SEARCH

SEARCH BY CITATION

Keywords:

  • prostate-specific antigen;
  • advanced prostate cancer;
  • patient management;
  • androgen-deprivation therapy

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References
  • To review current prostate-specific antigen (PSA) metrics used in monitoring treatment of advanced prostate cancer, with a specific focus on castration-resistant prostate cancer (CRPC) therapies.
  • Explore what is known about the correlation between PSA and androgen levels as well as underlying reasons for persistent PSA expression and serum elevation in CRPC, and outline suggestions for use of PSA in managing patients with advanced prostate cancer.
  • A comprehensive search of the PubMed database for English language articles through April 2012 was performed using the following Medical Subject Headings (MeSH) keywords or terms, alone or in combination: ‘prostate cancer’; ‘prostate cancer treatment’; ‘prostate cancer outcomes’; ‘prostate-specific antigen’; ‘androgen receptor’; ‘advanced prostate cancer’; ‘castration-resistant prostate cancer’; ‘biomarkers’.
  • Bibliographies of relevant articles were searched for additional references.
  • Relevant medical society and regulatory agency web sites from the USA and Europe were accessed for issued guidance on PSA use.
  • PSA doubling time (PSADT) is a useful metric for determining which patients should be considered for androgen-deprivation therapy (ADT) after failing local treatment or for second-line therapies after failing ADT. However, it is not a validated surrogate for survival and no therapy has received regulatory approval based upon PSADT characteristics.
  • PSA nadir and time-to-nadir have been identified as possible prognostic markers for patients receiving ADT.
  • There is no universally accepted definition for PSA progression, nor is PSA progression a regulatory-approved surrogate for clinical progression in drug approval trials.
  • PSA responses to second-line therapies can vary and are not considered by regulatory agencies as valid surrogates for clinical endpoints, so they must be assessed in the context of each individual therapy and trial design.
  • PSA expression in CRPC is often a reflection of persistent androgen receptor activity.
  • While we can provide guidance for use of PSA monitoring in managing patients with advanced prostate cancer based on the data at hand, there is an urgent need for prospective analyses of refined PSA metrics in conjunction with newer prostate cancer biomarkers in clinical trials to provide stronger evidence for their roles as surrogate endpoints.

Abbreviations
ACM

all-cause mortality

ADT

androgen-deprivation therapy

AR

androgen receptor

CRPC

castration-resistant prostate cancer

DHT

dihydrotestosterone

EAU

European Association of Urology

FDA

US Food and Drug Administration

NCCN

National Comprehensive Cancer Network

OS

overall survival

PCSM

prostate cancer-specific mortality

PSADT

PSA doubling time

PCWG

Prostate Cancer Clinical Trials Working Group

RP

radical prostatectomy

RT

radiation therapy

SWOG

Southwest Oncology Group

TTN

time-to-nadir

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

Since its identification in 1969 and characterisation in the 1980s, PSA has become the most widely used molecular marker for the diagnosis and management of a malignancy [1-4]. Despite its widespread adoption for early detection of prostate cancer beginning in 1986, PSA screening has become increasingly controversial for several reasons: the morbidity associated with treatment of prostate cancer, an increasing recognition that the cancer is more often than not unrelated to the cause of death, the heterogeneity of tumours, and, for most men, an indolent natural history of prostate cancer [5, 6]. While PSA screening for prostate cancer detection is increasingly debated, particularly after the US Public Health Services Task Force recommendation against PSA screening, PSA is an increasingly important tool for monitoring patients with prostate cancer, a use that aligns with the initial clinical indication for PSA.

As new therapies are developed for treating advanced prostate cancer and castration-resistant prostate cancer (CRPC), understanding the role of PSA measurements in assessing treatment efficacy, as a prognostic marker, and as a tool for identifying appropriate subjects for clinical trials becomes more important. This review is based on an expert roundtable discussion among urologists and medical oncologists which revealed that, despite routine use of PSA among men with prostate cancer, there are many gaps in the knowledge and applications surrounding PSA as a biomarker for clinical patient management. The objectives of the present article were to review current PSA metrics used in advanced prostate cancer, discuss PSA responses in CRPC therapies, explore what is known about the correlation between PSA and androgen levels as well as the underlying reasons for persistent PSA expression and serum elevation in CRPC, and put forth suggestions for appropriate use of PSA in managing patients with advanced prostate cancer.

Data Acquisition

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

We comprehensively searched the PubMed database for English language articles through April 2012 using the following Medical Subject Headings (MeSH) keywords or terms, alone or in combination: ‘prostate cancer’; ‘prostate cancer treatment’; ‘prostate cancer outcomes’; ‘prostate-specific antigen’; ‘androgen receptor’; ‘advanced prostate cancer’; ‘castration-resistant prostate cancer’; ‘biomarkers’. In addition, bibliographies of relevant articles were searched for additional references. Relevant medical society and regulatory agency web sites from the USA and Europe were accessed for issued guidance on PSA use.

PSA Metrics in Advanced Prostate Cancer

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

PSA Monitoring

Routinely monitoring PSA levels in patients with prostate cancer treated with localised therapies for curative intent (i.e. retropubic radical prostatectomy [RP], radiation therapy [RT]) is a key tool in detecting disease recurrence and is recommended by the AUA [7], the National Comprehensive Cancer Network (NCCN) [8], and the European Association of Urology (EAU) [9]. The AUA PSA Best Practice Statement does not provide recommendations about the timing of PSA measurements after local therapies [7]; however, the NCCN recommends PSA measurements every 6–12 months for the first 5 years after therapy, then annually, along with DREs if there are detectable PSA levels [8]. EAU guidelines also recommend PSA and DRE assessments be performed at 3, 6, and 12 months after treatment, then every 6 months through the third year, then annually [9].

For patients receiving androgen-deprivation therapy (ADT) to treat metastatic prostate cancer, regular PSA assessments, along with serum testosterone measurements, aid in monitoring treatment response and detecting disease progression despite castrate testosterone levels. Any rise in PSA levels after a nadir has been reached is often the first indication of the development of metastatic CRPC prior to any evidence of clinical progression (with the exception of post-RT, where PSA ‘bounces’ are not uncommon) [10, 11]. Therefore, the ability to identify patients early who are likely to progress and should be considered for enrolment in clinical trials may prove clinically beneficial.

Data collected from regular PSA measurements, whether in patients who have received localised therapies or in those receiving ADT, can be used in different PSA metrics, as described below. Some of these metrics have been incorporated into clinical trial designs as secondary endpoints or inclusion criteria for enrolment, or have been found to be statistically associated with clinical outcomes. However, no PSA metric has been approved by a regulatory agency as an acceptable surrogate for clinical endpoints.

PSA Doubling Time (PSADT)

PSADT is a concept that has been evaluated in several circumstances to aid in clinical decision making. However, there is no standardisation of this calculation; multiple means of calculating PSADT have been reported and the use of different calculation methods can provide differing values for the same patient [12, 13]. These formulations can differ by what is considered the lowest PSA to be used in the calculation, the number of PSA values used, and the duration between PSA measures. One of the more prevalent post-treatment PSADT calculations is the natural log of 2 divided by the slope of the linear regression line of the natural log of PSA against time (months), with the post-treatment nadir as the lowest value, at least three PSA measures, and PSA measures taken at least 3 months apart [14]. There are several calculator tools available online, but similar to those available in the literature, resulting PSADT values using the same patient data may differ. The use of this concept has not been validated as a surrogate for other aspects of prostate cancer biology. For this PSA metric to be validated and adopted as a clinical management tool, the PSADT calculations need to be standardised. That notwithstanding, PSADT reflects the rate of PSA increase as a function of time and may provide some information along with other parameters in determining which patients are more likely to clinically progress to metastatic disease or die from prostate cancer [15].

More rapid PSADT after RP or RT is strongly associated with increased risk of metastasis, all-cause mortality (ACM), and prostate cancer-specific mortality (PCSM) [16-19]. The NCCN and EAU recommend that patients with a shorter PSADT after definitive treatment be encouraged to consider early ADT [8, 20]. Among patients who have evidence of disease persistence after localised therapies, PSADT before initiation of ADT has been identified as an independent risk factor for PCSM at thresholds of ≤3 months [21], ≤6 months [22], and as a continuous variable [23]. The importance of PSADT was also identified as one of several metrics that should be considered for determining the timing of initiation of ADT in patients who biochemically relapse after definitive localised therapy [19]. To minimise the development of non-metastatic CRPC, a recent US Food and Drug Administration (FDA) Panel advised against indiscriminately administering ADT in men with very prolonged PSADT after biochemical relapse [24, 25]. To date, there is no level I evidence to support early initiation of ADT after curative localised treatment in asymptomatic patients with only rising PSA levels. Well-powered randomised trials are needed to determine whether the early application of androgen blockade will improve survival and delay the development of metastatic disease. For patients who have progressed to CRPC after ADT, more rapid PSADT before docetaxel chemotherapy is a significant independent risk factor for death [26-29]. As such, patients with CRPC with rapid PSADT should be considered for enrolment in clinical trials. Despite the utility described above, PSADT has not been validated or approved as a surrogate endpoint for prostate cancer drug development.

Absolute PSA Levels

The value of absolute PSA levels before therapy as a prognostic marker is not clear, as the data are conflicting. Some report that PSA levels before the initiation of ADT, as a categorical and/or a continuous variable, are significantly associated with progression, PCSM, and ACM [21, 30-32]. In addition, a higher PSA level before ADT may be a significant factor in the failure to achieve a desired PSA nadir while on ADT [33-35]. However, in other studies, PSA levels at the time of ADT initiation were not found to be significant predictors of PSA progression [36], PCSM or ACM [37], or its predictive significance after univariate analyses was lost after multivariate regression analyses [22, 38-40]. Differences in the published data underscore the limited value of using absolute PSA values in isolation.

For patients who have progressed to metastatic CRPC, exploratory analysis of the relationship between PSA levels and the natural history of bone disease progression in patients not receiving cytotoxic chemotherapy found that higher PSA levels were correlated with an increased risk of all bone-related clinical outcomes, along with overall survival (OS) [41]. In addition, in men with progressive, nonmetastatic CRPC not receiving any secondary hormonal or cytotoxic chemotherapy, a higher PSA is associated with shorter time to first bone metastasis, OS, and metastasis-free survival [42].

PSA Nadir and Time-to-Nadir during ADT

PSA nadir achieved during ADT has been put forth as a prognostic indicator of clinical outcomes (Table 1) [21-23, 30, 32, 34-40, 43-50]. Although there is no absolute threshold recognised by any society or regulatory agency, the most commonly cited ADT-induced PSA nadir threshold associated with increased risk of PSA progression, PCSM, and/or ACM is 0.2 ng/mL [21, 22, 32, 34, 36, 39, 43, 45]. This has been shown in various hormone-naïve patient populations, including those receiving ADT for recurrence after RP or external-beam RT [21, 22] as primary therapy for localised or metastatic prostate cancer [32, 34, 39, 49], and in heterogeneous study cohorts [36, 43, 45]. Achieving castrate testosterone levels should be confirmed if a patient fails to achieve any pre-determined PSA nadir, as unsatisfactory PSA response to ADT may reflect a poor response to therapy and indicate the need to modify the type of hormonal suppression being used (i.e. switch LHRH agonist formulations, switch to an LHRH antagonist, or add an antiandrogen).

Table 1. Studies that have examined PSA nadir and TTN as markers for outcomes
Study; study typePatientsTreatmentPSA thresholdsOutcome association
Nadir, ng/mLTTN, months
  1. EBRT, external-beam RT; IADT, intermittent ADT; HR, hazard ratio; NR, not reported; OR, odds ratio.

Benaim et al. (2002) [43]; retrospective chart reviewN = 132; heterogeneous staging; 39% had prior RP or EBRTLHRH agonist or orchidectomy, with or without antiandrogen0.2No specific threshold given

OR 14.7 (95% CI 2.2–97.2; P = 0.005) for PSA progression within 24 months for patients with detectable nadir

Significant positive correlation (r = 0.35; P = 0.01) between shorter time to progression and faster TTN

Choueiri et al. (2009) [45]; retrospective chart reviewN = 179; hormone-naïve, metastatic; 47.5% had prior local therapyOrchidectomy or LHRH agonists with or without antiadrogen0.26In multivariate analysis, higher nadir (≥0.2 ng/mL) and shorter TTN (<6 months) were associated with shorter OS (P < 0.0001 and P = 0.005, respectively)
Chung et al. (2008) [30]; retrospective chart reviewN = 162; biochemical recurrence after RP or EBRT and did not reach nadir of <0.2 ng/mL after 8 months of ADTOrchidectomy or LHRH agonist with or without antiandrogen0.9 (among men who did not achieve nadir of <0.2)4

Nadir (continuous variable) was significantly associated with PCSM (HR 1.04; 95% CI 1.02–1.06; P < 0.001) and ACM (HR 1.02; 95% CI 1.01–1.04; P = 0.005)

Longer TTN (continuous variable) was predictive of greater risk of PCSM (HR 2.53; 95% CI 1.24–5.14; P = 0.01) but not ACM (HR 1.03; 95% CI 0.96–1.09; P = 0.41)

Nadir >0.9 ng/mL had significantly higher 5-year ACM rates compared with those with nadir ≤0.9 ng/mL regardless of TTN.

When nadir (>0.9 ng/mL) stratified by TTN, patients with slower TTN had significantly higher 5-year PCSM rates (P = 0.04)

D'Amico et al. (2007) [23]; retrospective database analysisN = 585; PSA recurrence after RP or RTLHRH agonist, bilateral orchidectomy, antiandrogen monotherapy, or combined blockadeAll men achieved PSA level of <0.2Continuous variableLonger TTN was significantly associated with shorter time to PCSM (HR 9.2; 95% CI 3.8–22.1; P < 0.001)

Hori et al. (2011) [49];

cohort study, two sites

Site 1: N = 155; localised or locally advanced, non-metastatic, n = 109; bone metastasis, n = 46 Site 2: N = 84 non-metastaticOrchidectomy or LHRH agonist with or without antiandrogen

Site 1: continuous

Site 2: 0.1

Site 1: continuous

Site 2: 24

Site 1: high PSA nadir (P ≤ 0.002) and shorter TTN (P < 0.001) were significant independent predictors for biochemical recurrence

Site 2: men with low PSA nadir had significantly lower rate of biochemical recurrence than those with a higher nadir (P = 0.005), as did those with longer TTN (P = 0.01) than those with a shorter TTN

Huang et al. (2011 and 2012) [36, 37]; prospective cohort studyN = 650; advanced or metastatic; 15% had recurrence after RP or RTOrchidectomy or LHRH agonist with or without antiandrogen0.210

Both higher nadir and shorter TTN were significant predictors of:

PSA progression (HR 2.38; 95% CI 19.1–2.97; P < 0.001 and HR 1.31; 95% CI 1.07–1.61; P = 0.009, respectively)PCSM (HR 3.64; 95% CI 2.08–6.37; P < 0.001 and HR 1.68; 95% CI 1.03–2.75; P = 0.039, respectively)

ACM (HR 2.83; 95% CI 1.80–4.45; P < 0.001 and HR 1.66; 95% CI 1.09–2.53; P = 0.019)

When combined effect of nadir and TTN analysed, TTN was only a significant predictor of any outcome in patients whose nadir was >0.2 ng/mL (P < 0.001 for all three outcomes)

Hussain et al. (2006) [34]; retrospective analysis of SWOG 9346, phase III study comparing IADT with continuous ADTN = 1395; newly diagnosed stage D27-month induction period with goserelin and bicalutamide0.2 and 4.0NR

Compared with patients with higher nadir >4.0 mg/mL, risk of death for those with lower nadir was much reduced

Nadir ≤4.0 to >2.0 ng/mL – HR 0.30; 95% CI 0.24–0.38

Nadir ≤0.2 ng/mL – HR 0.17; 95% CI 0.13–0.21

Keizman et al. (2011) [48]; retrospective study of case reportsN = 96; biochemical relapse after RP or RT, non-metastaticIADT; LHRH agonist with or without bicalutamide. Treatment stopped 6–9 months when PSA nadired0.1NRDuring first treatment cycle, nadir (<0.1 vs ≥0.1 ng/mL) was a significant predictor of PSA or clinical progression-free survival (HR 4.7; P = 0.002)
Kwak et al. (2002) [38]; retrospective chart reviewN = 177; stage C or D at diagnosisOrchidectomy or LHRH agonist with or without antiandrogen

<0.2

0.2–1.0

1.1–10.0

≥10.1

NRHigher nadir was a significant predictor of shorter time to progression to CRPC and shorter OS
Miyamoto et al. (2012) [50]; retrospective chart reviewN = 94; all metastatic at initial diagnosisLHRH agonist with delayed combined androgen blockade if PSA increased during LHRH agonist monotherapy.

≤0.3

≥0.4 to ≤4.5

≥4.6

NRLower nadir was a significant predictor of OS. HR 3.91 (95% CI 2.01–7.60; P < 0.001)
Morote et al. (2004) [40]; retrospective database analysisN = 283; 65% metastaticOrchidectomy or LHRH agonist with or without antiandrogen0.212Higher nadir and shorter TTN were significant predictors of shorter time to biochemical progression (OR 6.07; 95% CI 2.21–9.43; P = 0.001 and OR 3.11; 95% CI 2.21–4.39; P = 0.001, respectively)
Park et al. (2009) [46]; database analysisN = 131; newly diagnosed stage D2, metastaticLHRH agonist or orchidectomy, with or without antiandrogen0.9

(half-time to nadir)

1

Lower nadir and shorter half-time to were predictive of prostate cancer-specific survival (HR 1.038; 95% CI 1.022–1.054; P < 0.001 and HR 3.021; 95% CI 1.281–7.126; P = 0.012, respectively)
Park et al. (2009) [35]; retrospective chart reviewN = 231; newly diagnosed; 36% and 49% with lymph node and bone metastases, respectivelyLHRH agonist or orchidectomy plus antiandrogen1.0NRRate of progression-free survival was significantly higher in patients with lower nadir (P < 0.05)
Rodrigues et al. (2006) [22]; retrospective chart reviewN = 67; recurrence after RP or EBRTLHRH agonist with or without antiandrogen0.2continuous

Higher nadir was significantly associated with increased risk of PCSM (HR 8.0; 95% CI 1.7–38.7; P = 0.009)

TTN was not associated with risk of PCSM

Sasaki et al. (2011) [39]; retrospective chart reviewN = 87; metastaticOrchidectomy or LHRH agonist with antiandrogen0.26 and 12

HR for OS for nadir was 3.73 (95% CI 1.54–9.03; P = 0.003).

OS rate decreased significantly with decreasing TTN (P ≤ 0.002)

When nadir stratified by TTN, nadir did not influence OS when TTN was longer

Scholz et al. (2007) [44]; retrospective chart reviewN = 160; heterogeneous, non-metastaticLHRH agonist plus antiandrogen0.05NRDetectable nadir was a significant predictor of clinical progression (HR 17.3; P < 0.001) and PCSM (HR 11.6; P < 0.001)
Soga et al. (2008) [32]; retrospective analysis of an open-label, delayed-complete androgen blockade studyN = 92; local or locally advancedLHRH agonist monotherapy. Bicalutamide added if PSA nadir not achieved0.2NRFailure to achieve nadir with LHRH monotherapy was associated with a significantly greater risk of PSA-progression (HR 16.96; 95% CI 4.58–62.81). Similarly, failure to achieve nadir after addition of bicalutamide lead to an increased risk of biochemical progression (HR 16.05; 95% CI 1.54–167.10)
Stewart et al. (2005) [21]; retrospective database analysisN = 747; rising PSA after local therapyOrchidectomy, LHRH agonist monotherapy, or LHRH agonist plus antiandrogen0.2NRCompared with patients who achieved a nadir ≤2.0 ng/mL, those with a nadir >2.0 had a significantly higher risk of PCSM (HR 18.5; 95% CI 6.0–56.7)
Yu et al. (2010) [47]; retrospective analysis of data from an open-label IADT studyN = 72; rising PSA after local therapy9 months of LHRH agonist plus flutamide0.1NRPSA nadir was not associated with increased risk of development of CRPC or death

The association between a very low or undetectable PSA nadir and improved outcomes is easy to understand. In hormone-sensitive patients with well-differentiated or moderately well-differentiated cancer, PSA levels are reflective of androgen receptor (AR) activity. Inability to achieve undetectable PSA levels may suggest incomplete androgen suppression or the presence of prostate cancer cells in which there is persistent AR activity despite castrate levels of testosterone [43]. This is supported by gene expression and immunohistochemical analyses of primary tumour samples obtained after 3 months of neoadjuvant ADT. Patients who biochemically relapsed had consistently greater PSA mRNA levels and more intense PSA immunostaining than patients who did not relapse, yet they had similar levels of AR gene expression and protein staining [51].

To date, only one study has used the Prentice criteria for validating surrogate clinical markers [52] to validate an absolute PSA nadir value as a surrogate endpoint for PCSM [53]. For a surrogate endpoint, such as PSA nadir, to qualify as a valid surrogate for PCSM under the Prentice criteria, the surrogate endpoint must be statistically associated with the clinical outcome and also capture the net effect of treatment on that clinical outcome (Table 2) [52, 54, 55]. D'Amico et al. [53] retrospectively reviewed data from two randomised controlled trials and found that RT plus 6 months of ADT provided significant reduction in PCSM compared with RT alone. A PSA nadir of 0.5 ng/mL satisfied the Prentice criteria for surrogacy, which adjusts for treatment effect and variation. Based on their analysis, the authors suggest that men with PSA nadir of >0.5 ng/mL after RT and 6 months of ADT be considered for treatmentwith docetaxel or other therapies that have been shown to extend survival in men with metastatic CRPC. It is important to note, as the authors discuss, that this validation of an absolute PSA nadir value as a surrogate marker is only relevant to the specific therapeutic regimens (RT doses and mode of ADT) used in the two randomised controlled trials. Thus, to validate any PSA nadir as a surrogate endpoint, a similar type of rigorous assessment of data must be performed and fulfilment of the restrictive Prentice criteria must be met for each type of ADT method available.

Table 2. The four Prentice operational criteria for validating a surrogate endpoint [49, 52]
Qualification for surrogacy requires that a potential surrogate endpoint yields a valid test of the null hypothesis of no association between treatment effect and the true endpoint (a clinical outcome such as prostate cancer-specific death).
Criterion 1The investigational treatment has a statistically significant impact on the true endpoint.
Criterion 2The investigational treatment has a statistically significant impact on the surrogate endpoint.
Criterion 3The surrogate endpoint has a statistically significant impact on the true endpoint.
Criterion 4The treatment effect on the true endpoint is captured by the surrogate endpoint.

A handful of studies have also shown that time-to-nadir (TTN) has prognostic significance along with PSA nadir. However, the data are conflicting, as a few studies [23, 30] have found that a longer TTN was predictive of higher risk of PCSM, while other studies have shown that a shorter TTN, particularly in patients who do not achieve the set threshold nadir, indicates a poor prognosis [36, 37, 39, 40, 45, 46, 49]. Yet other studies have reported that TTN is not a predictor of outcomes [22, 43] (Table 1). The wide range of results may, in part, be due to differences in study designs, analyses, and patient populations, or may reflect the extremely heterogeneous nature of prostate cancer. Intuitively, one would expect that a more rapid PSA decline in response to ADT would be positively associated with extended survival, as this would suggest apoptosis of androgen-sensitive clones within the tumour. However, in studies that show the opposite association, it has been suggested that the rapid fall in PSA may not be related to cell viability but may reflect down-regulation of AR-responsive PSA expression rather than cell death [39, 45]. Park et al. [46] speculated that the observed association between rapid PSA decline and reduced prostate cancer-specific survival in their study could be due to the presence of neuroendocrine differentiation, as the extent of neuroendocrine differentiation has been significantly associated with relatively low serum PSA and poor prognosis [56]. Huang et al. [36, 37] suggested that while the exact mechanism for these observations is unknown, PSA nadir and TTN may represent the ability of prostate cancer to become castration resistant. A rapid decrease in PSA may be due to ablation of AR function leading to cell-cycle arrest rather than cell death. It will be interesting to see if the more rapid declines in PSA induced by LHRH antagonists compared with LHRH agonists [57] impact outcomes.

PSA Progression

PSA progression is a commonly used term for rising PSA over a given threshold after any therapy. It is often used in clinical trials as an endpoint, as it usually precedes clinical progression and death, or as an inclusion criterion for enrolment. However, this metric has multiple definitions, making it difficult to compare results from studies that define PSA progression differently.

In 1999 [58], the Prostate Cancer Clinical Trials Working Group 1 (PCWG1) met to standardize the definition of PSA progression in patients with CRPC when used as a criterion for enrolment or for reporting outcomes in prostate cancer clinical trials. The convening of PCWG2 in 2007 [59] arose in response to the 2000 publication of Response Evaluation Criteria in Solid Tumors (RECIST), which did not capture some of the key characteristics of prostate cancer, and in response to a call-to-action from the FDA to rework the PCWG1 eligibility and outcome measures so that they could be applied across the clinical spectrum of disease [59]. Table 3 [55, 56] provides a comparison of the definitions put forth by these two groups.

Table 3. PCMG1 and PCWG2 definitions for PSA progression and response in castration resistant prostate cancer [55, 56]
PSA parameterPCWG1, 1999 [55]PCWG2, 2008 [56]
PSA progression for trial eligibilityTwo consecutive increases over a previous documented reference value (nadir), taken ≥1 week apart. To account for fluctuations, increase should be confirmed with a third rise. If a rising PSA is the only evidence of progressive disease, PSA level should be ≥5 ng/mL before enrolling in a clinical trial.Similar to PCWG1, but absolute PSA level reduced to 2 ng/mL. In addition, if three or more values are available, calculate PSADT.
PSA responseDecline of ≥50% confirmed by a second value ≥4 weeks later.Record individual percentage change from baseline in a waterfall plot.
PSA progression as a trial outcome

In patients with no decrease in PSA or who did not achieve response criteria, progression is a 25% increase over baseline value or nadir and an increase in absolute value ≥5 ng/mL and confirmed by a second value.

If ≥50% response, progression is defined as a PSA increase 50% above the nadir at a minimum value of 5 ng/mL.

A ≥25% increase from baseline (if no decline) or nadir (if there is a response) and an absolute value of ≥2 ng/mL.

Multiple groups have published retrospective analyses of phase III trial data to determine if either definition of PSA progression was associated with OS. Hussain et al. [60] retrospectively examined data from two large-scale clinical trials: 1078 hormone-responsive patients who received 7-month ADT in the Southwest Oncology Group (SWOG) trial 9346 for intermittent ADT, and 597 patients with metastatic CRPC treated with chemotherapy in the SWOG 9916 trial. For either study, both the PCWG1 and PCWG2 definitions of PSA progression (Table 3) were strongly associated with OS (P < 0.001). In the SWOG 9346 study, both definitions predicted a 2.4-fold increase in risk of death and a more than 4-fold increase in risk of death if PSA progression occurred in the first 7 months. Both definitions of PSA progression also predicted a 40% increase in risk of death for patients in the SWOG 9916 trial and a 2-fold increase in risk if progression occurred at 3 months [60]. The authors suggested that the PCWG2 definition was clinically more attractive, as it identified progression earlier than the PCWG1 definition and it was also simpler, as it did not differentiate based on PSA response to therapy (Table 3). Analyses of pooled data from >1200 patients with CRPC who participated in nine multi-institutional clinical trials also reported that PSA progression by 3 months, by either definition, was associated with a significant increased risk of death after multivariate analysis [61]. PSA progression, along with PSADT, is not considered a valid surrogate endpoint for drug approval by any regulatory agency; however, it is often an enrolment criterion for clinical trials or a trigger for clinical decision making.

PSA Response in CRPC

In trials assessing the efficacy of therapies for CRPC, a confirmed PSA decline of ≥50% is often an efficacy endpoint, presented as the percentage of patients responding to therapy [58, 59, 62]. Studies reporting an overall PSA response of ≥50% in their efficacy analyses include the phase III registration trials for docetaxel [63, 64] and abiraterone [65], and clinical investigations into second-line antiandrogens [66, 67] and ketoconazole [68]. Interestingly, two separate analyses of data from the docetaxel phase III registration studies, SWOG 9916 and TAX327, found that a PSA decline of ≥30% was a better surrogate for survival than a decline of ≥50% when the Prentice criteria for surrogacy were applied [54, 69]. However, the ≥30% PSA decline was a stronger surrogate for survival in the SWOG 9916 [54], which used docetaxel plus estramustine, while this PSA decline was only considered a modest surrogate in the TAX327 study [69], which used docetaxel alone. This exemplifies the difficulty in establishing a PSA metric as a surrogate endpoint, as determination of surrogacy for PSA response using the Prentice criteria is only valid for the treatment method tested and cannot be applied to other therapies or different therapeutic regimens with the same agent. Similar to PSA progression, PSA response, along with other PSA metrics described above, is not considered a valid surrogate for hard clinical endpoints. Recognising that an overall PSA response had uncertain clinical meaning, the PCWG2 recommended plotting individual PSA responses as a percentage change from baseline to week 12, as well as maximum decline, in a waterfall plot [59]. This provides a broader, more informative display of data. Waterfall plots presenting PSA responses to abiraterone [70-72] and MDV3100 (enzalutamide) [73] in their phase II trials provided initial support for these agents for treatment of CRPC.

For cytotoxic or hormonal therapies, early (≤12 weeks) PSA responses (rises or declines) should not be used for clinical decision making [59] because discordance between PSA change and clinical responses, e.g. pain response [74] or progression/regression of bone metastases, have been reported [75]. Newer therapies, e.g. sipuleucel-T, improved survival in the registration trial without decreasing PSA levels [76]; therefore, PSA declines are not a useful biomarker to monitor response to immunotherapy. However, the anti-cancer immunological mechanism of sipuleucel-T is unclear and there is a need to develop markers of immunotoxicity when monitoring its use in the clinical setting [77].

Recently, there has been a focus on bone as a primary target of treatments for advanced prostate cancer. Saad et al. [41] performed a retrospective exploratory analysis to evaluate possible correlations between PSA levels and clinical outcomes (survival, bone disease progression, and risk of skeletal-related events) in men with bone metastatic CRPC who received either zolendronic acid or placebo, but no cytotoxic chemotherapy. While PSA levels continued to rise in both treatment groups, the rate of PSA increase (PSA velocity) was significantly reduced in the zolendronic acid group [41]. In both treatment groups, PSA increases correlated with significantly increased risk of death, bone disease progression, and skeletal-related events.

PSA Expression throughout the Prostate Cancer Life Cycle

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

Although PSA expression is known to be regulated by AR and thus by androgens (testosterone and dihydrotestosterone [DHT]), studies examining direct correlation between PSA and androgen levels (both serum and intraprostatic or intratumoral) during various stages of prostate cancer are limited. Increasing serum PSA levels in untreated patients with prostate cancer are attributed to the number of cancer cells, the extent of PSA expression per cell, and changes in prostate tissue architecture that allows greater amounts of PSA to leak into the circulation [78]. In prostate cancer cells, PSA expression is reduced compared with healthy prostatic epithelium, reflecting the reduced degree of differentiation of malignant cells [78]. In untreated patients with prostate cancer, there is no correlation between serum androgen and PSA levels. However, after 6-month ADT, PSA levels correlate with serum testosterone (P < 0.001) and DHT levels (P = 0.001) [79]. Bruchovsky et al. [80] closely tracked PSA and serum testosterone levels in men with hormone-sensitive prostate cancer receiving intermittent ADT. PSA declines on treatment paralleled those in serum testosterone and exhibited a specific pattern with each cycle of ADT: a steep decline, representing initial inhibition of PSA expression, followed by a shallow slope due to further reductions in PSA from apoptotic cell loss [80].

In healthy men medically castrated for 1 month with the LHRH antagonist acyline, serum testosterone levels were reduced by 94%; however, prostatic testosterone was only reduced by 70% [81]. Serum PSA levels in these men declined significantly from baseline and became lower than those in the placebo group, but not significantly [81]. Interestingly, analysis of prostate tissue PSA mRNA and protein levels in these castrated healthy men showed substantial variability, much more so than that in the placebo-treated controls [82]. In localised prostate cancer, primary tumour samples examined after up to 9 months of ADT, PSA expression was also widely and variably detected, both via mRNA detection and immunohistochemical protein staining [82]. Importantly, in contrast to these histological observations, serum PSA levels declined substantially in response to hormonal suppression. These data suggest that serum PSA is not a reliable surrogate for assessing tissue response to ADT and may explain some of the conflicting data regarding PSA metrics described earlier. In both healthy men and patients with prostate cancer who had been medically castrated, AR expression was heterogeneous and comparable with placebo [82].

In CRPC, rising PSA levels reflect the growth/activity of tumour foci that may harbour a dysfunctional AR: that is AR with increased expression resulting in AR-dependent gene expression at very low androgen levels, with gene mutations that increase or broaden its function (increased ligand promiscuity), or activation of AR by other signalling pathways independent of androgens [83]. In addition, mutations in AR co-activators or co-repressors, endogenous tumoral androgen synthesis, and the development of proliferative pathways that completely circumvent AR could account for rising PSA levels in a castrate environment [83, 84].

Persistent signalling of the AR axis in CRPC has been shown by the detection of PSA protein in rapid autopsy samples of multiple metastatic tumours [85, 86]. The percentage of cells expressing PSA varied considerably between patients, between tumour samples from single patients, and even between samples taken from the same tissue site [85]. Quantification of mRNA levels revealed increased AR expression and equivalent levels of PSA expression in CRPC metastases compared with primary prostate cancer and benign prostate tissue samples, attributed to elevated levels of intratumoral testosterone levels [86].

Recommendations for Use of PSA for Clinical Decision Making

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

A recent systematic review found a lack of scientific or systematic approach to the development of society-based guidelines and best practice statements for PSA monitoring during the management of prostate cancer, due to inadequate available evidence and inappropriate use of this evidence [87]. As such, PSA metrics should not be used in isolation to make clinical decisions, but in conjunction with other clinical markers.

In patients with a rising PSA after local therapies, highly predictive markers of metastasis include not only a PSADT of <9 months, but also a Gleason score 8–10 [18]. Other clinical factors, e.g. age, comorbidities, and time to treatment, should be included in the decision to initiate ADT. Most importantly, the decision to begin hormonal therapy without evidence of metastasis is controversial. In a heterogeneous patient population with PSA recurrence after RP and not treated until metastases occurred, median metastasis-free survival was 7.9 years while median OS was determined to be >23 years [17]. It was also reported in this study that a PSADT of <3 months was the most significant predictor of shorter metastasis-free survival and shorter OS [17]. No mode of ADT has received regulatory approval in the USA for use based on rising PSA without evidence of metastasis.

We recommend that PSA measurements should be taken every 3 months in patients receiving ADT. This interval can be modified based on a patient's PSA level before ADT, initial response to treatment, Gleason score, and existing comorbid illnesses. A stable PSA nadir at undetectable levels (<0.2 ng/mL) is desirable. If PSA levels do not achieve a desired nadir, a serum testosterone measurement should be done to confirm the achievement of castrate testosterone levels. While the regulatory- and industry-accepted definition of castrate testosterone levels is ≤50 ng/dL, a lower threshold of 20 ng/dL has been suggested as more appropriate due to the availability of more sensitive assays [88]. If the patient is not at castrate levels, a change in testosterone-lowering agents may achieve this goal, or an antiandrogen can be added to therapy to further lower androgen levels. Additional bone scans should be done to monitor clinical progression. Prostate cancer progression after localised therapy or ADT in a setting of very low PSA levels (<2 ng/mL) has been reported but is rare [89]. This most often occurs in patients presenting with undifferentiated prostate cancer at the time of diagnosis and when ductal, neuroendocrine, or small cell cancer variants are present [89-91].

Once the presence of CRPC is established, PSA monitoring should continue, along with imaging. Patients with a rapid PSADT should be considered for enrolment in clinical trials. As mentioned above, PSA responses to second- and third-line therapies vary depending on the type of therapy. It is important for clinicians to be aware of how various therapies affect PSA levels, if at all.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

The PSA test is a quick, nonintrusive, and inexpensive means of monitoring patients with prostate cancer. Yet, despite numerous studies, mostly retrospective, exploring various PSA metrics as indicators of treatment response or prognostic markers, validation remains difficult and is becoming increasingly complicated as the therapeutic armamentarium for CPRC widens. As such, society- and regulatory-based guidelines for the incorporation of PSA metrics in clinical management of advanced prostate cancer are limited. Importantly, PSA responses to therapies for CRPC vary and need to be assessed within the individual context of each therapeutic strategy. This requires prospective incorporation of PSA endpoints in clinical trials. While we await these data, we have provided consensus guidance based on available data about the appropriate use of PSA for clinical decision making.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

The authors wish to thank Janssen Biotech, Inc., for its sponsorship of the roundtable, topic selection, and resultant transcript, which led to the development of this manuscript. The authors developed the manuscript independently without contribution, oversight, or review of the content by the sponsor. Thus, the content reflects the exclusive knowledge and expert opinions of the authors. Medical writing support was provided by Meryl Gersh, PhD, of AlphaBioCom, LLC, in King of Prussia, PA, USA.

Funding/Support and Role of the Sponsor

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

Janssen Pharmaceuticals, Inc., provided financial support for the development of this article and the roundtable discussion on which it was based. The company invited the authors to be involved in the roundtable discussion and defined the scope of the meeting in collaboration with the chair. Janssen Pharmaceuticals, Inc., funded independent medical writing support but had no role in the preparation or approval of the manuscript. Editorial control resides with the authors and the editor.

Conflict of Interest

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References

E. David Crawford serves as a consultant for Ferring, Dendreon, Bayer, Janssen, and Amgen, and as an investigator for Precision Biopsy.

Gerald L. Andriole is a consultant for Amarex, Amgen, Augmenix, Bayer, Bristol Myers Squibb, Cambridge Ende, GlaxoSmith Kline, Janssen Biotech, Myfiad Genotics, Ortho-Clinical Diagnostics, Steba Biotech, Theroneo Medical Isolopes, and Viking Medical. He serves as an investigator for Johnson & Johnson, Medivation, and Wilex.

Marc B. Garnick serves as editor-in-chief of the Harvard Annual Report of Prostate Disease, for which he reseives an honorarie, and is a member of the US FDA ODAC that reviewed chemoprevention drugs.

Daniel P. Petrylak is a consultant for Amgen, Novartis Otsuka, Egenix, Bellicum, Ferring, AstraZenica, Johnson and Johnson, Dendreon, and Milleneum; has received research support from Progenics, Celgene, Sanofi-Aventis, Dendreon, Pfizer, GlaxoSmithKline, Boehringer Ingleheim, Medivation, Esai, Abbott, and Johnson and Johnson; and is a board member for the Prostate Cancer Education Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data Acquisition
  5. PSA Metrics in Advanced Prostate Cancer
  6. PSA Expression throughout the Prostate Cancer Life Cycle
  7. Recommendations for Use of PSA for Clinical Decision Making
  8. Conclusions
  9. Acknowledgements
  10. Funding/Support and Role of the Sponsor
  11. Conflict of Interest
  12. References