona WJ, Smith DS, and Ornstein DK: Prostate can-
fore more difficult to measure, the utility of cer detection in men with serum PSA concentrations of 2.6 to
4 ng/mL and benign prostate examinations. JAMA 277: 1452–
PSA-TZ was significantly diminished. Hence, 1455, 1997.
PSA-TZ is unlikely to be well suited for screening 7. Partin AW, Criley SR, Subong EN, et al: Standard versus
in populations with comparatively small prostates. age-specific prostate specific antigen reference ranges among
Another limitation is that PSA-TZ measurements men with clinically localized prostate cancer: a pathological
are based on transrectal ultrasonography and thus analysis. J Urol 155: 1336 –1339, 1996.
8. Smith DS, Catalona WJ, and Herschman JD: Longitudi-
subject to operator-dependent variability. Al- nal screening for prostate cancer with prostate-specific anti-
though we have reported that transition zone vol- gen. JAMA 276: 1309 –1315, 1996.
ume measurement is accurate and reproducible,14 9. Gann PH, Hennekens CH, and Stampfer MJ: A prospec-

UROLOGY 54 (3), 1999 521


tive evaluation of plasma prostate-specific antigen for detec- serum of patients with prostate carcinoma and benign prostate
tion of prostate cancer. JAMA 273: 289 –294, 1995. hyperplasia. Cancer 79: 104 –109, 1997.
10. Stenman UH, Hakama M, Knekt P, et al: Serum concen- 14. Zlotta AR, Djavan B, Roumeguère T, et al: Transition
trations of prostate specific antigen and its complex with al- zone volume on transrectal ultrasonography is more accurate
pha-1-antichymotrypsin before diagnosis of prostate cancer. and reproducible than the total prostate volume (abstract).
Lancet 344: 1594 –1598, 1994. Br J Urol 80(suppl 2): A926, 1997.
11. Vashi AR, and Oesterling JE: Percent free prostate-spe- 15. Roehrborn CG, Gregory A, McConnell JD, et al:
cific antigen: entering a new era in the detection of prostate Comparison of three assays for total serum prostate-specific
cancer. Mayo Clin Proc 72: 337–344, 1997. antigen and percentage of free prostate-specific antigen in
12. Espana F, Martinez M, Montserrat R, et al: Reference predicting prostate histology. Urology 48(suppl): 23–32,
ranges for the concentrations of total and complexed plasma 1996.
prostate-specific antigen and their ratio in patients with be- 16. Partin AW, Catalona WJ, Southwick PC, et al: Analysis
nign prostate hyperplasia. Eur Urol 32: 268 –272, 1997. of percent free prostate-specific antigen (PSA) for prostate
13. Stephan C, Lein M, Jung K, et al: The influence of pros- cancer detection: influence of total PSA, prostate volume, and
tate volume on the ratio of f/t prostate specific antigen in age. Urology 48(suppl): 55– 61, 1996.

522 UROLOGY 54 (3), 1999

Potrebbero piacerti anche