REVIEWS

Prostate-specific antigen and

prostate cancer: prediction,
detection and monitoring
Hans Lilja*, David Ulmert‡ and Andrew J. Vickers*
Abstract | Testing for prostate-specific antigen (PSA) has profoundly affected the diagnosis
and treatment of prostate cancer. PSA testing has enabled physicians to detect prostate
tumours while they are still small, low-grade and localized. This very ability has, however,
created controversy over whether we are now diagnosing and treating insignificant cancers.
PSA testing has also transformed the monitoring of treatment response and detection of
disease recurrence. Much current research is directed at establishing the most appropriate
uses of PSA testing and at developing methods to improve on the conventional PSA test.

Prostate-specific antigen (PSA) is a serine protease, the in benign prostatic hypertrophy (BPH) and in prostate
physiological role of which is believed to be liquefy- cancer of all grades and stages2,3. It is released into semi-
ing the seminal fluid1. Prostate cancer causes PSA to nal fluid, in which PSA concentrations range from ~0.3
be released into the circulatory system, increasing the to 3 mg/ml (10–100 µmol/l)4. The majority of PSA is an
level in blood up to 105-fold. Increased PSA levels can, active serine protease with chymotrypsin-like activity,
however, also be caused by benign prostate diseases. and it proteolyses the gel proteins of seminal fluid (for
PSA testing has come into widespread use as a cancer example, semenogelin I (SEMG1) and SEMG2)5. PSA
marker, both for initial diagnosis and for monitoring is therefore believed to have the physiological role of
of response to treatment. Use of PSA has also aided liquefying seminal fluid1.
the prediction of prostate cancer risk and of treat- PSA belongs to the family of glandular kallikrein-
ment outcome. As PSA screening has become more related peptidases. In humans, the genes for all 15
widely used, it has generated increasing controversy: glandular kallikreins are clustered in a locus that
some believe that PSA screening has led to diagnosis spans approximately 280 kb of chromosome 19q133–4
and treatment of prostate cancers that pose no real (Ref. 6). The androgen regulation of PSA is conferred
threat without, importantly, reducing prostate cancer by androgen response elements in the promoter of the
mortality. There is also continuing disagreement over gene encoding PSA, KLK3. The closest paralogue of
the threshold level of PSA that should indicate biopsy. KLK3, KLK2 (encoding human kallikrein 2 (hK2)) is
*Departments of Surgery PSA testing cannot be expected to resolve all ambigu- also androgen-regulated through androgen response
(Urology), Clinical ity with respect to prostate cancer; instead, it may be elements.
Laboratories, Medicine best considered as an indicator of risk to be weighed in Although 11 of the 15 human kallikrein genes appear
(Genitourinary-Oncology),
Epidemiology and
combination with other factors. This Review examines to be evolutionarily conserved, KLK3 and KLK2 show
Biostatistics, Memorial current evidence for the association between PSA and major changes through mammalian evolution7. Mice
Sloan-Kettering Cancer prostate cancer risk, disease outcome and recurrence. and rats carry no functional KLK2 or KLK3 genes,
Center New York, New York We also describe the use of PSA testing for long-term but instead carry 1–10 pseudogene copies of a KLK2
10065, USA.
prediction of prostate cancer and the use of PSA deriva- and/or KLK3 progenitor7. Functional KLK3 genes have
‡
Department of Laboratory
Medicine, Lund University, tives (PSA subforms and PSA dynamics) to improve on been found only in certain primates, and functional
University Hospital UMAS, the PSA test. KLK2 genes in primates and in dogs7. This is notable
205 02 Malmö, Sweden. because humans and dogs are among the few species
Correspondence to H.L. Biology of PSA known to spontaneously develop prostate cancer and
e-mail: [email protected]
doi:10.1038/nrc2351
Transcription of PSA is governed by androgens, BPH. The canine KLK2 promoter contains androgen
Published online restricting high-level production to the prostate epi- response elements, whereas the rodent pseudogenes
13 March 2008 thelium. PSA is synthesized in healthy prostate tissue, lack these elements6.

268 | april 2008 | volume 8 www.nature.com/reviews/cancer

© 2008 Nature Publishing Group
 REVIEWS

At a glance The role of PSA in early detection and screening

In the United States, PSA was approved by the Food and
• Prostate-specific antigen (PSA) is one of the few molecular markers routinely used for Drug Administration as a marker to monitor patients
detection, risk stratification and monitoring of a common cancer. treated for prostate cancer in 1986, and as a diagnostic
• PSA is specific to the prostate but not to prostate cancer: benign prostate diseases marker in 1994. As PSA-based screening has become
often cause increases in serum PSA and most men with increased PSA do not have widespread in the United States, it has led to increased
prostate cancer. incidence of prostate cancer and to stage migration, with
• PSA strongly discriminates different cancer stages: it is higher in men with localized a decreased proportion of metastatic or locally advanced
disease than in cancer-free controls, is associated with stage and grade in localized cancers at diagnosis13.
disease and is higher in patients with metastatic compared with localized disease. There is unquestionable evidence that prostate
• Men with a higher PSA at the time of initial therapy have increased risk of recurrence. cancer risk varies with the levels of PSA circulating in
• PSA is a sensitive indicator of recurrence after radical prostatectomy, but far less the blood14–18 (TABLE 1). However, assessing the value of
sensitive as an indicator of recurrence after radiation therapy. PSA screening involves several difficulties. As noted
• PSA before age 50 is a strong predictor of prostate cancer occurring up to 25 years above, serum PSA levels vary with age and various
later. other factors, and this hampers comparison between
• The introduction of PSA as a screening test has led to a sharp increase in the studies. In addition, most studies of PSA testing are
incidence of prostate cancer because there has been a shift to diagnosis at earlier subject to verification bias, which is described in more
stages and there is probably substantial ‘overdiagnosis’ — men diagnosed with detail below.
prostate cancer whose cancer would never have affected their lives if they had not In the context of detecting prostate cancer, an
had a PSA test. increased PSA level prompts a recommendation that the
• The effects of PSA screening on prostate cancer mortality are not yet clear. man undergo prostate biopsy, with PSA of 4 ng/ml being
the traditional threshold level. In three studies of large
cohorts without extensive prior screening, men with
PSA in blood PSA above this threshold had cancer detection rates of
The normal prostate architecture keeps PSA tightly 27–44%19–21. The specificity and sensitivity of PSA testing
confined, so that only a minute proportion leaks into have been harder to determine, because the prostate can-
the circulatory system. In healthy adult males aged cer status of men with low PSA values is seldom verified
≤50 years, the concentrations are 106-fold higher in by biopsy. The Prostate Cancer Prevention Trial (PCPT),
seminal fluid than in blood, in which the median PSA however, is unique in requesting that all participants
level is ~0.6 ng/ml (Ref. 8). In blood, PSA manifests undergo biopsy. Among 5,112 men in the placebo arm
little or no catalytic activity 9,10. This is mainly due of this trial, a PSA level >4 ng/ml had specificity of 93%
to a ≥105-fold excess of protease inhibitors such as and sensitivity of 24% (Ref. 22).
α1-antichymo­trypsin (ACT) and α2-macroglobulin, The threshold of PSA ≥4 ng/ml is frequently criti-
which inactivate any catalytic PSA by forming stable cized, on the one hand for being too lax and on the
Stage migration covalent complexes10,11. PSA in blood exists in multiple other for being too strict. Numerous studies have
The decrease over time in the
forms: free or in complexes with the various protease reported that prostate cancer is not rare in men with
proportion of men with
prostate cancer who are found inhibitors; as proprotein or mature protein; intact PSA values <4 ng/ml. For example, among men with a
to have advanced stage or nicked. These forms, and their implications for median age of 72 years in the placebo arm of the PCPT,
disease at diagnosis, prostate cancer, will be discussed further below. prostate cancer was detected by biopsy in 6.6% of men
commonly attributed to the PSA levels in the blood span a ~105-fold range, from with PSA <0.5 ng/ml and in 27% of men with PSA
introduction of PSA testing,
which identifies prostate
<0.1 to 104 ng/ml, with levels above 102 ng/ml found 3.1–4 ng/ml (Ref. 17). Similarly, a large European study
cancer at an earlier stage in almost exclusively in men with advanced prostate can- reported a cumulative prostate cancer detection rate of
the disease process. cer. PSA levels in blood are also influenced by other 21–25% for men with PSA of 2.0–2.99 ng/ml, and 33%
prostate disease conditions, such as BPH and prosta- at PSA 3.0–3.99 ng/ml (Ref. 14). Results from the PCPT
Specificity
titis, and by age, body mass index and race (see BOX 1 in particular, however, demonstrate that use of lower
The number of people who
test negative for a disease and for further discussion of racial differences). Whether PSA thresholds increases sensitivity at the expense of
who are disease-free, divided individual variations in PSA level are influenced by specificity, and that no cut-point achieves both high
by the total number of people genetics is an area of active research, but, as yet, we lack sensitivity and high specificity23.
who are disease-free and who definitive data. There is increasing appreciation, although no con-
were tested. For a PSA test this
is the proportion of men with
The increased blood levels of PSA in men with can- sensus, that a single PSA cut-point for recommending
no prostate cancer who have a cer cannot be explained by increased PSA expression; prostate biopsy might be inappropriate. First, no single
low level of PSA. during the development and progression of prostate cut-point distinguishes men into two homogenous
cancer, PSA expression may actually decrease slightly12. groups of high and low cancer risk15,16,18. Second, the risk
Sensitivity
The increased blood PSA levels must instead be caused at which men would opt for biopsy is strongly influenced
The number of people who
test positive for a disease and by increased release of PSA into blood. Although there by personal preference, age, heredity, race and presence
who have the disease, divided are no experimental data on the mechanisms of increased of concurrent disease. Accordingly, there has been a
by the total number of people release, they are believed to result from the disruption of move towards calculating a percentage probability of
who have disease and who prostate architecture seen in prostate tumours, such as prostate cancer based on the PSA level and on additional
were tested. For a PSA test this
is the proportion of men with
disordering of the basement membrane and loss of basal predictors of cancer risk, then presenting this probability
prostate cancer who have cell layer, ductal lumen architecture and epithelial cell to the patient as part of shared decision on biopsy. An
increased PSA. polarity. online calculator, developed by the PCPT investigators,

nature reviews | cancer volume 8 | april 2008 | 269

© 2008 Nature Publishing Group
REVIEWS

Box 1 | PSA and race In a prospective study with a larger number of cases, the
lead time between PSA levels ≥4 ng/ml and subsequent
Prostate-specific antigen (PSA) levels can differ in different racial and ethnic groups. clinical diagnosis of prostate cancer was estimated as
In a study of 1,802 white and 1,673 black men without prostate cancer, Morgan and 5.5 years15. Similarly, Fang et al. studied the risk of pros-
co-workers reported that age-specific PSA values for blacks are both higher and more
tate cancer diagnosis among 549 men after a baseline
variable than those for whites29. Even after adjusting for prostate volume, PSA levels
PSA measurement at age 40–60 (median follow-up
are higher in black men than in white men109. Also, black men with local-stage cancers
have higher PSA levels110. The percentage of free PSA, however, has been reported to ~13 years)32. A PSA above the age-adjusted median carried
be of equal value as an additional predictor with total PSA in black and white men111. a relative risk of cancer diagnosis of ~3.6.
Race appears to be a key variable in prostate cancer: the incidence of prostate cancer Two larger studies have extended prediction to lower
is 34% higher, mortality is twofold higher and stage at diagnosis is less favourable in PSA ranges and longer intervals. Loeb et al. studied
black men than in white men112,113. Despite their importance, our understanding of 1,178 men in their 40s with risk factors for prostate
racial differences in PSA and their clinical implications is far from complete. cancer33. The risk of subsequent prostate cancer diagno-
sis was 14.6-fold higher for men with a baseline PSA
level between 0.7 and 2.5 ng/ml than for men with PSA
uses PSA and other risk factors to provide a continuous <0.7 ng/ml. Lilja et al. investigated the relation between
estimate of cancer risk, rather than a binary positive or PSA levels measured in archived plasma taken from
negative test result24. men aged 44–55 years and subsequent diagnosis of
prostate cancer34. This cohort consisted of 21,277 men
Effects of 5α-reductase inhibitors. 5α-Reductase inhibi- participating in the Malmö Preventive Medicine study, a
tors, such as finasteride and dutasteride, are mostly used cardiovascular risk assessment study conducted between
to treat BPH. These agents, which inhibit the conversion 1974 and 1986 in Sweden. Among these men, 462 were
of testosterone to dihydrotestosterone, also lower PSA diagnosed with prostate cancer at a median of 18 years
levels by approximately 50% in the first year of use, and later. The rate of PSA testing was low in Sweden during
further with longer-term use25,26. Two means have been the study period, leaving this study largely free of verifi-
proposed to adjust biopsy criteria for men on 5α-reductase cation bias. PSA level at age 44–55 was associated with
inhibitors. The first is simply doubling the PSA level; diagnosis of prostate cancer up to 25 years later (FIG. 2).
the second is using a PSA increase of ≥0.3 ng/ml above the The association applied even across low levels of PSA;
nadir as the indicator for biopsy25,27. With these meth- for example, the odds ratio for developing prostate can-
ods the predictive value of PSA appears to be as good as cer for PSA 0.51–1.0 ng/ml compared with ≤0.5 ng/ml
or better than conventional criteria in men not taking was 2.51. PSA level at age 44–50 has also been shown
5α-reductase inhibitors. to be highly predictive of advanced prostate cancer,
defined as T3, T4 or metastatic at diagnosis35. In a fur-
Age-specific reference ranges. After age 50, the median ther analysis of the Malmö Preventive Medicine cohort,
PSA level increases (FIG. 1), presumably as a result of the these younger men were compared with 1,167 men of
increasing frequency of benign prostate diseases with ages 59–61 years (Ref. 36). This analysis showed that the
age: BPH, for example, is found in only a few percent of prognostic accuracy of PSA (both the conventional total
40 year olds but in a quarter of 60 year olds28. Because PSA and complexed PSA, described below) decreased
of this rise in PSA, age-specific reference ranges have with age, a difference hypothesized to result from greater
been proposed as a means of increasing the sensitivity prevalence of BPH (and therefore of non-cancer-related
of detection in younger men and the specificity in older PSA increase) among older men.
men29,30. Age-specific ranges have, however, been criti- All these studies indicate that men who will eventu-
cized, mainly for missing clinically significant cancers in ally develop prostate cancer have increased PSA levels
older men31, and have not become uniformly accepted. years or decades before the cancer is diagnosed. This
finding suggests that release of PSA into blood might
Long-term prediction. Several studies have indicated that begin early in the process of carcinogenesis, and could
PSA indicates the risk of prostate cancer years, or even even, perhaps, be a causative factor (FIG. 3). A practical
decades, before diagnosis. However, there are methodo- implication of these studies is that results of a PSA test
logical problems in using PSA to predict future prostate before age 50 could be used to risk-stratify men for fre-
cancer in a population subject to PSA testing. If PSA test- quency or type of later prostate cancer screening.
ing is the primary means of detecting cancer, then early
increases in PSA, which are correlated with subsequent Prediction of disease characteristics and treatment out-
high PSA levels, would be statistically associated with come. Numerous subsequent studies have confirmed
prostate cancer diagnosis irrespective of any biological the original data from Stamey et al. that higher PSA
association. This is known as ‘verification bias’. The first level is associated with larger tumour volume37,38 and
long-term prediction study, which reported that PSA higher clinical stage38, pathological stage39 and Gleason
levels >2.5 ng/ml predicted diagnosis of prostate cancer grade 37,39. With repeated testing, however, diagno-
over the subsequent decade, was free from verification sis of high-Gleason-grade disease and particularly
bias as all cancers were diagnosed before the PSA era16. advanced disease stages becomes less common. This
However, conclusions from this study are limited by sub- has been seen in both the Swedish screening arm of the
stantial degradation of PSA in the archived samples and European Randomized Study of Screening for Prostate
by the fact that the study included only 44 cancer cases. Cancer (ERSPC), where men were offered biennial

270 | april 2008 | volume 8 www.nature.com/reviews/cancer

© 2008 Nature Publishing Group
 REVIEWS

Table 1 | Prostate cancer detection rates for PSA cut-points in various age groups
Cut-point Prostate cancer detection rate (%)
(ng/ml)
Lane117 Donovan118 Aus14 Postma44 Thompson22
n = 496 n = 7,383 n = 5,855 n = 19,970 n = 5,112
45–49 years 50–59 years 60–69 years 50–66 years 55–74 years 62–90 years
1.0 – – – – – 28
1.5 19 – – – – 31
2.0 – – – – – 34
2.5 – – – – – 37
3.0 – 23 33 42* 29† 40
4.0 – 25 34 47* 31 48
5.0 – 29 38 – – –
6.0 – 37 42 – – 43
7.0 – 40 48 64* – –
10.0 – 55 63 77* 57 –
*Cumulative rate of diagnosis over a median follow-up of 7.6 years. †n = 10,191 for this cut-point.

re-screening21, and in the Rotterdam arm of the ERSPC, in black men. By contrast, investigators in the Rotterdam
which had a 4 year screening interval40. Although the section of the ERSPC reported that, after two rounds
association between PSA and stage and grade is insuffi- of screening, 43% of the men who underwent radical
cient to allow useful predictions for individual patients prostatectomy had cancers considered to be minimal44.
based on PSA as a single marker, PSA is used in several Whatever definition is used for insignificant cancer,
multivariable models that predict prostate cancer stage there are currently no means that allow a clinician to
and grade. distinguish insignificant from potentially lethal cancer,
The PSA level before treatment is also a predic- and therefore to determine which men require treat-
tor of risk of recurrence after treatment. Nearly all ment. This determination would probably require
multi­variable models that predict disease recurrence consideration of a man’s general medical condition,
incorporate the pre-treatment PSA. In models designed rather than merely focusing on the tumour, because
for use before treatment PSA tends to be the strongest the likelihood that a given prostate cancer will affect a
single predictive factor, but in models for use after pros- man’s quality of life or survival depends on his risk of
tatectomy but the strongest factor tends to be either the death from other causes.
Gleason score or the proportion of high-grade tumour It is evident that any harm from overdiagnosis and
in the removed prostate. overtreatment of insignificant cancers must be balanced
against the benefit presumed to result from early detec-
Early diagnosis versus overdiagnosis tion and early treatment of prostate cancers with lethal
One of the principal challenges in the diagnosis and potential. Several studies have addressed the question
treatment of prostate cancer is that clinically insigni­ of whether prostate cancer screening reduces prostate
ficant prostate cancer is quite common. Autopsy of men cancer-specific mortality; TABLE 2 gives an overview of
who died of unrelated causes has demonstrated that randomized and observational studies published since
latent prostate cancer is present in approximately 70% of 2000. Interpretation of the observational studies is com-
men in their 60s41. This contrasts with the estimated 17% plicated not only by well-known problems, such as con-
lifetime risk of prostate cancer diagnosis in the United founding by unmeasured differences between groups,
States42. Clearly, many prostate cancers pose no threat to but also by two related factors more specific to PSA.
life or health. PSA screening leads to detection of some First, studies of screening programmes conducted just
of these insignificant cancers, creating the potential for after they are introduced (1987–1992 in the case of PSA
patient anxiety and unnecessary treatment, with attend- testing) are subject to a variety of biases. For example,
ant side effects. Several studies have estimated the rate in a case–control study, a proportion of men who had
of overdiagnosis that results from PSA screening. These PSA tests in 1987–1992 and who later died of pros-
estimates cover a wide range, in part because of dif- tate cancer would have been screened too late in the
ferent definitions of overdiagnosis. In a recent report, course of their disease for screening to be of benefit.
Telesca et al. defined overdiagnosed cases as those that This effect is attenuated after screening programmes
would not have been detected by clinical means within become established and the spike in incidence associ-
the man’s lifetime43. The estimated rates of overdiagnosis ated with pre-existing, more advanced cases passes.
varied with age, but for men of ages 60–64 PSA screening Second, prostate cancer is a slowly growing malig-
in the United States was estimated to result in overdiag- nancy and effects on survival are likely to become
nosis of 10% of cancers in white men and 21% of cancers apparent only after many years. In the Scandinavian

nature reviews | cancer volume 8 | april 2008 | 271

© 2008 Nature Publishing Group
REVIEWS

0.5 Complexed and free PSA. A majority of PSA in blood

Malmö 40–50 years; n = 16,286; median = 0.63 ng/ml occurs in stable covalent complexes with protease
Malmö 51–55 years; n = 2,436; median = 0.81 ng/ml inhibitors (FIG. 4); these forms are collectively known
Malmö 60 years; n = 1,167; median = 1.11 ng ng/ml as complexed PSA (cPSA). Complexes with ACT, a
0.4
ERSPC 59–61 years; n = 950; median = 1.08 ng/ml
SERPIN-type inhibitor, are predominant and are readily
detectable by most immunoassays, whereas complexes
0.3
with α2-macroglobulin escape detection by commercial
PSA assays10,49,50. The non-complexed forms, known as
Proportion

free PSA (fPSA), are unreactive with plasma protease

inhibitors. Epitopes exposed by fPSA but not by cPSA
0.2
enabled the development of assays specifically measur-
ing fPSA or cPSA as complements to the conventional
PSA assay, which measures total PSA (tPSA)49,50
0.1 In proportion to tPSA, fPSA is lower in men with
prostate cancer than in men with BPH51. The ratio of
fPSA to tPSA (generally expressed as the percentage of
0.0 free PSA (%fPSA)) adds significant information when
0.00–0.50 0.51–1.00 1.01–2.00 2.01–4.00 4.01–10.00 >10.00 separating men with BPH from those with prostate
PSA (ng/ml) cancer, although the magnitude of the effect has varied
Figure 1 | Distribution of prostate-specific antigen (PSA) values among men of across studies51,52. One possible explanation for this
different age groups. Data are derived from studies of the MalmöNature Reviews
Preventive | Cancer
Medicine variability is that fPSA in blood is much more labile
(MPM) cohort18,34,36 and the Swedish section of the European Randomized Study of
than cPSA, so suboptimal handling of blood samples can
Screening for Prostate Cancer (ERSPC)14. PSA data from the MPM study cohort in
Malmö have been multiplied by 1.13 to compensate for the assay yielding results 13%
greatly lower fPSA levels53 and thus weaken the associa-
lower than those calibrated against the World Health Organization standard. tion between %fPSA and prostate cancer. Catalona and
co-workers found prostate cancer in 56% of the subjects
with %fPSA <10 but in only 8% of those with %fPSA
trial of radical prostatectomy versus watchful waiting45, >25; they therefore proposed %fPSA of ≤25% as the
surgery was associated with important benefit at criterion for biopsy52. Moreover, %fPSA has shown evi-
10 years (incidence of cancer-specific death 5.3% dence of increasing the diagnostic accuracy at both low
lower), but at 5 years the difference between groups and high levels of tPSA54. The most convincing results
was only 2.0%. Therefore, it appears unlikely that a came from a careful meta-analysis of 66 studies, which
population-based study with short follow-up will be clearly verified better diagnostic performance (that is, a
able to detect differences of this magnitude. higher prostate cancer detection rate) for %fPSA than
Two large randomized trials are currently in for tPSA55. However, larger prostate volume in patients
progress: the Prostate, Lung, Colorectal, and Ovarian with prostate cancer is correlated with higher %fPSA,
screening trial and the ERSPC. The results are expected and thus patients with concomitant BPH might receive
to mature in 2009 or 2010. Meanwhile, as shown in false-negative %fPSA results due to a dilution effect
TABLE 2, the Swedish arm of ERSPC has yielded data on caused by large prostate volumes56.
the effect of screening on the incidence of metastatic Despite the evidence that the combination of %fPSA
disease at the time of diagnosis46. Among the 9,972 and tPSA can outperform tPSA (or cPSA) alone as
men who were randomized for biennial screening, 24 predictor of immediate biopsy results, %fPSA might be
of 810 men diagnosed with cancer had metastases at less useful as a long-term predictor of prostate cancer.
diagnosis, compared with 47 of 442 of those diagnosed In the retrospective study of men of ages 44–50 in the
with cancer in the control arm, a risk reduction of 49%. Malmö Preventive Medicine cohort, a multivariable
Similar results were noted in the Rotterdam arm of model incorporating tPSA, fPSA, %fPSA, and hK2 had
ERSPC. no better predictive power than tPSA alone34. However,
among study participants of ages 59–61, the multivari-
PSA derivatives able model was superior to tPSA alone (AUC 0.758 versus
PSA density. Prostate cancer causes release of more PSA 0.819, respectively), presumably because the time from
into circulation per unit volume than does BPH38. The PSA measurement to diagnosis was shorter and the fre-
ratio of serum PSA to prostate volume (measured by quency of increased PSA levels due to BPH was higher
transrectal ultrasound) might therefore help distinguish in the older men36.
prostate cancer from BPH. This ratio, the PSA density,
AUC
has shown discriminatory power in many studies47, ProPSA, intact PSA, nicked PSA and BPSA. PSA is
Area under the receiver-
operating characteristic curve. although other studies have been negative48. A low PSA initially produced as a 261-amino-acid preproprotein.
This value gives the probability density has also been associated with tumour aggres- Co-translational removal of an amino-terminal leader
that, in a pair of patients, one siveness and with unfavourable pathological features. generates a non-catalytic zymogen (proPSA). Subsequent
of whom had the event and the However, PSA density requires a transrectal ultrasound, removal of the 7-residue propeptide generates the cata-
other of whom did not, the
patient who had the event was
which is expensive, time-consuming and causes signi­ lytically active mature form, a 237-residue single-chain
given the higher risk by the ficant patient discomfort. This might help to explain why enzyme containing five intrachain disulphide bonds
predictive model. PSA density is not widely used clinically. (FIG. 4).

272 | april 2008 | volume 8 www.nature.com/reviews/cancer

© 2008 Nature Publishing Group
 REVIEWS

Free PSA in blood is a mixture of mature PSA 80 25 50 75 90

and proPSA, including some proPSA forms with the
propeptide N-terminally truncated at various posi- 70

Probability of prostate cancer (%)

tions57. Several studies have suggested that proPSA,
60
particularly the –2 truncation, might aid in discrimi-
nating prostate cancer from benign disease58–60. Free 50
PSA in serum can also be divided into intact forms
40
(both mature and proPSA) and nicked forms, the lat-
ter cleaved between Lys145 and Lys146. The level of 30
intact PSA and the ratio of nicked to total PSA have
shown potential for improving the discrimination 20
of prostate cancer from BPH 61,62. A second distinct 10
cleaved form, BPSA, has been associated with prostate
volume and therefore might also help to discriminate 0
0 1 2 3 4
prostate cancer from BPH63. All of these PSA forms are
worthy of further research to determine whether they PSA at baseline venipuncture (ng/ml)
can improve the accuracy of prostate cancer detection, Figure 2 | Early prediction of prostate cancer risk. The
particularly if combined into a panel of markers. Nature diagnosis
predicted probability of a prostate cancer Reviews | Cancer
before the age of 75 years by total prostate-specific
PSA dynamics: velocity and doubling time. PSA veloc- antigen (PSA) measured at age 44–50 years, with 95%
ity, the change in PSA level over a specified time inter- confidence intervals. The vertical lines represent the 25th,
50th, 75th and 90th percentiles of baseline PSA, and the
val, has been much advocated as a means of identifying
horizontal line represents the average lifetime risk (10%)
men with prostate cancer. PSA velocity is strongly of a prostate cancer diagnosis before the age of 75 years.
associated with diagnosis of prostate cancer64 and with The illustration is based on data from Ref. 34. Note that
risk of recurrence or cancer-specific death after treat- the PSA levels reported from this study are approximately
ment65,66. PSA doubling time, the time required for the 13% lower than values derived from assays calibrated
PSA level to double, is mostly used to monitor disease against the World Health Organization standard.
progression for patients after initial surgery and radio-
therapy and for patients who choose surveillance rather
than definitive treatment67. What has not been shown, Monitoring of PSA after definitive local therapy for
however, is that either of these measures increases diag- prostate cancer provides a means of detecting recurrent
nostic or predictive accuracy over that provided by a prostate cancer long before the tumour is detectable by
simple PSA level test. PSA velocity added no predictive any other means (for example, imaging or symptoms).
value to PSA level in either a study of long-term predic- Increasing PSA levels after treatment generally indicate
tion of prostate cancer18 or in the PCPT study of biopsy the presence of tumour cells; however, there is debate
outcome24. on the best criteria for defining biochemical recurrence.
The literature frequently refers to ‘undetectable PSA’
PSA as an indicator of disease recurrence after treatment, but the meaning of undetectable has
changed with improvements in the sensitivity of PSA
assays. Before the mid-1990s, the typical limit of detec-
a tion was around 0.3 or 0.4 ng/ml; today, the typical limit
Premalignant
Carcinogenesis
changes
Rise in PSA of detection is below 0.10 ng/ml.
Prostate cancer can recur in the absence of a PSA
increase or detectable PSA level, although the frequency
b of this event appears to be low68,69.
Premalignant
Carcinogenesis Rise in PSA
changes Prostatectomy. Some believe that any detectable PSA
level after prostatectomy is evidence of treatment failure,
but most define biochemical recurrence as PSA above
c
some selected level, ranging from 0.2 to 0.5 ng/ml.
Carcinogenesis
Premalignant
Rise in PSA Stephenson et al. compared ten different definitions of
changes
biochemical recurrence by testing their ability to pre-
dict metastatic progression70. The definition that was the
Figure 3 | Three non-exclusive hypotheses to explain the association between best predictor was a PSA value ≥0.4 ng/ml followed by
prostate-specific antigen (PSA) level in younger men and prostate cancer
Nature Reviews | Cancer a second PSA test at a higher level. Definitions using
diagnosed up to 25 years subsequently. a | A carcinogenic process causes premalignant
two PSA values above a given threshold have, however,
changes in prostate cells, which in turn increases leaking of PSA into the bloodstream.
b | A carcinogenic process causes premalignant changes in prostate cells. These changes
been criticized for skewing outcome and delaying the
are not sufficient to cause increased levels of serum PSA; however, carcinogenesis appearance of treatment failure71.
independently causes increased serum PSA by a separate process. c | An unknown process Some patients develop a detectable PSA after radical
causes an increase in serum PSA; extracellular PSA is causally influencing the carcinogenic prostatectomy but have no further increases in PSA71.
process, which leads to premalignant changes in the prostate. Possible explanations for this low, stable PSA level

nature reviews | cancer volume 8 | april 2008 | 273

© 2008 Nature Publishing Group
REVIEWS

Table 2 | Studies on the effects of prostate cancer screening

First author Design Main findings Comments and limitations
and reference
Observational studies
Shibata119 Ecological: comparison of Decrease in death rates more rapid in US Geographical differences in mortality rates might
Oliver120 prostate cancer incidence and than UK, suggesting a favourable effect of be associated with various factors other than
mortality rates in UK and US screening screening; relatively short follow-up
Roehl121 Case series Recurrence-free survival after surgery better Not population-based; end point not mortality
for patients in a screening study than for
referred patients (83% versus 77% at 7 years).
Oberaigner122 Ecological: comparison of Reduction in mortality seen in Tyrol (risk ratio of Geographical differences in mortality might be
mortality rates in different 0.81 per decade), where intensive screening is due to factors other than screening; mortality
regions of Austria offered, but not other regions (risk ratio of 1.00) reductions not formally compared
Shaw123 Ecological: comparison of Decreases in prostate cancer mortality greater Focused on the early period of screening, which
mortality rates in different in regions where screening more intensive, but is vulnerable to a variety of statistical artefacts;
regions of the US differences were small and less than predicted geographical differences in mortality rates
by a statistical model affected by factors other than screening
Kopec124 Case–control: cases were Lower rate of PSA tests in men with metastatic Men who choose to undergo PSA tests may differ
men with metastatic prostate cancer (OR 0.65) from those who forgo testing
cancer
Concato125 Case–control: cases were men No difference in rate of PSA tests in men who Focused on the early period of screening, which
who died from prostate cancer died from prostate cancer (OR 1.08) is vulnerable to a variety of statistical artefacts;
low number of patients undergoing screening;
extremely short survival after diagnosis
Weinmann126 Case–control: cases were men Lower rates of PSA tests in men who died from Focused on the early period of screening, which is
who died from prostate cancer prostate cancer (OR 0.73) vulnerable to a variety of statistical artefacts; men
who choose to undergo PSA tests may differ from
those who forgo testing; could not distinguish
effects of PSA from DRE testing
Lu-Yao127 Ecological: comparison of No reduction in prostate cancer mortality in Focused on a very early period of PSA screening
prostate cancer mortality in region where rate of PSA testing was higher with limited follow-up; only older men included;
two regions of the US (OR 1.03) geographical differences in mortality rates might
be associated with a variety of factors other than
screening
Ciatto128 Cohort: comparison of actual Death from prostate cancer in the cohort Cohort included men screened by DRE only; in
and expected death rates from lower than expected after exclusion of some analyses, benefits also seen for men who
prostate cancer in a cohort prevalent cancers (standardized mortality rate did not attend screening, suggesting bias
invited to screening of 0.72)
Coldman129 Ecological: comparison of Decreases in prostate cancer mortality greater Limited follow-up; geographical differences in
mortality rates in different in regions where PSA testing was likely to be mortality rates might be associated with various
regions of British Columbia, low (29% versus 14%), suggesting no effect of factors other than screening
Canada screening
Randomized studies
Labrie130 46,486 men in Quebec, Lower death rates in men in screening group Contamination: only 24% of men randomized
Canada who accepted screening compared with to screening accepted the invitation; 7% of
control group (19.8 and 52.3 per 100,000 man men in the control group were screened; main
years) comparison was not between randomized groups
and was therefore subject to bias
Sandblom131 9,026 men in Norway No difference in prostate cancer mortality Only 1 in 6 men randomized to screening;
between groups screening included PSA testing only in later rounds
Aus46 20,000 men in the Swedish Fewer cases of metastatic cancer in the Interim analysis of one centre in a multicentre
section of the European screening arm (risk ratio 0.51) study
Randomized Study of Prostate
Cancer Screening (ERSPC)
van der 35,000 men in the Dutch 7 cases of metastasis in 17,635 screened Interim analysis of one centre in a multicentre
Cruijsen- section of the ERSPC patients compared with 27 with 17,513 study
Koeter132 controls
DRE, digital rectal examination; OR, odds ratio; PSA, prostate-specific antigen.

include benign prostate tissue left behind by surgery, Radiation therapy. The issues relating to PSA after radia-
spurious signal caused by non-specific interaction of a tion therapy are complex, and we will describe them only
blood component (for example, a heterophilic antibody) in brief. After radiation therapy, PSA levels decrease
with an assay reagent72, or (less likely) PSA produced slowly. The time to reach PSA nadir can be months to
and released into blood from other organs73. years after treatment. The level of the post-radiation

274 | april 2008 | volume 8 www.nature.com/reviews/cancer

© 2008 Nature Publishing Group
 REVIEWS

A common phenomenon that confounds monitor-

Prostate
ing of patients after radiation therapy is ‘PSA bounce’
ProhK2 ProPSA — transient benign increases in PSA that subsequently
decrease to pre-increase levels79,80. Attempting to sum-
marize findings on PSA bounce is difficult because
bounce is not defined in a uniform manner. However, it
is estimated that 30–55% of patients treated with brachy-
hK2 PSA therapy experience PSA bounce, starting a median of
15–20 months after implantation81–83. PSA bounce might
be less common after external beam therapy, and bounce
occurs later, at a median of around 30 months after treat-
ment84–86. The median magnitude of the bounce depends
on the definition used, but ~0.6 ng/ml is typical for
Prostate fluid Blood
both brachytherapy and external beam therapy. Among
patients receiving brachytherapy, PSA bounce appears
PCI PCI hK2
more common in younger patients81.
PSA hK2 The definition of biochemical recurrence after
PSA hK2 Free PSA radiotherapy is much debated. In an attempt to stand-
ardize definitions, the American Society for Therapeutic
PSA PSA ProPSA
Radiology and Oncology (ASTRO) proposed that bio-
hK2 PSA chemical recurrence after external beam radiotherapy
Complexed PSA be defined as three consecutive PSA increases after PSA
PSA
PSA nadir has been reached, with the date of failure backdated
API to the mid-point between the nadir and the first of these
Ejaculatory mixing ACT three increases. Although this definition has been used
extensively, it was found to be suboptimal as a predictor
SEMG1
Invisible PSA of clinical treatment failure87, and the backdating causes
various biases, including underestimation of failure rates
SEMG2 A2M when follow-up is short. As a result, in 2006 a consen-
PSA
sus group developed a revised definition (known as the
Figure 4 | Prostate-specific antigen (PSA) subforms and interactions. Active forms Phoenix definition or the RTOG–ASTRO definition), in
of PSA and kallikrein-related peptidase 2 (hK2) are shown in red, inactive forms in which biochemical recurrence is defined as an increase
blue or green. In the prostate, propeptides (grey wedge) are removed
Nature from proPSA
Reviews | Cancer by ≥2 ng/ml above the nadir, with no backdating88. This
and prohK2, leaving the mature, catalytic forms. hK2 might be one of the proteases definition predicts clinical failure with higher sensitivity
responsible for these processing events. PSA and hK2 are released at high and specificity than does the original ASTRO definition
concentrations into prostatic fluid, then into seminal fluid, and at low concentrations for both external beam therapy89 and brachytherapy90.
into blood. PSA forms in prostatic fluid are active PSA, nicked PSA and PSA
PSA bounce, however, remains a problem for diagnosing
complexed with protein C inhibitor (PCI, encoded by SERPINA2), a protease
inhibitor. The sizes in the figure indicate the relative abundances of the forms. In
recurrence. As many as 6% of brachytherapy patients can
seminal fluid, active PSA is believed to be responsible for liquefaction of seminal have PSA bounce of ≥2 ng/ml and thus be misclassified
fluid by proteolysing gel proteins (SEMG1 and SEMG2, which are secreted primarily under the Phoenix definition as having recurrence81.
by the seminal vesicles, though SEMG2 is also secreted in small amounts by the
epididymis). Blood contains a variety of forms of PSA: free PSA forms (nicked, intact Androgen deprivation therapy. The standard treatment
and proPSA) and complexed PSA. The most abundant form in blood is PSA for advanced prostate cancer is androgen deprivation
complexed with α1-antichymotrypsin (ACT); complexes with α2-macroglobulin therapy (ADT). After initiation of ADT, PSA in blood
(A2M) or α1-protease inhibitor (API) are estimated to comprise only a 1–2% or lower almost always decreases, then stabilizes for varying
proportion of PSA in blood114–116. A2M envelopes PSA, masking the epitopes intervals91. The failure of ADT to produce a reduction
recognized by commercial PSA assays and thus rendering this form invisible to the
in PSA in a given patient most commonly, although
assays. PSA levels in seminal fluid are 0.5–3.0 mg/ml (~106-fold higher than in blood)
and hk2 levels in seminal fluid are 2–12 mg/ml (~104-fold higher than in blood).
not inevitably, indicates its failure to arrest growth and
to produce cytotoxicity in the tumour. Progression to
androgen-independent disease is generally defined as
PSA nadir appears to be related to the radiation dose74. two consecutive increases of PSA after the post-ADT
Perhaps because of this, PSA typically remains detectable PSA nadir92,93. This PSA increase invariably predicts
External beam therapy after external beam therapy, but it often becomes undetect- objective tumour progression (defined by radiologi-
The use of radiation from a
able after brachytherapy, in which interprostatic radiation cal or clinical evidence) and commonly precedes it
high-energy source external to
the patient as a treatment for intensities are higher75. The time to PSA nadir is influ- by 6–12 months94. ADT causes the initial decrease in
cancer. enced by the size of the prostate and the pre-treatment PSA, not only because of tumour regression, but also
PSA76. Further, a shorter time to nadir after external because ADT suppresses transcription of the PSA
Brachytherapy beam radiotherapy appears to be associated with treat- gene95, which is androgen-dependent. Similarly, the
Implantation of radioactive
pellets, approximately the size
ment failure77. The PSA nadir after radiation therapy also androgen-independent PSA increase indicates not only
of a grain of rice, into the has predictive value; a low PSA nadir is associated with renewed tumour growth but also reactivation of the
tissue being treated for cancer. freedom from biochemical recurrence78. androgen receptor despite castrate levels of testosterone.

nature reviews | cancer volume 8 | april 2008 | 275

© 2008 Nature Publishing Group
REVIEWS

Androgen receptor reactivation, which may be caused trials has been questioned103. There is great need for
by mutation, gene duplication or other mechanisms, is more accurate markers of response.
a common feature of androgen-independent prostate Little is known about PSA derivatives in this disease
cancer96. state, and this topic is worth further research. Free PSA
Whether pre-treatment PSA predicts response to might be particularly well suited to rapid measurement
ADT is not yet clear. The time to PSA nadir91,93,94,97 and of responses to therapy because it is eliminated from
the percentage decrease in PSA after treatment97–100 blood quickly (terminal half-life ~14 hours), compared
have been reported as predictive factors. For example, with the much slower elimination of cPSA (linear
patients whose PSA decreased by at least 80% during elimination at ≤1 ng/ml/day)104. Because cPSA is the
the first month of treatment had significantly longer predominant component of tPSA, elimination of the
disease-free interval97. Further, Furuya and co-workers bulk of PSA measured in the conventional PSA test will
have reported that 5 year survival was 67% in patients be similarly slow.
with normalization of PSA levels compared with 0% in
those without normalization101. Future directions
One of the most important prognostic parameters Whether or not PSA screening is found to be of benefit
is PSA level at its post-ADT nadir. PSA nadir has been at the population level, PSA will continue to be used
associated with time to androgen-independent progres- in the care of the individual prostate cancer patient.
sion92,93, clinical progression91,93,99 and death91,93,98. Kwak Moreover, investigators are actively exploring means
et al. reported that nadir PSA predicted progression to of improving the accuracy of testing for prostate can-
hormone-refractory prostate cancer within 2 years of cer by combining the conventional PSA test with PSA
initiation of ADT with 86.2% accuracy (determined derivatives or with other markers. One marker under
in receiver-operating characteristic analysis)93. Patients investigation is hK2, the protein most closely related to
whose PSA nadir is less than 0.2 ng/ml have significantly PSA (~80% amino acid sequence identity). hK2 resem-
longer intervals to androgen-independent progres- bles PSA in having androgen-regulated expression and
sion75,93,102. Interestingly, the PSA nadir has not been being released into seminal fluid, though its levels are
shown to correlate with the interval from initiation of approximately 100-fold lower than those of PSA. hK2
ADT to the time of PSA nadir93. Also, PSA nadir does also resembles PSA in being increased in the serum
not seem to be associated with PSA level at diagnosis, of patients with prostate cancer. Several studies have
or with stage or Gleason score92,93,102. shown that measuring hK2 adds important informa-
tion to tPSA for detecting prostate cancer36,105, and for
PSA forms in monitoring advanced cancer predicting prostate cancer stage, grade and volume106,107
The conventional total PSA test is frequently used to and recurrent cancer107.
monitor the effects of treatment in men with androgen- An intriguing area of research is PSA-activated anti-
deprivation-refractory prostate cancer. In general, PSA cancer drugs, in which an anticancer drug is linked to
in men with such advanced disease stages is only mod- an inhibitory peptide that is removed by PSA-catalysed
estly associated with survival. Moreover, post-treatment cleavage. In theory, the activation by PSA could make
changes in PSA (or lack thereof) may not accurately the agent highly prostate-specific in its action. This
indicate presence or absence of a response to treatment, approach has been explored with an agent based on the
and the value of PSA as a surrogate measure in clinical cytotoxic thapsigargin108.

1. Lilja, H. A kallikrein-like serine protease in prostatic identification of a progenitor of prostate-specific 14. Aus, G. et al. Individualized screening interval for
fluid cleaves the predominant seminal vesicle protein. antigen. Genomics 84, 147–156 (2004). prostate cancer based on prostate-specific antigen level:
J. Clin. Invest. 76, 1899–1903 (1985). 8. Savblom, C. et al. Blood levels of free-PSA but not results of a prospective, randomized, population-based
2. Herrala, A. M., Porvari, K. S., Kyllonen, A. P. & complex-PSA significantly correlates to prostate study. Arch. Intern. Med. 165, 1857–1861 (2005).
Vihko, P. T. Comparison of human prostate specific release of PSA in semen in young men, while blood 15. Gann, P. H., Hennekens, C. H. & Stampfer, M. J.
glandular kallikrein 2 and prostate specific antigen levels of complex-PSA, but not free-PSA increase with A prospective evaluation of plasma prostate-specific
gene expression in prostate with gene amplification age. Prostate 65, 66–72 (2005). antigen for detection of prostatic cancer. Jama 273,
and overexpression of prostate specific glandular 9. Niemela, P., Lovgren, J., Karp, M., Lilja, H. & 289–294 (1995).
kallikrein 2 in tumor tissue. Cancer 92, 2975–2984 Pettersson, K. Sensitive and specific enzymatic assay 16. Stenman, U. H. et al. Serum concentrations of
(2001). for the determination of precursor forms of prostate- prostate specific antigen and its complex with
3. Lintula, S., Stenman, J., Bjartell, A., Nordling, S. & specific antigen after an activation step. Clin. Chem. α1-antichymotrypsin before diagnosis of prostate
Stenman, U. H. Relative concentrations of hK2/PSA 48, 1257–1264 (2002). cancer. Lancet 344, 1594–1598 (1994).
mRNA in benign and malignant prostatic tissue. 10. Piironen, T. et al. Measurement of circulating forms of This classic paper was the first to suggest that
Prostate 63, 324–329 (2005). prostate-specific antigen in whole blood immediately PSA is powerful as a long-term predictor of
4. Ahlgren, G., Rannevik, G. & Lilja, H. Impaired after venipuncture: implications for point-of-care prostate cancer. The study is, however, limited by
secretory function of the prostate in men with oligo- testing. Clin. Chem. 47, 703–711 (2001). the small number of cancer cases analysed and
asthenozoospermia. J. Androl. 16, 491–498 11. Christensson, A., Laurell, C. B. & Lilja, H. Enzymatic by the degradation of PSA in archived serum
(1995). activity of prostate-specific antigen and its reactions samples.
5. Lilja, H., Oldbring, J., Rannevik, G. & Laurell, C. B. with extracellular serine proteinase inhibitors. Eur. J. 17. Thompson, I. M. et al. Prevalence of prostate cancer
Seminal vesicle-secreted proteins and their reactions Biochem. 194, 755–763 (1990). among men with a prostate-specific antigen level
during gelation and liquefaction of human semen. 12. Qiu, S. D. et al. In situ hybridization of prostate- ≤ 4.0 ng per milliliter. N. Engl. J. Med. 350,
J. Clin. Invest. 80, 281–285 (1987). specific antigen mRNA in human prostate. J. Urol. 2239–2246 (2004).
6. Lundwall, A., Clauss, A. & Olsson, A. Y. Evolution of 144, 1550–1556 (1990). Unique study showing cancer incidence among
kallikrein-related peptidases in mammals and 13. Ung, J. O., Richie, J. P., Chen, M. H., Renshaw, A. A. & men of ages 62–91 with PSA levels of 0–4 ng/ml,
identification of a genetic locus encoding potential D’Amico, A. V. Evolution of the presentation and based on end-of-study biopsies of the men
regulatory inhibitors. Biol. Chem. 387, 243–249 pathologic and biochemical outcomes after radical randomized to the control arm of the Prostate
(2006). prostatectomy for patients with clinically localized Cancer Prevention Trial.
7. Olsson, A. Y., Lilja, H. & Lundwall, A. Taxon-specific prostate cancer diagnosed during the PSA era. 18. Ulmert, D. et al. Long-term prediction of prostate
evolution of glandular kallikrein genes and Urology 60, 458–463 (2002). cancer: PSA velocity is predictive but does not

276 | april 2008 | volume 8 www.nature.com/reviews/cancer

© 2008 Nature Publishing Group
 REVIEWS

improve the predictive accuracy of a single PSA involvement in the radical prostatectomy specimen This thorough meta-analysis demonstrates the
measurement 15 years or more before cancer among men with an elevated PSA. Urol. Oncol. 7, improvement in diagnostic performance gained by
diagnosis in a large, representative, unscreened 195–198 (2002). measuring the percentage of free PSA in addition
population. J. Clin. Oncol. 26, 835–841 (2008). 38. Stamey, T. A. et al. Prostate-specific antigen as a to tPSA.
19. Andriole, G. L. et al. Prostate cancer screening in the serum marker for adenocarcinoma of the prostate. 56. Stephan, C., Lein, M., Jung, K., Schnorr, D. & Loening,
prostate, lung, colorectal and ovarian (PLCO) cancer N. Engl. J. Med. 317, 909–916 (1987). S. A. The influence of prostate volume on the ratio of
screening trial: findings from the initial screening This classic paper demonstrated that PSA is free to total prostate specific antigen in serum of
round of a randomized trial. J. Natl Cancer Inst. 97, elevated in blood of patients with prostate cancer patients with prostate carcinoma and benign prostate
433–438 (2005). or BPH, and that it is associated with clinical stage hyperplasia. Cancer 79, 104–109 (1997).
20. Crawford, E. D. et al. Serum prostate-specific antigen and tumor volume. 57. Peter, J., Unverzagt, C., Krogh, T. N., Vorm, O. &
and digital rectal examination for early detection of 39. Pinsky, P. F. et al. Prostate-specific antigen velocity Hoesel, W. Identification of precursor forms of free
prostate cancer in a national community-based and prostate cancer gleason grade and stage. Cancer prostate-specific antigen in serum of prostate cancer
program. The Prostate Cancer Education Council. 109, 1689–1695 (2007). patients by immunosorption and mass spectrometry.
Urology 47, 863–869 (1996). 40. Hoedemaeker, R. F. et al. Pathologic features of Cancer Res. 61, 957–962 (2001).
21. Hugosson, J., Aus, G., Lilja, H., Lodding, P. & Pihl, prostate cancer found at population-based screening 58. Catalona, W. J. et al. Serum pro prostate specific
C. G. Results of a randomized, population-based study with a four-year interval. J. Natl Cancer Inst. 93, antigen improves cancer detection compared to free
of biennial screening using serum prostate-specific 1153–1158 (2001). and complexed prostate specific antigen in men with
antigen measurement to detect prostate carcinoma. 41. Sakr, W. A. et al. Age and racial distribution of prostate specific antigen 2 to 4 ng/ml. J. Urol. 170,
Cancer 100, 1397–1405 (2004). prostatic intraepithelial neoplasia. Eur. Urol. 30, 2181–2185 (2003).
22. Thompson, I. M. et al. Effect of finasteride on the 138–144 (1996). 59. Mikolajczyk, S. D. et al. Proenzyme forms of prostate-
sensitivity of PSA for detecting prostate cancer. J. Natl 42. Jemal, A. et al. Cancer statistics, 2007. CA Cancer J. specific antigen in serum improve the detection of
Cancer Inst. 98, 1128–1133 (2006). Clin. 57, 43–66 (2007). prostate cancer. Clin. Chem. 50, 1017–1025 (2004).
23. Thompson, I. M. et al. Operating characteristics of 43. Telesca, D., Etzioni, R. & Gulati, R. Estimating lead time 60. Mikolajczyk, S. D. et al. A precursor form of prostate-
prostate-specific antigen in men with an initial PSA and overdiagnosis associated with PSA screening from specific antigen is more highly elevated in prostate
level of 3.0 ng/ml or lower. JAMA 294, 66–70 (2005). prostate cancer incidence trends. Biometrics 14 May cancer compared with benign transition zone prostate
24. Thompson, I. M. et al. Assessing prostate cancer risk: 2007 (doi:10.1111/j.1541–0420.2007.00825.x). tissue. Cancer Res. 60, 756–759 (2000).
results from the Prostate Cancer Prevention Trial. This study provides estimates of lead time and 61. Nurmikko, P., Pettersson, K., Piironen, T., Hugosson, J.
J. Natl Cancer Inst. 98, 529–34 (2006). rates of overdiagnosis of prostate cancer resulting & Lilja, H. Discrimination of prostate cancer from benign
Describes an online calculator that estimates a from PSA testing, based on trends in the incidence disease by plasma measurement of intact, free prostate-
man’s risk of prostate cancer from his PSA level, of prostate cancer. specific antigen lacking an internal cleavage site at
age, race, family history of prostate cancer, results 44. Postma, R. et al. Cancer detection and cancer Lys145-Lys146. Clin. Chem. 47, 1415–1423 (2001).
of digital rectal examination and any prior biopsy characteristics in the European Randomized Study of 62. Steuber, T. et al. Association of free-prostate specific
results. Screening for Prostate Cancer (ERSPC) — Section antigen subfractions and human glandular kallikrein 2
25. Marks, L. S., Andriole, G. L., Fitzpatrick, J. M., Rotterdam. A comparison of two rounds of screening. with volume of benign and malignant prostatic tissue.
Schulman, C. C. & Roehrborn, C. G. The interpretation Eur. Urol. 52, 89–97 (2007). Prostate 63, 13–18 (2005).
of serum prostate specific antigen in men receiving 45. Bill-Axelson, A. et al. Radical prostatectomy versus 63. Slawin, K. M., Shariat, S. & Canto, E. BPSA: A novel
5α-reductase inhibitors: a review and clinical watchful waiting in early prostate cancer. N. Engl. J. serum marker for benign prostatic hyperplasia. Rev.
recommendations. J. Urol. 176, 868–874 (2006). Med. 352, 1977–1984 (2005). Urol. 7 (Suppl. 8), 52–56 (2005).
26. D’Amico, A. V. & Roehrborn, C. G. Effect of 1 mg/day 46. Aus, G., Bergdahl, S., Lodding, P., Lilja, H. & 64. Carter, H. B. et al. Longitudinal evaluation of prostate-
finasteride on concentrations of serum prostate- Hugosson, J. Prostate cancer screening decreases the specific antigen levels in men with and without
specific antigen in men with androgenic alopecia: absolute risk of being diagnosed with advanced prostate disease. JAMA 267, 2215–2220 (1992).
a randomised controlled trial. Lancet Oncol. 8, 21–25 prostate cancer--results from a prospective, 65. D’Amico, A. V., Chen, M. H., Roehl, K. A. & Catalona,
(2007). population-based randomized controlled trial. Eur. W. J. Preoperative PSA velocity and the risk of death
27. Thompson, I. M. et al. Prediction of prostate cancer Urol. 51, 659–664 (2007). from prostate cancer after radical prostatectomy.
for patients receiving finasteride: results from the Early results from a randomized trial of PSA N. Engl. J. Med. 351, 125–135 (2004).
Prostate Cancer Prevention Trial. J. Clin. Oncol. 25, screening, showing that screening reduced both 66. D’Amico, A. V., Renshaw, A. A., Sussman, B. & Chen,
3076–3081 (2007). frequency and absolute number of prostate cancer M. H. Pretreatment PSA velocity and risk of death
28. Verhamme, K. M. et al. Incidence and prevalence of cases that were metastatic at diagnosis. from prostate cancer following external beam
lower urinary tract symptoms suggestive of benign 47. Karazanashvili, G. & Abrahamsson, P. A. Prostate radiation therapy. JAMA 294, 440–447 (2005).
prostatic hyperplasia in primary care — the Triumph specific antigen and human glandular kallikrein 2 in 67. D’Amico, A. V. et al. Predictors of mortality after
project. Eur. Urol. 42, 323–328 (2002). early detection of prostate cancer. J. Urol. 169, prostate-specific antigen failure. Int. J. Radiat. Oncol.
29. Morgan, T. O. et al. Age-specific reference ranges for 445–57 (2003). Biol. Phys. 65, 656–660 (2006).
prostate-specific antigen in black men. N. Engl. J. 48. Ohori, M., Dunn, J. K. & Scardino, P. T. Is prostate- 68. Leibovici, D. et al. Prostate cancer progression in the
Med. 335, 304–310 (1996). specific antigen density more useful than prostate-specific presence of undetectable or low serum prostate-
30. Oesterling, J. E. et al. Free, complexed and total antigen levels in the diagnosis of prostate cancer? specific antigen level. Cancer 109, 198–204 (2007).
serum prostate specific antigen: the establishment of Urology 46, 666–671 (1995). 69. Nishio, R., Furuya, Y., Nagakawa, O. & Fuse, H.
appropriate reference ranges for their concentrations 49. Lilja, H. et al. Prostate-specific antigen in serum occurs Metastatic prostate cancer with normal level of serum
and ratios. J. Urol. 154, 1090–1095 (1995). predominantly in complex with α1-antichymotrypsin. prostate-specific antigen. Int. Urol. Nephrol. 35,
31. Borer, J. G., Sherman, J., Solomon, M. C., Plawker, Clin. Chem. 37, 1618–1625 (1991). 189–192 (2003).
M. W. & Macchia, R. J. Age specific prostate specific 50. Stenman, U. H. et al. A complex between prostate- 70. Stephenson, A. J. et al. Defining biochemical
antigen reference ranges: population specific. J. Urol. specific antigen and α 1-antichymotrypsin is the major recurrence of prostate cancer after radical
159, 444–448 (1998). form of prostate-specific antigen in serum of patients prostatectomy: a proposal for a standardized
32. Fang, J. et al. Low levels of prostate-specific antigen with prostatic cancer: assay of the complex improves definition. J. Clin. Oncol. 24, 3973–3978 (2006).
predict long-term risk of prostate cancer: results from clinical sensitivity for cancer. Cancer Res. 51, 222–226 71. Amling, C. L., Bergstralh, E. J., Blute, M. L., Slezak,
the Baltimore Longitudinal Study of Aging. Urology (1991). J. M. & Zincke, H. Defining prostate specific antigen
58, 411–416 (2001). 51. Christensson, A. et al. Serum prostate specific progression after radical prostatectomy: what is the
33. Loeb, S. et al. Baseline prostate-specific antigen antigen complexed to α 1-antichymotrypsin as an most appropriate cut point? J. Urol. 165, 1146–1151
compared with median prostate-specific antigen for indicator of prostate cancer. J. Urol. 150, 100–105 (2001).
age group as predictor of prostate cancer risk in men (1993). 72. Vaisanen, V., Peltola, M. T., Lilja, H., Nurmi, M. &
younger than 60 years old. Urology 67, 316–320 52. Catalona, W. J. et al. Use of the percentage of free Pettersson, K. Intact free prostate-specific antigen and
(2006). prostate-specific antigen to enhance differentiation of free and total human glandular kallikrein 2.
34. Lilja, H. et al. Long-term prediction of prostate cancer prostate cancer from benign prostatic disease: a Elimination of assay interference by enzymatic
up to 25 years before diagnosis of prostate cancer prospective multicenter clinical trial. JAMA 279, digestion of antibodies to F(ab′)2 fragments. Anal.
using prostate kallikreins measured at age 44 to 50 1542–1547 (1998). Chem. 78, 7809–7815 (2006).
years. J. Clin. Oncol. 25, 431–436 (2007). This report, based on a large cohort, was the first 73. Diamandis, E. P. & Yu, H. Nonprostatic sources of
35. Ulmert, D. et al. Prostate-specific antigen at or before to propose a cut-off for the percentage of free PSA prostate-specific antigen. Urol. Clin. North Am. 24,
age 50 as a predictor of prostate cancer of as criterion for prostate biopsy. 275–282 (1997).
unquestionable significance up to 25 years later: a 53. Ulmert, D. et al. Reproducibility and accuracy of 74. Kupelian, P. et al. Improved biochemical relapse-free
case-control study. BMC Med. 6, 6 (2008). measurements of free and total prostate-specific survival with increased external radiation doses in
This paper shows that a single PSA test taken at antigen in serum vs plasma after long-term storage at patients with localized prostate cancer: the combined
or before age 50 is a strong predictor of advanced –20 degrees C. Clin. Chem. 52, 235–239 (2006). experience of nine institutions in patients treated in
prostate cancer diagnosed up to 25 years later, 54. Morote, J. et al. The percentage of free prostatic- 1994 and 1995. Int. J. Radiat. Oncol. Biol. Phys. 61,
with advanced cancer defined as cancer that is specific antigen is also useful in men with normal 415–419 (2005).
locally advanced or metastatic at the time of digital rectal examination and serum prostatic-specific This study showed that PSA level after radiation
diagnosis. antigen between 10.1 and 20 ng/ml. Eur. Urol. 42, therapy reflects the radiation dose, which is in turn
36. Vickers, A. J. et al. The predictive value of prostate 333–337 (2002). associated with relapse-free survival.
cancer biomarkers depends on age and time to 55. Roddam, A. W. et al. Use of prostate-specific antigen 75. Miller, N., Smolkin, M. E., Bissonette, E. &
diagnosis: towards a biologically-based screening (PSA) isoforms for the detection of prostate cancer in Theodorescu, D. Undetectable prostate specific
strategy. Int. J. Cancer 121, 2212–2217 (2007). men with a PSA level of 2–10 ng/ml: systematic review antigen at 6–12 months: a new marker for early
37. Grossklaus, D. J. et al. The free/total prostate-specific and meta-analysis. Eur. Urol. 48, 386–399; success in hormonally treated patients after prostate
antigen ratio (%fPSA) is the best predictor of tumor discussion 398–399 (2005). brachytherapy. Cancer 103, 2499–2506 (2005).

nature reviews | cancer volume 8 | april 2008 | 277

© 2008 Nature Publishing Group
REVIEWS

76. Hanlon, A. L., Moore, D. F. & Hanks, G. E. Modeling therapy of metastatic prostate cancer. J. Urol. 147, prostate-specific antigen and α2-macroglobulin. Clin.
postradiation prostate specific antigen level kinetics: 956–961 (1992). Chem. 44, 2471–2479 (1998).
predictors of rising postnadir slope suggest cure in 95. Ryan, C. J. et al. Persistent prostate-specific antigen 117. Lane, J. A. et al. Detection of prostate cancer in
men who remain biochemically free of prostate expression after neoadjuvant androgen depletion: an unselected young men: prospective cohort nested
carcinoma. Cancer 83, 130–134 (1998). early predictor of relapse or incomplete androgen within a randomised controlled trial. BMJ 335, 1139
77. Ray, M. E. et al. PSA nadir predicts biochemical and suppression. Urology 68, 834–839 (2006). (2007).
distant failures after external beam radiotherapy for 96. Scher, H. I. & Sawyers, C. L. Biology of progressive, 118. Donovan, J. et al. Prostate Testing for Cancer and
prostate cancer: a multi-institutional analysis. Int. J. castration-resistant prostate cancer: directed Treatment (ProtecT) feasibility study. Health Technol.
Radiat. Oncol. Biol. Phys. 64, 1140–1150 (2006). therapies targeting the androgen-receptor signaling Assess 7, 1–88 (2003).
78. Critz, F. A. et al. Post-treatment PSA ≤0.2 ng/mL axis. J. Clin. Oncol. 23, 8253–8261 (2005). 119. Shibata, A. & Whittemore, A. S. Re: Prostate cancer
defines disease freedom after radiotherapy for 97. Arai, Y., Yoshiki, T. & Yoshida, O. Prognostic incidence and mortality in the United States and the
prostate cancer using modern techniques. Urology significance of prostate specific antigen in endocrine United Kingdom. J. Natl Cancer Inst. 93, 1109–1110
54, 968–971 (1999). treatment for prostatic cancer. J. Urol. 144, (2001).
79. Cavanagh, W., Blasko, J. C., Grimm, P. D. & 1415–1419 (1990). 120. Oliver, S. E., Gunnell, D. & Donovan, J. L. Comparison
Sylvester, J. E. Transient elevation of serum prostate- 98. Furuya, Y. et al. Prognostic significance of changes in of trends in prostate-cancer mortality in England and
specific antigen following 125I/103Pd brachytherapy for prostate-specific antigen in patients with metastasis Wales and the USA. Lancet 355, 1788–1789 (2000).
localized prostate cancer. Semin. Urol. Oncol. 18, prostate cancer after endocrine treatment. Int. Urol. 121. Roehl, K. A. et al. Survival results in patients with
160–165 (2000). Nephrol 32, 659–663 (2001). screen-detected prostate cancer versus physician-
80. Reed, D., Wallner, K., Merrick, G., Buskirk, S. & True, 99. Matzkin, H., Eber, P., Todd, B., van der Zwaag, R. & referred patients treated with radical prostatectomy:
L. Clinical correlates to PSA spikes and positive repeat Soloway, M. S. Prognostic significance of changes in early results. Urol. Oncol. 24, 465–471 (2006).
biopsies after prostate brachytherapy. Urology 62, prostate-specific markers after endocrine treatment of 122. Oberaigner, W. et al. Reduction of prostate cancer
683–688 (2003). stage D2 prostatic cancer. Cancer 70, 2302–2309 mortality in Tyrol, Austria, after introduction of
81. Crook, J. et al. PSA kinetics and PSA bounce following (1992). prostate-specific antigen testing. Am. J. Epidemiol.
permanent seed prostate brachytherapy. Int. J. 100. Zanetti, G. et al. Prognostic significance of prostate- 164, 376–384 (2006).
Radiat. Oncol. Biol. Phys. 69, 426–433 (2007). specific antigen in endocrine treatment for prostatic 123. Shaw, P. A. et al. An ecologic study of prostate-specific
82. Patel, C. et al. PSA bounce predicts early success in carcinoma. Eur. Urol. 21 (Suppl. 1), 96–98 (1992). antigen screening and prostate cancer mortality in
patients with permanent iodine-125 prostate implant. 101. Furuya, Y., Nagakawa, O. & Fuse, H. Prognostic nine geographic areas of the United States. Am. J.
Urology 63, 110–113 (2004). significance of changes in short-term prostate volume Epidemiol. 160, 1059–1069 (2004).
83. Toledano, A. et al. PSA bounce after permanent and serum prostate-specific antigen after androgen 124. Kopec, J. A. et al. Screening with prostate specific
implant prostate brachytherapy may mimic a withdrawal in men with metastatic prostate cancer. antigen and metastatic prostate cancer risk: a
biochemical failure: a study of 295 patients with a Urol. Int. 70, 195–199 (2003). population based case-control study. J. Urol. 174,
minimum 3-year followup. Brachytherapy 5, 122–126 102. Morote, J., Trilla, E., Esquena, S., Abascal, J. M. & 495–499; discussion 499 (2005).
(2006). Reventos, J. Nadir prostate-specific antigen best 125. Concato, J. et al. The effectiveness of screening for
84. Hanlon, A. L., Pinover, W. H., Horwitz, E. M. & Hanks, predicts the progression to androgen-independent prostate cancer: a nested case-control study. Arch.
G. E. Patterns and fate of PSA bouncing following 3D- prostate cancer. Int. J. Cancer 108, 877–881 (2004). Intern. Med. 166, 38–43 (2006).
CRT. Int. J. Radiat. Oncol. Biol. Phys. 50, 845–849 103. Armstrong, A. J. et al. Prostate-specific antigen and pain 126. Weinmann, S. et al. Screening by prostate-specific
(2001). surrogacy analysis in metastatic hormone-refractory antigen and digital rectal examination in relation to
85. Horwitz, E. M. et al. Biochemical and clinical prostate cancer. J. Clin. Oncol. 25, 3965–3970 (2007). prostate cancer mortality: a case-control study.
significance of the posttreatment prostate-specific 104. Bjork, T. et al. Rapid exponential elimination of free Epidemiology 16, 367–376 (2005).
antigen bounce for prostate cancer patients treated prostate-specific antigen contrasts the slow, 127. Lu-Yao, G. et al. Natural experiment examining impact
with external beam radiation therapy alone: a capacity-limited elimination of PSA complexed to of aggressive screening and treatment on prostate
multiinstitutional pooled analysis. Cancer 107, α1-antichymotrypsin from serum. Urology 51, cancer mortality in two fixed cohorts from Seattle area
1496–1502 (2006). 57–62 (1998). and Connecticut. BMJ 325, 740 (2002).
86. Rosser, C. J. et al. Is patient age a factor in the 105. Becker, C., Piironen, T., Pettersson, K., Hugosson, J. & 128. Ciatto, S. et al. Prostate cancer specific mortality in
occurrence of prostate-specific antigen bounce Lilja, H. Testing in serum for human glandular the Florence screening pilot study cohort 1992–
phenomenon after external beam radiotherapy for kallikrein 2, and free and total prostate specific 1993. Eur. J. Cancer 42, 1858–1862 (2006).
prostate cancer? Urology 66, 327–331 (2005). antigen in biannual screening for prostate cancer. 129. Coldman, A. J., Phillips, N. & Pickles, T. A. Trends in
87. Thames, H. et al. Comparison of alternative J. Urol. 170, 1169–1174 (2003). prostate cancer incidence and mortality: an analysis of
biochemical failure definitions based on clinical 106. Raaijmakers, R. et al. hK2 and free PSA, a prognostic mortality change by screening intensity. CMAJ 168,
outcome in 4839 prostate cancer patients treated by combination in predicting minimal prostate cancer in 31–35 (2003).
external beam radiotherapy between 1986 and screen-detected men within the PSA Range 4–10 ng/ml. 130. Labrie, F. et al. Screening decreases prostate cancer
1995. Int. J. Radiat. Oncol. Biol. Phys. 57, 929–943 Eur. Urol. 52, 1358–1364 (2007). mortality: 11-year follow-up of the 1988 Quebec
(2003). 107. Steuber, T. et al. Comparison of free and total forms of prospective randomized controlled trial. Prostate 59,
88. Roach, M. 3rd et al. Defining biochemical failure serum human kallikrein 2 and prostate-specific 311–318 (2004).
following radiotherapy with or without hormonal antigen for prediction of locally advanced and 131. Sandblom, G., Varenhorst, E., Lofman, O., Rosell, J. &
therapy in men with clinically localized prostate recurrent prostate cancer. Clin. Chem. 53, 233–240 Carlsson, P. Clinical consequences of screening for
cancer: recommendations of the RTOG-ASTRO (2007). prostate cancer: 15 years follow-up of a randomised
Phoenix Consensus Conference. Int. J. Radiat. Oncol. 108. Denmeade, S. R. et al. Prostate-specific antigen- controlled trial in Sweden. Eur. Urol. 46, 717–723;
Biol. Phys. 65, 965–974 (2006). activated thapsigargin prodrug as targeted therapy for discussion 724 (2004).
89. Horwitz, E. M. et al. Definitions of biochemical failure prostate cancer. J. Natl Cancer Inst. 95, 990–1000 132. van der Cruijsen-Koeter, I. W. et al. Comparison of
that best predict clinical failure in patients with (2003). screen detected and clinically diagnosed prostate
prostate cancer treated with external beam radiation 109. Fowler, J. E., Bigler, S. A., Kilambi, N. K. & Land, S. A. cancer in the European randomized study of screening
alone: a multi-institutional pooled analysis. J. Urol. Relationships between prostate-specific antigen and for prostate cancer, section Rotterdam. J. Urol. 174,
173, 797–802 (2005). prostate volume in black and white men with benign 121–125 (2005).
A demonstration that the Phoenix definition of prostate biopsies. Urology 53, 1175–1178 (1999).
biochemical recurrence after radiation therapy (a 110. Moul, J. W. et al. Prostate-specific antigen values at Acknowledgements
rise of 2 ng/ml above PSA nadir) is a better the time of prostate cancer diagnosis in African- We thank A. M. Cronin of Memorial Sloan–Kettering Cancer
predictor of clinical failure and of metastases than American men. JAMA 274, 1277–1281 (1995). Center for compiling data on the distribution of PSA levels.
is the ASTRO definition. 111. Catalona, W. J. et al. Percentage of free PSA in black We also thank J. Novak of Helix Editing for assistance with
90. Kuban, D. A. et al. Comparison of biochemical failure versus white men for detection and staging of prostate writing the manuscript. She was paid for her work by
definitions for permanent prostate brachytherapy. Int. cancer: a prospective multicenter clinical trial. Urology Memorial Sloan–Kettering Cancer Center.
J. Radiat. Oncol. Biol. Phys. 65, 1487–1493 (2006). 55, 372–376 (2000).
91. Stamey, T. A., Kabalin, J. N., Ferrari, M. & Yang, N. 112. Stanford, J. L. et al. Prostate Cancer Trends Competing interests statement
Prostate specific antigen in the diagnosis and treatment 1973–1995 (SEER Program; National Cancer The authors declare competing financial interests: see web
of adenocarcinoma of the prostate. IV. Anti-androgen Institute, Bethesda, 1999). version for details.
treated patients. J. Urol. 141, 1088–1090 (1989). 113. Mettlin, C. J., Murphy, G. P., Ho, R. & Menck, H. R.
92. Benaim, E. A., Pace, C. M., Lam, P. M. & Roehrborn, The National Cancer Data Base report on DATABASES
C. G. Nadir prostate-specific antigen as a predictor of longitudinal observations on prostate cancer. Cancer Entrez Gene: http://www.ncbi.nlm.nih.gov/entrez/query.
progression to androgen-independent prostate cancer. 77, 2162–2166 (1996). fcgi?db=gene
Urology 59, 73–78 (2002). 114. Finne, P. et al. Use of the complex between prostate KLK2 | KLK3 | SEMG1 | SEMG2 | SERPINA2
93. Kwak, C., Jeong, S. J., Park, M. S., Lee, E. & Lee, S. E. specific antigen and α 1-protease inhibitor for National Cancer Institute: http://www.cancer.gov/
Prognostic significance of the nadir prostate specific screening prostate cancer. J. Urol. 164, 1956–1960 prostate cancer
antigen level after hormone therapy for prostate (2000). National Cancer Institute Drug Dictionary:
cancer. J. Urol. 168, 995–1000 (2002). 115. Lilja, H. et al. Significance and metabolism of http://www.cancer.gov/drugdictionary/
This study demonstrated that PSA nadir after complexed and noncomplexed prostate specific dutasteride | finasteride
initiation of hormonal therapy is an important antigen forms, and human glandular kallikrein 2 in
predictor of both time to hormone-refractory clinically localized prostate cancer before and after FURTHER INFORMATION
disease and survival. radical prostatectomy. J. Urol. 162, 2029–2034; Hans Lilja’s homepage:
94. Miller, J. I., Ahmann, F. R., Drach, G. W., Emerson, discussion 2034–2035 (1999). http://www.mskcc.org/prg/prg/bios/874.cfm
S. S. & Bottaccini, M. R. The clinical usefulness of 116. Zhang, W. M. et al. Characterization and
All links are active in the online pdf
serum prostate specific antigen after hormonal immunological determination of the complex between

278 | april 2008 | volume 8 www.nature.com/reviews/cancer

© 2008 Nature Publishing Group
Corrigendum

Prostate-specific antigen and prostate cancer: prediction, detection

and monitoring (Corrigendum)
Hans Lilja, David Ulmert and Andrew J. Vickers
Nature Reviews Cancer 8, 268–278 (2008)
There is an error in the legend of FIG. 4 on page 275 of this article. The hk2 level in seminal fluid is given incorrectly;
it should read 2–12 µg/ml.

© 2008 Nature Publishing Group