Clinical Investigation
A Comparison of the Prognostic Value of Early PSA Test-Based Variables Following External Beam Radiotherapy, With or Without Preceding Androgen Deprivation: Analysis of Data From the TROG 96.01 Randomized Trial

https://doi.org/10.1016/j.ijrobp.2009.10.071Get rights and content

Purpose

We sought to compare the prognostic value of early prostate-specific antigen (PSA) test-based variables for the 802 eligible patients treated in the Trans-Tasman Radiation Oncology Group 96.01 randomized trial.

Methods and Materials

Patients in this trial had T2b, T2c, T3, and T4 N0 prostate cancer and were randomized to 0, 3, or 6 months of neoadjuvant androgen deprivation therapy (NADT) prior to and during radiation treatment at 66 Gy to the prostate and seminal vesicles. The early PSA test-based variables evaluated were the pretreatment initial PSA (iPSA) value, PSA values at 2 and 4 months into NADT, the PSA nadir (nPSA) value after radiation in all patients, and PSA response signatures in men receiving radiation. Comparisons of endpoints were made using Cox models of local progression-free survival, distant failure-free survival, biochemical failure-free survival, and prostate cancer-specific survival.

Results

The nPSA value was a powerful predictor of all endpoints regardless of whether NADT was given before radiation. PSA response signatures also predicted all endpoints in men treated by radiation alone. iPSA and PSA results at 2 and 4 months into NADT predicted biochemical failure-free survival but not any of the clinical endpoints. nPSA values correlated with those of iPSA, Gleason grade, and T stage and were significantly higher in men receiving radiation alone than in those receiving NADT.

Conclusions

The postradiation nPSA value is the strongest prognostic indicator of all early PSA-based variables. However, its use as a surrogate endpoint needs to take into account its dependence on pretreatment variables and treatment method.

Introduction

Prostate cancer (PC) often has a long natural history, so it can take many years to determine whether new treatment strategies for localized cancers result in improved PC survival. A recent publication showed that the death rate from PC rose threefold after 15 years had elapsed since diagnosis (1). There is, therefore, a need to identify accurate predictors of PC survival that are expressed early in the course of the disease. Such predictors would allow new treatments to be assessed much earlier and sequential clinical trials to be designed in which the best arm in the first trial is selected as the control arm in the second trial.

Using data from the Trans-Tasman Radiation Oncology Group (TROG) 96.01 randomized trial in PC (2), we have already shown that the time to biochemical failure (using the Phoenix method, i.e., a PSA rise of 2 ng/ml above the posttreatment nadir value) and the PSA value doubling time can be used as surrogate endpoints for predicting PC death. In the same dataset, we found that the occurrence of biochemical failure itself, however, is a far weaker predictor of cancer death than time to biochemical failure (3).

Several earlier PSA test-based variables have been shown to have prognostic value. The initial PSA (iPSA) value at diagnosis has been shown to be related to the likelihood of extra-prostatic disease (4). Moreover, metastatic disease is more likely to be detectable once the iPSA is 20 ng/ml or more (5) and algorithms incorporating iPSA results have been developed that define the risk category of individual cancers at diagnosis (6). Its value in predicting survival is less clear, however. Like many other investigators, we have also found the iPSA level to be a good predictor of biochemical failure-free survival (BFFS) in the TROG 96.01 dataset 3, 7. However, in this same dataset, the iPSA level did not predict PC-specific survival (PCSS). Exploratory analyses indicated that increasing iPSA values were no longer prognostic for PCSS at the time of biochemical failure. In fact, men with iPSA values below 10 ng/ml were at greatest risk of PC death from this point in time (3).

A low-nadir PSA (nPSA) value after radical radiation is already recognized as being associated with better PC survival in many series 8, 9, 10, 11, 12. In the TROG 96.01 dataset, the nPSA value was also strongly prognostic. In exploratory analyses, it was found that men with double or polyexponential PSA descent patterns after radiation alone had significantly lower nPSA values than men with exponential descent patterns. Of great interest was the finding that these descent patterns (subsequently named “PSA response signatures” [PRS]) were highly prognostic values independent of the nPSA level (13). The predictive value of nPSA after the radiation component of short-term neoadjuvant androgen deprivation therapy (NADT) plus radiation protocols is less certain. Low PSA values at the end of several months of ADT before radiation or surgery have been shown to predict better BFFS rates in men receiving NADT 14, 15, 16. Although it remains unknown whether PSA-based values after 3 to 8 months of NADT will also predict survival, it is clear that measures of response to NADT deserve evaluation as prognostic variables in men undergoing “curative treatments,” just as they have already been in the palliative treatment setting 17, 18.

In this report we use TROG 96.01 trial data to compare the prognostic impact of all of the pre-biochemical failure PSA test-based variables mentioned above on local and distant failure, as well as BFFS and PCSS rates, to determine if any of them merit evaluation as surrogate endpoint candidates in trials.

Section snippets

Methods and Materials

The TROG 96.01 randomized controlled trial was opened in 1996 following institutional ethical approval of the protocol at participating Australian and New Zealand sites. A total of 802 eligible patients receiving radiotherapy (66 Gy delivered, using 33 weeks of daily fractions of 2 Gy) for locally advanced PC were randomized to receive 0, 3, or 6 months' maximal NADT (goserelin [Zoladex; AstraZeneca Pty Ltd, Sydney Australia], at 3.6 mg subcutaneously per month and flutamide [Eulexin;

Results

Between June 1996 and February 2000, 802 eligible men with locally advanced PC were randomized. Minimum follow-up time in this report is 7.4 years since randomization. Pretreatment characteristics are presented in Table 1.

Results of the comparison of early PSA test-based variables for the radiation-only arm of the trial are summarized in Table 2. The hazard ratios (HR) and 95% confidence intervals (CI) are those for the early PSA test-based variables in Cox models, which also include the

Discussion

In this report we have found that the nPSA value following radiation is a predictor of every outcome measure (including distant failure and PC death) after radiation alone or after NADT and radiation for locally advanced PC. This finding contrasts with all the other early PSA test-based variables except for the PRS variable in the radiation-alone arm of the trial. While these variables predict Phoenix-based BFFS both for men treated with radiation alone and for those treated with NADT, these

Conclusions

In the TROG 96.01 trial, the postradiation nPSA value was a powerful predictor of LPFS, DFFS, BFFS, and PCSS for all 802 trial patients, regardless of treatment arm. Unfortunately, however, its use as a surrogate endpoint needs to take into account its dependence on pretreatment variables and treatment method. PSA test-based variables measured before nPSA, including iPSA and PSA levels during androgen deprivation before radiation, predicted only BFFS.

Acknowledgment

We thank Rosemary Bradford for skillful preparation of the manuscript.

References (19)

There are more references available in the full text version of this article.

Cited by (33)

  • Posttreatment Prostate-Specific Antigen 6 Months After Radiation With Androgen Deprivation Therapy Predicts for Distant Metastasis–Free Survival and Prostate Cancer–Specific Mortality

    2016, International Journal of Radiation Oncology Biology Physics
    Citation Excerpt :

    We chose to look specifically at a 6-month PSA despite many variables that showed statistical significance on UVA for our clinical endpoints. Many studies examining PSA response have used PSA nadir to predict those men most likely to have disease progression (10-13). However, we chose not to include time-dependent variables for our multivariable analysis (PSA nadir and time to PSA nadir) because we were interested in having an early predictor at a well-defined time point to predict patient outcomes.

  • Prostate-specific antigen kinetics following hypofractionated stereotactic body radiotherapy for low- and intermediate-risk prostate cancer

    2016, Cancer Treatment Communications
    Citation Excerpt :

    In our study, the PSA decline of SBRT was not significantly notable in the first year but constantly decreased during the period of 2 and 3 to achieve lower PSA nadir. Lamb et al. showed that the post-radiation nadir PSA is the strongest indicator [33]. Zelefsky et al. demonstrated that nadir PSA values of ≤1.5 ng/mL at 2 years after radiation therapy for prostate cancer predict for long-term distant metastases and cause-specific mortality [34].

  • Prostate-specific antigen kinetics after stereotactic body radiotherapy as monotherapy or boost after whole pelvic radiotherapy for localized prostate cancer

    2015, Prostate International
    Citation Excerpt :

    In our study, the rate of PSA decline was not significantly different between SBRT as monotherapy and SBRT. Lamb et al27 showed that the postradiation nadir PSA is the strongest indicator. Zelefsky et al28 demonstrated that nadir PSA values of ≤1.5 ng/mL at 2 years after radiation therapy for prostate cancer could predict long-term distant metastases and cause-specific mortality.

View all citing articles on Scopus

This work was funded by the Australian Government National Health and Medical Research Council (project grant applications 9936572, 209801, and 455520); the Hunter Medical Research Institute (Newcastle, Australia); AstraZeneca Pty. Ltd. (Sydney, Australia); and Schering-Plough Pty. Ltd., Sydney, Australia.

Conflict of interest: Allison Steigler and David Lamb received support from AstraZeneca to attend one or more meetings; John Matthews received support from AstraZeneca and Schering Plough to attend an Australasian meeting; David Joseph received honoraria associated with membership of AstraZeneca's Breast Cancer Medical Advisory Board; Gillian Duchesne received a travel grant from AstraZeneca to attend a European prostate cancer symposium in 2006; and Nigel Spry received honoraria associated with AstraZeneca and Schering Plough and research funding from Schering Plough.

View full text