Professional Documents
Culture Documents
Artículo Italiano
Artículo Italiano
P
rostate cancer (PCa) shows an extremely heteroge- patients with PCa into risk groups.1 Unfortunately, current
neous clinical course, ranging from indolent and methods of clinically assessing PCa grade and stage are not
organ-confined to aggressive and metastatic disease. accurate, and upgrading or upstaging occurs in more than
An accurate assessment of the tumor characteristics is crucial one-third of the cases.2-4 Recently, Epstein et al proposed
so that the appropriate treatment options can be considered.1 a new grading system to provide a more accurate tumor
D’Amico et al was the first to combine the use of pre- stratification, simplifying at the same time patients’ com-
operative prostate-specific antigen (PSA) levels, clinical prehension of their disease.5
stage (cT), and biopsy Gleason score (bGS) to stratify Some authors proved that prostate-specific antigen density
(PSAD) may be useful in predicting adverse pathologic find-
ings in radical prostatectomy (RP) specimens and unfa-
Financial Disclosure: The authors declare that they have no relevant financial vorable outcomes after local treatment.6 However, when
interests.
From the Department of Urology, San Giovanni Hospital, Rome, Italy; and the De- assessing the ability of PSAD to predict upgrading and up-
partment of Urology, La Sapienza University of Rome, Italy staging, conflicting results were reported.7,8 Moreover, to
Address correspondence to: Aldo Brassetti, M.D., Department of Urology, San Giovanni our knowledge, no study has yet evaluated the accuracy of
Hospital, Via Oslo 7, Fiumicino, Rome 00054, Italy. E-mail: aldo.brassetti@
gmail.com PSAD in predicting upgrading after RP using the new
Submitted: April 10, 2017, accepted (with revisions): July 28, 2017 Epstein grading system.
© 2017 Elsevier Inc. https://doi.org/10.1016/j.urology.2017.07.071 129
All rights reserved. 0090-4295
The aim of the present study was to analyze the perfor- out using only the patients for whom the complete sets of data
mance of PSAD as a predictor of upstaging and prognos- were available. The variables considered for entry into the model
tic grade group (PGG) upgrading. included age, the number of positive cores, the percentage of core
involvement, and PSAD (PSA and PV were excluded from the
multivariable model for the risk of multicollinearity). Receiver
MATERIALS AND METHODS operating characteristic (ROC) curves were produced to evalu-
ate the area under the curve (AUC) and the diagnostic perfor-
Patient Selection mance of PSAD as a predictor of upgrading and upstaging. The
After obtaining institutional review board approval, a retrospec- diagnostic performance of PSAD was also expressed by the cal-
tive analysis of our prospectively maintained database of men who culation of the sensitivity, the specificity, the positive predic-
underwent robot-assisted laparoscopic prostatectomy (RALP) at tive value (PPV), and the negative predictive value (NPV) at
our center between January 2014 and December 2015 was carried the best cutoff value of 0.130. A P value of 5% was considered
out. We included patients with a histologically confirmed pros- as the threshold for significance. Data were presented as
tate biopsy diagnosis of PCa (at our institution) within 3 months mean ± standard deviation and median with interquartile
before surgery. Patients who underwent neoadjuvant therapy were range.
excluded from the study. Previous benign prostatic hyperplasia
surgery and treatment with 5α-reductase inhibitors at the time
of surgery were also considered exclusion criteria.
RESULTS
Prostate biopsies were conducted through a transrectal
ultrasound-guided approach. During the sampling, 12 cores were Overall, 379 patients were enrolled and their character-
taken from all the patients, regardless of the size of the gland; istics are described in Table 1. Before surgery, 38 (10%)
prostate volume (PV) was also evaluated (using the ellipsoid patients presented with HG-PCa and 1 of these patients
formula) and PSAD was further calculated. (0.2%) was suspected to have a locally advanced disease.
RALP was performed by experienced surgeons. Extended lymph On pathologic examination, 83 (22%) were diagnosed with
node dissection was carried out according to European Associa- a high-grade tumor and 109 (29%) with a locally ad-
tion of Urology (EAU) guidelines.9 vanced one.
The American Joint Committee on Cancer TNM, sixth edition After RALP, 157 (41.4%) patients were found up-
(2002), was used for pathologic staging. The Gleason score (GS)
graded and the rate of clinically relevant upgrading was
and PGG were also assigned according to the International Society
14.5% (n = 55). Of 280 patients diagnosed with PGG 1
of Urological Pathology recently described by Epstein et al.10,11
cancer on biopsy, 145 (51.7%) were upgraded on final pa-
thology; the same happened to 6 of 61 patients (9.8%) with
Data Collected
Evaluated variables included age, body mass index (calculated as a bPGG of 2 and to 6 of 30 patients with a bPGG of 3.
weight in kilogram divided by height in meter squared [kg/m2]), Downgrading occurred in 9 patients (2.4%): 7 had a bPGG
preoperative PSA, PV, PSAD, cT and pathologic stage (pT), of 2 and 2 had a bPGG of 5.
bioptic and pathologic GS and PGG, the number of positive cores, Of all the patients with a bioptic diagnosis of PGG 1
and the percentage of tumor involvement of each bioptic core. cancer, 61 (22.1%) were diagnosed with a locally ad-
We defined a PGG ≥3 tumor as a high-grade prostate cancer vanced disease; the same happened to 21 patients (34.4%)
(HG-PCa); pT ≥ 3a malignancies were considered locally ad- with a bPGG of 2, to 20 patients (66.6%) with a bPGG
vanced diseases. Any increase in PGG between core biopsy and of 3, and to 7 patients (87.5%) with a bPGG of ≥4.
RP specimen was considered upgrading;12 this increase was con- Patients presenting with upgrading in the pathology
sidered clinically relevant when switching from low-grade PCa at
report showed smaller prostates, slightly higher PSA values,
biopsy to HG-PCa after RALP. Upstaging was defined as the patho-
and higher PSAD (Table 1). On multivariable analysis, core
logic diagnosis of a non–organ-confined disease not clinically sus-
pected before RP by means of routine investigation. involvement and PSAD were found to be predictors
Preoperative data were used to identify patients theoretically of upgrading (Table 2). The latter showed an AUC of 0.712
eligible for active surveillance (AS) according to EAU guidelines,9 (95% confidence interval [CI] 0.645-0.780, P = .000) for
defining a binary variable for the presence or the absence of a the prediction of upgrading. At the best cutoff value of
low-risk PCa (>10 years of life expectancy, cT T1/2, bGS ≤6, PSA 0.13 ng/mL2, the PSAD showed a sensitivity of 83%, a speci-
≤10 ng/mL, fewer than 3 prostate biopsy cores positive with ≤50% ficity of 57%, a PPV of 25%, and an NPV of 95%.
of cancer in each core). Patients presenting with upstaging in the pathology report
showed smaller prostates, a larger number of positive cores,
Statistical Analysis a higher percentage of core involvement, and higher PSA
Statistical analysis was performed using the SPSS 12.0 software. values and PSADs (Table 2). The multivariate analysis
Evaluation of data distribution showed a non-normal distribu- showed that core involvement and PSAD were found to
tion of the study data set. Differences between groups of pa-
be predictors of upstaging (Table 3). The latter showed
tients in medians for quantitative variables and differences in
distributions for categorical variables were tested with the Kruskal-
an AUC of 0.628 (95% CI 0.566-0.689, P = .000) for
Wallis 1-way analysis of variance and chi-square test, respec- the prediction of upstaging. At the best cutoff value of
tively. Using multiple logistic regression with the enter method, 0.13 ng/mL2, the PSAD showed a sensitivity of 62%, a speci-
the statistically significant variables as assessed in the univari- ficity of 57%, a PPV of 38%, and an NPV of 78%.
ate analysis were entered and investigated as predictors of up- Overall 90 patients (24%) were possible candidates for
grading and upstaging. The logistic regression analysis was carried AS. Of these patients, 13 (14%) presented with upstaging
.583
.000
.065
.004
.000
.000
.000
P
accuracy of PSAD in these patients is described in Table 3.
(0.170, 0.120-0.280)
(n = 111, 29%)
0.250 ± 0.313
(8.1, 6.1-12.0)
COMMENTS
(63, 59-68)
(47, 40-56)
(28, 25-31)
(40, 18-55)
28.2 ± 4.0
10.7 ± 9.9
Upstaging
4.5 ± 3.1
63.0 ± 6
51 ± 19
38 ± 20
(4, 2-6)
Bioptic GS and cT, together with the PSA level at diag-
nosis, are widely adopted variables used to stratify pa-
tients with PCa in prognostic groups so that appropriate
treatment options can be considered.1,9 Several studies,
however, have reported a poor concordance between clini-
cal assessment and pathologic findings that might result
in inappropriate therapeutic recommendations.2,3,8,12-16
(0.110, 0.080-0.0.198) Recently, Epstein et al proposed a new grading system
(n = 268, 71%)
0.160 ± 0.123
to provide a more accurate tumor stratification, simplify-
(7.1, 5.5-9.9)
No Upstaging
25.0 ± 20.5
(63, 58-68)
(54, 44-71)
(26, 25-28)
26.6 ± 3.2
59.8 ± 23
(20, 9-40)
8.1 ± 4.1
3.1 ± 2.4
62.6 ± 6
.000
.610
.053
.000
.783
.097
P
0.241 ± 0.185
(8.1, 6.7-11.0)
(46, 40-52)
(27, 24-31)
(35, 10-50)
28.2 ± 5.1
10.0 ± 6.0
33.7 ± 23
Upgrading
3.5 ± 2.4
49 ± 19
(3, 2-5)
63 ± 6
D’Elia et al (14.1%).15
0.177 ± 0.206
(7.3, 5.5-10.0)
No Upgrading
27.9 ± 21.1
(63, 58-68)
(54, 43-70)
(20, 10-43)
26.9 ± 3.2
3.6 ± 2.7
62.7 ± 6
58 ± 22
(3, 2-5)
(0.13, 0.09-0.22)
0.186 ± 0.204
(7.4, 5.5-10.5)
28.8 ± 21.4
(63, 58-68)
(52, 42-68)
(22, 10-45)
27.1 ± 3.5
8.9 ± 6.5
3.6 ± 2.7
(n = 379)
62.7 ± 6
(3, 2-5)
Overall
Positive cores
Table 3. Predictive characteristics of PSAD in patients eli- sensitivity of PSAD was 83% for upgrading and 62% for
gible for AS upstaging.
Upgrading Upstaging In our study, moreover, in a subpopulation of 90 pa-
tients theoretically eligible for AS according to the EAU
Sensibility 84 40
(%) guidelines,9 13 (14%) were found upstaged on pathologic
Specificity 85 78 examination and 15 (17%) were upgraded. In this group
(%) of patients, in the ROC analysis, PSAD showed AUCs of
PPV (%) 50 27 0.894 (95% CI 0.808-0.97, P = .000) and 0.689 (95% CI
NPV (%) 97 87 0.539-0.840, P = .021) for the prediction of upgrading and
Accuracy 85 72
(%) upstaging, respectively.
AUC 0.894 (0.808-0.979) 0.689 (0.539-0.840) Our results are consistent with those from recent pro-
spective AS studies that have demonstrated that a more
AUC, area under the curve; NPV, negative predicting value; PPV,
positive predictive value. aggressive disease is found in up to one-third of the pa-
tients eligible, on confirmation biopsy.26,27 Noteworthy is
that Ha et al recently proved that removing the PSAD cri-
In our cohort, patients with a locally advanced PCa at terion from those adopted by the National Comprehen-
the pathologic examination had smaller prostates sive Cancer Network or the Prostate Cancer Research
(51 ± 19 mL vs 59.8 ± 23, P = .000), a higher percentage International Active Surveillance (PRIAS) would expand
of core involvement (38% ± 20% vs 25.0% ± 20.5%, eligibility at the expense of the significant increase in the
P = .000), higher PSA values (10.7 ± 9.9 ng/mL vs 8.1 ± 4.1, rate of upgrading and upstaging. Contrariwise, reducing the
P = .004), and higher PSADs (0.250 ± 0.313 ng/mL2 vs PSAD threshold from 0.2 to 0.15 ng/mL2 in the PRIAS
0.160 ± 0.123 ng/mL2, P = .000). On multivariate analy- study or adding a PSAD cutoff of 0.15 ng/mL2 to the cri-
sis, core involvement and PSAD were found to be predic- teria adopted by the University of California, San Fran-
tors of upstaging (OR 1.020, 95% CI 1.020-1.034, P = .003; cisco, would reduce the rate of unexpected adverse pathology
and OR 5.656, 95% CI 1.285-24.894, P = .022, respec- diagnosed at prostatectomy.27
tively). Similar pieces of evidence were previously pub- We must acknowledge several limitations to the present
lished by other authors.7,20 trial.
Initially introduced to improve PSA testing sensitivity Firstly, the database was prospectively maintained but
and specificity for PCa screening,21 PSAD has clearly shown was analyzed retrospectively. Furthermore, in the present
a strong relationship with PCa aggressiveness,17 proving to study, no pathologic review of all the specimens was per-
be helpful in predicting high-grade and extracapsular disease, formed, and interpretative modifications might in part affect
seminal vesicle invasion, and lymph node involvement.6 the results of our study. However, the lack of a pathologic
Recent papers, moreover, have supported the inclusion of review was, common to other studies validating the Epstein
PSAD into the risk stratification system for candidates for classification.28
AS, proving that it is significantly more accurate than PSA Another limitation is that we calculated PSAD taking
in predicting upgrading and upstaging in this specific into consideration PV values assessed during transrectal ul-
population.22 Notwithstanding these pieces of evidences, trasound, made by 4 different experienced operators. Even
although widely adopted as a risk stratification tool,23-25 though a standard protocol was followed, the interobserver
PSAD has not been recognized by the EAU as an AS in- variability may have negatively affected the validity of the
clusion criteria.9 results. Some other authors, to overcome the limitation of
Overall, in the present series, PSAD proved to be ef- an operator-dependent assessment of the prostate size, re-
fective in predicting adverse pathologic features. In the ROC curred to the measured volume of the specimen ex vivo
analysis, PSAD showed AUCs of 0.712 (95% CI 0.645- (often before seminal vesicle removal) for the PSAD
0.780, P = .000) and 0.628 (95% CI 0.566-0.689, P = .000) calculation.4,7 In our opinion, however, this choice may limit
for the prediction of upgrading and upstaging, respec- the clinical applicability of their findings, completely vio-
tively. At the best cutoff value of 0.130 ng/mL 2 , the lating the nature of PSAD as a preoperative prognostic tool.