INTRODUCION : PSA is the most important tu-
mor marker in all solid tumor, indispensable in
the management of prostate cancer. Screening for
prostate cancer is still not recomended, although
performed in many countries, which introduced
questions about the usefulnes of PSA in detection
of prostate cancer. The PSA treshold has also
been changed, the value of PSA derivatives revised.
Whether such changes are applicable in non scrrened
population is questionable.
Aim of this study was to evaluate the predictive value
of PSA, free/ total PSA and PSA density in our non
screened population.
Patients and methods: TRUS guided prostate biopsy
was performed in 579 patients. The number of cores
was 6-12. Mean age of the patients was 67.5 years
(30-90). PSA was ranging from 0.41 to 2250 ( mean
38.6ng/ml, median: 11.95, SD 140,45). Digitorectal ex-
amination was considered positive in 351 patients.
Free PSA was measured in 352 patients with the index
ranging from 0.02 to 0.88 ( mean free/total PSA: 0.14,
median:0.13, ). The volume of the prostate was meas-
ured in all patients according the prostate ellipsoid
model, and PSA density calculated according to the
formula PSA/PV. Patients were stratified in 6 groups
according to PSA value ( I: PSA ng/ml, II: PSA 2.5-4,
III: PSA 4-10, IV: PSA 10-20, V: PSA:20 to 50, Group
6: PSA 50 ).
RESULTS: Non homogenicity of the patients can be
seen through the wide range of PSA which was from
0.4 to 2025). Prostate cancer was diagnosed in 233 pts
(40.2%). As expected, the probability of detecting can-
cer was raised with PSA (p), and was extremely rare in
pts with PSA below 4 ng/ml. PSA, free/total PSA, vol-
ume of the prostate and PSA density were significantly
different according to the presence of cancer. Most of
our patients had PSA between 4 and 20 ng/ml. Predic-
tive value of PSA was 20.6% for pts with PSA from 4
to 10 and 32.7% for those with PSA from 10 to 20
ng/ml. Sensitivity, specificity, positive and negative
predictive values for different cut off’s of PSA (4, 10
and 20) was performed. The best results were obtained
for PSA cut off of 10 ng/ml. In the group of patient
with PSA, PSA density more reliable than free/total
PSA index.
CONCLUSION: PSA is still valuable marker for de-
tection of prostate cancer in our non screened popula-
tion. According to our results PSA treshold should not
be lowered below 4 ng/ml. PSA density is a reliable
PSA derivative, free/total PSA index having less im-
portance in pts with PSA below 20 ng/ml.
Key words: prostate cancer, diagnosis, PSA value, non
screened population
INTRODUCION
P
SA PSA testing is responsible for the fact that most
prostate cancer are detected in early, curable stages. It
is also expected that screening for prostate cancer
should be manifested by a reduction in detection rate of
aggressive cancers during subsequent screening. PSA
testing with early diagnosis should lead to less mortality
from prostate cancer, but studies have not proved this
theory. Two studies in the United States and Canada
showed that regions of these countries with higher rates of
PSA testing, prostate cancer diagnosis, and early interven-
tion had prostate cancer mortality rates not significantly
lower than those in areas with less intense testing and
treatment, so the American College of Preventive Medi-
cine recommends against routine population screening
with DRE and PSA. European attitude toward scrrening is
similar stressing that there are many arguments against in-
troducing population-based screening for prostate cancer.
PSA is a prostate tissue marker which can be elevated as
a result of prostate cancer, but BPH or infection can also
cause PSA augmentation. Suspicion of prostate cancer in-
dicates prostatic biopsy. The most universally applied in-
dications include PSA 4.0 or irregular DRE. In all, 22% of
.........................................
The predictive value of PSA in diagnosis of prostate
cancer in non screened population
V. Vukotic
1
, S. Cerovic
2
, M. Kozomara
3
, M. Lazic
1
.
1
Department of urology, Health Center "Dr. D. Misovic"
Belgrade
2
Department of pathology, Military Academy, Belgrade
3
Urologic clinic, Clinical center of Serbia, Belgrade
/STRU^NI RAD
616.65-006.04-097
rezime
new CaP cases are discovered in men with normal PSA
and irregular DRE. Recent studies challenged the PSA
treshold of 4 ng/ml. Cancer detection rate can be as high
as 22.8% in patients whose PSA ranged from 2 do 4
ng/ml. Stamey suggested that PSA was related to prostate
cancer 20 years ago, but that in the last 5 years serum
PSA has only been related to benign prostatic hyperplasia.
In order to refine the specifity of PSA testing, it is sug-
gested that more frequent use of age-specific PSA refer-
ence ranges and isoforms should be encouraged, and the
need for biopsy determined on a case-by-case basis.
Most of the studies are from US, were screenig proto-
cols and early detection might have changed the value of
PSA testing. It is questionable whether the same trends
exists in other regions and other countries since demo-
graphic and racial characteristic as well as dietary habits
can have influnece on PSA concentration.
AIM of this study was to evaluate the predictive value
of PSA, free/ total PSA and PSA density in our non
screened population.
PATIENTS AND METHODS
We analyzed 579 patients in whome transrectal ultra-
sound guided (TRUS) guided biopsies were performed in
the period from 1997 to 200, in the private setting . The
patients came for some sort of urinary disturbance or were
referred by other urologist for TRUS biopsy. The mean
age of the patients was 67,5 years ( 30 – 90 years). Serum
PSA was performed at least once prior to biopsy in all pts
(mean PSA:38.6 ng/ml) with Hybritech method of mono-
clonal immunoassay. Digital rectal examination was clas-
sified as positive if any suspicious finding was noted such
as induration, palpable node or asymetry of the gland. The
description of DRE was available for 559 pts. The de-
scription of transrectal ultrasound was available for 481
patients, any abnormal finding were described as positive.
Free PSA was measured and F/T index calculated in 351
pts. . Biopsy was performed using transrectal biplanar
probe with automatic 18 GA core biopsy system. Biopsies
were performed by three urologist and all hystological
analysis by two pathologist .
The indication for biopsy was not made according strict
criteria but rather on clinical suspicion of prostate cancer
considering PSA, free/total PSA, DRE, TRUS. In one pa-
tient, adenocarcinoma was found in cervical lymph nodes
and prostate biopsy was performed in search for primary
tumor.
In most of the patients only one biopsy was done, but in
55 patient the biopsy was repeated. Of these patients, 10
had more than 2 biopsies. Classical sextant biopsy was
performed in all patients until 2001, afterwards at least ten
cores were obtained, unless the diagnosis of cancer was
obvious according to DRE or PSA. In the first years we
did not use any anesthetic procedures, afterwards we in-
troduced the use intrarectal Xylocain gel, while instilla-
tion of 5 ml of Lidocaine on both sides of the prostate is
routinely used in all patients in the last two years, which
also made possible the multiple core biopsies.
For histological processing all cores were divided in six
slices and than fixed in 10% phosphate buffered formalin,
processed into wax paraffin and stained with haema-
toxylin-eosin. Biopsy of transitional zone or seminal ves-
icles was not routinely performed.
RESULTS:
As expected, with the routine use of PSA, the number of
biopsies raised each year (Graph 1). The number of posi-
tive biopsies lowered with accumulating experience as a
result of broadened indicitations.
Prostate cancer was diagnosed in 233 patient (40.2%)
which is shown on table 1.
Most of our patients had more than 65 years. PSA , free
/total PSA (F/T PSA), volume, and PSA density, were
significantly different in patients according to the presec-
nce of cancer (Table 2 )
Non homogenicity of the patients can be seen through a
very wide range of PSA which was from 0.4 to 2025
(mean PSA : 38.60, median: 11,9, Standard deviation:
140,47). Digito rectal examination was positive in 352
patient (60,8%). Free / total PSA was done in 351 pa-
tients, the mean free / total PSA index beeing 0,14 (me-
dian 0,13, range from 0,02- 0,88). The mean volume of
the prostate in our patients was 44 ml, the biggest measur-
ing 140 ml.
Since PSA is the most important marker, we grouped
patients in 6 groups according to the level of PSA. Group
one were patients with PSA less than 2.5 ng/ml, group 2
were formed from patients with PSA between 2.5 and 4
ng/ml, group 3: PSA from 4 to 10, group 4: PSA ranging
from 10 to 20, group 5: PSA from 20 to 50 and groupe 6:
Year
200520042003200220012000199919981997
Number of patients
120
100
80
60
40
20
0
Cancer
No
Yes
GRAPH 1
NUMBER OF BIOPSIES
TABLE 1
DIAGNOSIS OF PROSTATE CANCER
Biopsy Freguency Percent
Negative 346 59.8
Positive 233 40.2
Total 579 100
82 V. Vukotic et al. ACI Vol. LII
PSA 50 ng/ml. Most of our patients had PSA between 4
and 20 ng/ml. As expected, cancer detection rate raised
with more elevated PSA (table 3), but we can point out to
the fact that a cancer detection rate of patients whose
PSA was from 4 to 10 was only 20.6%, being slightly
bigger for those who had PSA between 10 to 20 ng/ml
(32.7%).
Since patients with positive cancer with PSA less than
2.5 ng/ml were extremely rare (only 1 patient out of 19
with PSA less 2.5 had a cancer diagnosed), we analysed
the sensitivity, specificity, positive and negative predic-
tive values for PSA with different cutoff posints, the first
one being 4, the other 10 and the third one 20 ng/ml. (Ta-
ble 3a).
We obtained very low specificity with a cutoff of 4
ng/ml; with a cutoff of 10 we still had a good sensitivity
with acceptable specificity.
For those different cut off values of PSA , the value of
PSA density anf free / total PSA were compared using re-
ceiver operating characteristic (ROC) analysis. The area
under the curve (AUC) was larger for PSA density than
for free/total PSA index which is shown on table 4, mak-
ing a PSA density more reliable than free/total PSA index.
Table 4 :AUC for F/T PSA and PSA density acoording
different PSA
Although the possibilty of cancer detection is the most
important issue of prostatic biopsy, other pathological
findings are of interest suggesting the more or less urgent
need need for rebiopsy in cases that cancer was not found
on initial biopsy.
Prostatic intraepithelial neoplasia was frequently associ-
ated with cancer, 118 patients (50.8%) had PIN along
with cancer. PIN as a dominant pathological finding was
seen in 95 patient (16.4%). Other pathological findings
such as ASAP or PIA were not so frequent. ASAP was
present in 25 pts (4.3%), while PIA as a sole finding was
diagnosed in only 1 patient. The most frequent benign
condition was prostatitis. The PSA distribution according
to histological diagnosis was statistically significant
(p=0.012), as well as density (p=0.029), while there was
not significant difference in the distribution of free/total
PSA index.
Gleason score was not done in first 23 patients with can-
cer, so analysis of grade and gleason score could have
been performed in 218 patients. Most of the patients had a
Grade 2 cancer, while Gleason score in the majority of
patients was 6 or 7. (Table 6)
DISCUSSION:
The pattern of diagnosis of prostate cancer dramatically
changed in last ten years. The routine use of PSA as a tool
for early diagnosis is essential, although the value of
screening programs is doubtful. The pressure to detect
more cancers in localised stages led to PSA treshold low-
ering to 2.6 ng/ml with the detection rate of 15% . On the
other side is the concern of over detection, meaning that
cancer without potential clinical significance would be di-
agnosed and treated .
In non screened population, PSA is still a very relaible
marker which can be seen through the high detection rate
of cancer which in our group of patients is 40.2%. It is
striking that very few patients had cancers if PSA was
lower than 4ng/ml. This can be explained by the fact that
urologist still rely on digitorectal examination. In our pa-
tient DRE was estimated positive in 352 pts (60.8%), with
more then half (182) having prostatic cancer. In USA,
DRE is used as an indication for biopsy if PSA is less
than 4 ng/ml, in which cases it has a positive predictive
value of 8.8% . The other possible explanation for the low
cancer detection with PSA below 4 ng/ml is the age of our
patients, the mean age being 67.9 years. Patients under
60 years of age represented only 17.9% of the whole
group. This also explaines the low positive predictive
value of PSA which is 0.42 for PSA up to 4, and 0.55 for
PSA up to 10 ng/ml, while in other studies much more
cancer were detected in younger patients with PSA lower
than 4 ng/ml
Patients with PSA values between 4 and 20 constituted
the majority of our patients (374, 64.6%). Even in those
patients the cancer detection rate was low: 25,5% (94 pts).
The PSA density was a good predictor of positive biopsy
TEBLE 2
CHARACTERISTICS OF OUR PATIENT ACCORDING
TO THE PRESENCE OF CANCER
/ Ca positiveCa negative P value
Age
Mean 69,5 66.1 <0.001
median 71 67 /
min 30 38 /
max 90 85 /
Prostate
volume
mean 39.5 47.4
median 35.0 44.4 <0.001
min 10 15 /
max 127 140 /
PSA
Mean 78.27 12.12 <0.001
median 23.8 9.22 /
min 1.5 0.4 /
max 2025 145 /
F/T PSA
Mean 0.12 0.15 <0.001
median 0.10 0.14 /
min 0.02 0.02 /
max 0.88 0.74 /
PSA density
mean 2.27 0.35 <0.001
median 0.71 0.21 /
min 0.04 0.03 /
max 68.18 18.5 /
DRE
positive 182 170 <0.001
negative 42 155 /
TRUS
positive 107 162 <0.001
negative 172 40 /
Br. 4 The predictive value of PSA in diagnosis of prostate 83
cancer in non screened population
in those patients while free/total PSA index was not reli-
able. Free/total PSA index should be more valuable in pa-
tients with lower PSA , and can be improved if measur-
ment of testosterone is included .
The volume of the prostate is important in planning the
optimal number of cores in the first biopsy set. Prostate
volume is also a part of Vienna nomogram which adjust
the number of cores according to age of patients, PSA and
volume of the prostate. But volume is also important since
one of PSA derivative, PSA density is calculated by di-
viding PSA with gland volume. Benson et all suggested
that PSA density over 0.15 may improve the specificity of
cancer detection. Our results also showed the superiority
of PSA density over F/T PSA. The same result is obtained
by Stephan and others . Kobayashi et all found also PSA
density to be of value although they pointed out that
TRUS measured volume is superior over transabdominal
US measurment .
The number of cores is an important factor in detection
of prostate cancer. The standard protocol of sextant bi-
opsy is now rarely used, more extended biopsy with 8, 10,
12 or 14 cores are suggested not only for repeated biop-
sies. Transitional zone might be included. The reason of
augmentation of number of cores lies in the fact that more
cancers are detected with extended protocols, in lower
stages . The use of an extended biopsy protocol at initial
biopsy reduces the number of repeat biopsies required
and decreases the number of PC detection in the follow-
up period. But, the number of cores is also important as
more cores improves the accuracy of the biopsy Gleason
score in predicting the final Gleason score at RP .
In our patients , we rarely used more than 10 cores,
which might also explain the low number of positive bi-
opsies in patients with PSA up to 20 ng/ml.
One of benefits of screening programs is the possibility
to diagnose patients with low stage and low grade of PC .
Since are patients are not screened we would expect to
have more high grade tumors. Still, most of our patients
had grade 2 cancer, the most frequent Gleason score bee-
ing 6 or 7.
Findings of PIN (Prostatic intraepithelial neoplasia) or
ASAP (Atypical small acinar proliferation) on initial di-
agnosis is included in the nomogram for repeat biopsy . It
was recently published that the detection rate of prostate
cancer is higher for PIN (58%) then for ASAP (35%) , al-
though it is pointed from Bostwick laboratories that PIN
is declining in it’s predictive value .
Prostatis was the most frequent benign condition diag-
nosed with biopsy. Asymptomatic inflammation of the
prostate has been recognized to be an important con-
founding factor in patients with an elevated PSA .
Chronic prostatis has the same effect on PSA as prostate
cancer, while free/total PSA might distinguish beetwen
two entites according to Stancik et al , although in our pa-
tients free/total PSA was not valuable in making such dis-
tiction. The use of antibiotics in order to reduce the PSA
provoked by infection might be beneficial. It is suggested
that infection might contribute to prostate carcinogenesis,
so it remains to find out the appropriate way to follow up
those patients.
CONCLUSION:
Although there is a tendency to minimise the value of
PSA in diagnosis of prostate cancer, it is essential to make
difference concerning screening. In developped countries
screening program already detected patients having can-
cer, leaving only younger population in which, of course,
lowering of PSA treshold is mandatory.
But what can countries without screeing program learn
and accept as a policy ?
According to our results, PSA is PSA treshold should
not be lowered under 4, except if digitorectal examina-
tion is abnormal. The use of antibiotics in patients with
PSA up to 20 might add in discrimination of those having
prostatitis or cancer. PSA should be the most important
parameter for biopsy; PSA density is a reliable PSA de-
rivative while free /total PSA index is less important and
TABLE 3
CANCER DETECTION RATE ACCORDING TO PSA
GROUP
PSA Group
Cancer
Total
No Yes (%)
1171(5,5)18
2 13 2(13,3) 15
3 162 42(20.6) 204
4 115 56(32.7) 171
5 33 66(66.7) 99
6 5 64(92.7) 69
Total 345 231(40.1) 576
TABLE 4
AUC FOR F/T PSA AND PSA DENSITY ACCORDING
DIFFERENT PSA
Area under the curve
PSA<10ng/ml PSA <20ng/ml
F/T PSA 0.42 0.38
PSA density 0.68 0.72
TABLE 3A
SENSITIVITY, SPECIFICITY, POSITIVE AND
NEGATIVE PREDICTIVE VALUES ACCORDING O
DIFFERENT CUT OFF’S OF PSA
PSA cut off
4ng/ml 10 ng/ml 20 ng/ml
Sensitivity
0.98 0.81 0.43
Specificity
0.09 0.55 0.89
Positive predictive V
0.42 0.55 0.72
Negative predictive V
-.80 0.81 0.70
No of pts
34 342 140
84 V. Vukotic et al. ACI Vol. LII
as such is not obligatory in patient with PSA up to 20
ng/ml.
SUMMARY
UVOD: PSA je najpozdaniji tumorski marker u onkolo-
giji, neophodan u dijagnostici i terapiji karcinoma pros-
tate. Skrining kod karcinoma prostate nije op{te prihva}en
mada ga mnoge zemlje sprovode. To je ukazalo na ne-
pozdanost PSA u detekcija karcinoma prostate. Do sada
prihva}ena grani~na vrednost PSA je revidirana, kao i
vrednost PSA derivata. Da li ove promene imaju impli-
kacija i u na{oj populaciji bolesnika je veliko pitanje.
CILJ rada bio je da ispita prediktivnu vrednost PSA, od-
nosa izmedu slobodnog i totalnog PSA i PSA gustine u
na{oj grupi bolesnika koji nisu obuhva}eni skrining pro-
gramom.
Pacijenti i metode: TRUS vodjena biopsija prostate je
u~injena kod 579 bolesnika. Broj uzoraka je bio od 6 do
12. Prose~na starost bolesnika bila je 67.5 godina (30-90).
Vrednost PSA kretala se od 0.41 do 2025 (prose~na vred-
nost: 38.6 ng/ml, mediana: 11.95, SD:140.45). Rektalni
pregled je ocenjen kao pozitivan kod 351 bolesnika. F/T
PSA je odredjen kod 352 bolesnika i raspon je bio od 0.02
do 0.88 (prose~ni F/T PSA:0.14, mediana: 0.13).
Volumen prostate je izmeren kod svih na osnovu elipsoid-
nog modela, a PSA gustina izra~unata po formuli
PSA/volume prostate. Bolesnici su stratifikovani u 6
grupa u zavisnosti od vrednosti PSA: Grupa 1: PSA ,
Grupa 2: PSA 2.5-4, Grupa 3: PSA od 4-10, Grupa 4:
PSA od 10-20, Grupa 5: PSA 20-50 i Grupa 6: PSA50
ng/ml).
REZULTATI: Na nehomogenost bolesnika ukazuje
veliki raspon PSA (0.4- 2025). Karcinom je dijagnostik-
ovan kod njih 233 (40.2%). O~ekivano je detekcija karci-
noma bila ve}a ukoliko je PSA bio vi{i, ali je karcinom
bio ekstremno redak ukoliko je PSA bio ispod 4 ng/ml.
PSA, F/T PSAi PSA gustina su se bitno razlikovale u zav-
isnosti od prisustva karcinoma. Najve}i broj na{ih PSA je
imao PSA u rasponu od 4- 20 ng/ml. U grupi 3 ( PSA od
4-10) prediktivna vrednost PSA je bila samo 20.6%, a u
grupi 4 (PSA 10-20 ng/ml): 32.7%. Osetljivost, spe-
cifi~nost, pozitivna i negativna prediktivna vrednost PSA
bile su najbolje pri grani~noj vrednosti PSA od 10 ng/ml.
U grupi bolesnika sa PSA ispod 20 ng/ml, gustina PSA je
bila pouzdaniji parametar u detekciji karcinoma u odnosu
na F/T PSA.
ZAKLJU^AK: PSA predtavlja pouzdan marker u de-
tekciji karcinoma prostate u na{oj populaciji koja nije
podvrgnuta skrining programom. Na osnovu na{ih rezul-
tata, grani~na vrednost PSA ne bi trebalo da bude ispod 4
ng/ml. PSA gustina je pouzdan marker u detekcija karci-
noma kod bolesnika sa PSA do 20 ng/ml.
Klju~ne re~i: karcinom prostate, dijagnoza, PSA
vrednost
REFERENCES
1. van der ast TH, Ciatto S, Martikainen PM, Hoede-
maeker R, et al. Detection rates of high-grade prostate
cancer during subsequent screening visits. Results of the
European Randomized Screening Study for Prostate Can-
cer, Int J Cancer. 2005 Dec 13, ; (pub ahead of print)
2. U.S. Preventive Services Task Force. Screening for
Prostate Cancer: Recommendations and Rationale. Origi-
nally published in Ann Intern Med 2002;137:915-916.
Rockville, MD: Agency for Healthcare Research and
Quality; 2002.
3, Rebecca Ferrini R., Woolf HS., Screening For Pros-
tate Cancer in American Men. American College of Pre-
ventive Medicine Practice Policy Statement, aticle On-
line, 2005
4. Finsky, FP., AndrioleG., Kramer SB., Hayes BR,
Prorok PC., Gohagan JK., for the Prostate, Lung, Colorec-
tal and Ovarian Project Team, Prostate biopsy following a
positive screen in the colorectal and ovarian cancer
screening , J Urol , Vol. 173, 746-751, March 2005
5. Quinn M, Babb P. Patterns and trends in prostate can-
cer incidence, survival, prevalence and mortality. Part I:
international comparison, BJU Int. 2002 Jul;90(2):162-73
6. Busby1 JE, Evans CP., Determining variables for re-
peat prostate biopsy, Prostate Cancer and Prostatic Dis-
eases (2004) 7, 93-98
7. Kravchick S., Peled R., Dorfman R., Agulansky R.,
Ben-Dor D., Cytron S., Predictive criteria for postate can-
ver detection in men with serum PSA concentration of 2.0
to 4.0 ng/ml, Urology, 66: 542- 546, 2005
TABLE 5
PSA ACCORDING TO HISTOLOGICAL DIAGNOSIS
PH / PSA F/T PSA
PSA
density
Ca
mean 88.878 0.1507 2.5464
N 113 45 110
median 23.300 0.1100 0.7412
PIN
mean 12.084 0.1528 0.5004
n 947288
median 10.040 0.1400 0.2629
Ca+PIN
mean 67.596 0.01095 0.2629
N 119 65 118
median 24.000 9.000E-02 0.6955
Prostatitis
mean 12.364 0.1507 0.2927
N 197 137 188
median 9.100 0.1400 0.2015
BPH
mean 12.222 0.1738 0.3812
N 412640
median 7.270 0.1400 0.1953
BPH+
Prostatis
mean 8.304 0.1675 0.1811
N10410
median 7.660 0.1400 0.1568
Total
mean 38.751 0.1454 1,1442
N 574 349 554
median 11.900 0.1300 0.2959
Br. 4 The predictive value of PSA in diagnosis of prostate 85
cancer in non screened population
