The concept of a prostate-specific antigen (PSA) “nadir” has been used as a predictive marker for treatment success in patients treated with radiotherapy for localized prostate carcinoma. However, this approach is not applicable in patients who are concomitantly treated with short-term hormonal therapies. To address this, the authors sought to develop a new predictive marker in such patients after prostate brachytherapy (BT).
Between March 1997 and November 2002, 194 men with clinical Stage T1A-T3N0M0 prostate carcinoma (according to the 1992 International Union Against Cancer/American Joint Committee on Cancer TNM classification system) were treated with interstitial palladium (103Pd3) BT and androgen ablation therapy with or without external beam radiotherapy (EBRT). Based on tumor characteristics, 127 patients received an antiandrogen, finasteride, and BT whereas 67 received an antiandrogen, leuprolide, and EBRT followed by a BT boost. Hormonal therapy was initiated 2–3 months before any radiotherapy for a total duration of 8–9 months. Follow-up included physical examination and determining the PSA level at 3-month intervals. Postoperative serum testosterone was evaluated in preoperatively potent patients with erectile dysfunction > 6 months after therapy. A PSA level ≤ 0.06 ng/mL or ≤ 0.20 ng/mL detected during a 6–12-month window after the implant were evaluated as predictors of biochemically disease-free survival (DFS), defined as the time to a PSA level ≥ 1.0 ng/mL.
Of the 194 patients, 163 were available for analysis. The median length of follow-up was 48 months. In those patients with a PSA level ≤ 0.20 ng/mL at 6–12 months, the DFS at 48 months after the implant was 96% (95% confidence interval [95% CI], 91–99%) compared with the remainder of the patients, whose DFS decreased to 80% (95% CI, 65–89%) (P < 0.001). When a PSA level ≤ 0.06 ng/mL was used as an indicator, the 48-month DFS was 99% (95% CI, 91–100%) compared with that for patients with a PSA level > 0.06 ng/mL, in whom the DFS was 85% (95% CI, 74–92%) (P = 0.004). Furthermore, because testosterone levels may occasionally remain low after the cessation of luteinizing hormone-releasing hormone agonist therapy and result in erectile dysfunction and an artificially low PSA level, the authors reviewed the serum testosterone levels in 23 patients who were so treated and were experiencing erectile dysfunction. None had PSA values below the lower limit of normal.
Serum prostate-specific antigen (PSA) is a marker used to evaluate treatment success after the treatment of localized prostate carcinoma. After radical prostatectomy, the presence of any detectable serum PSA is considered evidence of treatment failure. After radiotherapy, the benign glandular epithelium producing PSA is not always completely destroyed and therefore some serum detection of PSA may be expected.
With the intent to delineate early markers of success after radiotherapy, several authors have attempted to define the level of posttreatment PSA that indicates cure after radiotherapy for the treatment of localized prostate carcinoma.1–10 The concept of the “PSA nadir” as a predictive marker for durable long-term treatment success was introduced in the context of external beam radiotherapy (EBRT) and has been shown to be a useful predictive marker for disease failure in many1, 4, 11 but not all studies.12 Subsequent to its use in patients treated with EBRT, it was used in patients treated with combined prostate brachytherapy (BT) and EBRT.8, 13–15
However, because neoadjuvant and adjuvant androgen ablation have been shown to be advantageous with respect to tumor control when combined with radiotherapy for the treatment of prostate carcinoma,16, 17 an increasing number of patients are undergoing such treatment. Unfortunately, in this situation, the “PSA nadir” approach is not applicable. To address this, we sought to develop a new predictive marker in patients treated with neoadjuvant androgen ablation after radiotherapy for prostate carcinoma.
MATERIALS AND METHODS
Patient Population, BT Procedure, and Follow-Up
The study was comprised of 194 consecutive men with clinical Stage T1c-T3N0M0 (according to the 1992 International Union Against Cancer [UICC]/American Joint Committee on Cancer [AJCC] TNM staging system18) carcinoma of the prostate who were treated with interstitial palladium (103Pd) BT, and androgen ablation therapy with or without EBRT from March 1997 to November 2002. Pretreatment bone scan and computed tomography (CT) scans of the pelvis were performed if a PSA level ≥ 10 ng/mL, a Gleason score ≥ 7, or clinical stage of disease ≥ T2b was present. Patients without evidence of metastases were assigned to advanced disease and early-stage disease groups based on disease stage, Gleason score, and pretreatment PSA level (Table 1). Patients with either a PSA level ≥ 10 ng/mL, a Gleason score ≥ 7, or clinical stage of disease ≥ T2b were stratified to receive BT combined with EBRT (patients with advanced disease), whereas all other patients were treated with BT as monotherapy (patients with early-stage disease).
Table 1. Classification of Patients Based On Failure to Reach an “Undetectable PSA” Level Using Both the 0.20 ng/mL and 0.06 ng/mL PSA Definitions as a Function of Various Clinical and Demographic Criteriaa
PSA: prostate-specific antigen; SD: standard deviation.
For each level of a clinical or demographic criterion, both the frequency and the rate of classification as a biochemical failure or a nonfailure were noted.
b “Failure” indicates biochemical failure during the follow-up period, defined as a prostate-specific antigen level > 1 ng/mL.
c “None” represents patients for whom there was insufficient information available to make a classification using this cutoff value. Percentages noted exclude those individuals who were not classified.
Mean age (yrs) (SD)
≤ 10 ng/mL
Missing Gleason score
Clinical stage of disease
Missing stage of disease
The implant procedure was performed as described previously.19 Patients in the early-stage disease group received a 103Pd implant dose of 115 grays (Gy). Patients in the advanced disease group were treated with an EBRT dose of 45 Gy over 6 weeks followed by a 103Pd implant of 90 Gy given 1 month after the completion of EBRT. All patients received 8–9 months of androgen ablation. The early-stage disease group (n = 127 patients) received an antiandrogen (bicalutamide [at a dose of 50 mg daily] or flutamide20) and finasteride (10 mg daily) beginning 2–3 months before the 103Pd implant. The patients with advanced disease (n = 67) received a combination of an antiandrogen and leuprolide (3-month or 4-month depot injection repeated 3 times and twice, respectively) beginning 2 months prior to the initiation of EBRT.
Patient follow-up after the implant procedure included digital rectal examination and a PSA level taken at 3-month intervals. Serum testosterone was evaluated in preoperatively potent patients with erectile dysfunction > 9 months after agonist therapy with luteinizing hormone-releasing hormone (LHRH) because testosterone levels may occasionally remain low in such situations and result in erectile dysfunction and an artificially low PSA level.
Endpoints and Patient Exclusion Criteria
Biochemical failure after radiotherapy for localized prostate carcinoma has been defined as a PSA level of > 0.2 ng/mL to 4.0 ng/mL.1–5, 7–15, 21–28 For the current study, we defined it as a PSA level of ≥ 1.0 ng/mL. “Disease-free survival” (DFS) is defined as the time from 103Pd implant to biochemical failure or last follow-up in patients without biochemical failure. “Undetectable PSA” is defined as a PSA level ≤ 0.06 ng/mL or ≤ 0.20 ng/mL as measured during routine follow-up. We used both of these definitions because our institution implemented the ultrasensitive PSA assay during the course of patient accrual. In this article, we tested the hypothesis that an undetectable PSA level 6–12 months after the implant is a predictor of DFS. We defined this 6–12-month interval as the “screening interval.”
Patients without any PSA measurements and those with biochemical failure (a PSA level ≥ 1.0 ng/mL) during the screening interval were excluded from the current analysis. These latter patients were in turn examined using the American Society for Therapeutic Radiology and Oncology (ASTRO) Consensus Panel definition of biochemical failure.20
Patients were classified into binary categories based on their first PSA measurement taken within the screening interval. Two separate classifications were made, based on both definitions of undetectable PSA as outlined earlier. Log-rank tests were performed to determine whether rates of DFS differed based on whether patients had a detectable or undetectable PSA level during the 6–12-month screening interval. Kaplan–Meier survival estimates and 95% confidence intervals (95% CI) were used to summarize DFS rates. Chi-square tests of association were used to examine differences in the rates that patients were classified into the undetectable PSA subgroup between those patients with and those without testosterone measurements. Log-rank tests and univariate Cox proportional hazards modeling were performed to determine whether clinical parameters including stage of disease, baseline PSA, and Gleason score were correlated with PSA failure.
Of the 194 patients, 17 patients were excluded due to the lack of PSA follow-up after > 12 months and 14 were excluded because they demonstrated biochemical failure within the screening interval (please note that 1 patient both lacked follow-up and failed treatment within the screening interval). The remaining 163 patients had a median follow-up from the date of 103Pd implant of 48 months (range, 13–72 months). Using the entire cohort of 163 men, patients with a PSA level ≤ 0.20 ng/mL during the screening interval had a DFS at 48 months of 96% (95% CI, 91–99%) (Fig. 1A). Conversely, if the PSA level was > 0.20 ng/mL during this interval, the DFS decreased to 80% (95% CI, 65–89%) (P < 0.001) (Fig. 1A). Using the “undetectable PSA” definition of a PSA level ≤ 0.06 ng/mL during the screening interval, the 48-month DFS was 99% (95% CI, 91–100%) (Fig. 1B) whereas for patients with a PSA level > 0.06 ng/mL during the screening interval, the DFS was reduced to 85% (95% CI, 74–92%) (P = 0.004) (Fig. 1B).
When a similar analysis was applied to the early-stage disease group alone (n = 102), we found that if the PSA level was ≤ 0.20 ng/mL during the screening interval, the DFS at 48 months was 97% (95% CI, 88–99%) (Fig. 2A). If the PSA level was > 0.20 ng/mL during the screening interval, then the DFS decreased to 78% (95% CI, 62–87%) (P < 0.001) (Fig. 2A). Using the definition of a PSA level ≤ 0.06 ng/mL, the DFS was 100% (Fig. 2B) whereas if the PSA level was > 0.06 ng/mL, the DFS was reduced to 85% (95% CI, 72–92%) (P = 0.011) (Fig. 2B). There were not enough patients in the advanced disease group alone to allow for independent analysis.
To assess whether there were correlations between PSA failure and clinical parameters including stage of disease, baseline PSA level, and Gleason score, log-rank tests and univariate Cox proportional hazards modeling were performed. None of these clinical parameters were found to be statistically significant for the PSA failure cutoff value when tested univariately.
When assessed using the ASTRO criterion,20 DFS was also examined among the 14 patients excluded from the analysis because of PSA failure (PSA level ≥ 1.0 ng/mL) within the screening interval. Among these patients, the DFS rate relative to the date of treatment was found to be 85% (95% CI, 52–96%) at 48 months.
DFS relative to the time of the PSA nadir was estimated after classifying patients according to both the 0.20 ng/mL and 0.06 ng/mL PSA nadir. Patients with a PSA level ≥ 1.0 at the PSA nadir were excluded. Defining biochemical failure as a PSA level ≥ 1.0 ng/mL, DFS rates 3 years from the PSA nadir were reported to be 89% (95% CI, 84–93%) for those with a PSA level ≤ 0.20 ng/mL and 0% for those with a PSA level > 0.20 ng/mL (P < 0.001; n = 222). However, it should be noted that the group of patients with a PSA level > 0.20 ng/mL was extremely small (n = 7). Using the cutoff value of 0.06 ng/mL, the DFS was 93% (95% CI, 86–97%) for a PSA level ≤ 0.06 ng/mL and 65% (95% CI, 34–84%) for a PSA level > 0.06 ng/mL (P < 0.001; n = 136). Furthermore, using ASTRO criteria20 to define biochemical failure, the DFS 3 years from the PSA nadir was 87% (95% CI, 82–91%) for a PSA level ≤ 0.20 ng/mL and was not found to be significantly different from the DFS for the group with a PSA level > 0.20 ng/mL (P = 0.56; n = 222). Using the 0.06 ng/mL criteria, the DFS was 85% (95% CI, 76–90%) for patients with a PSA level ≤ 0.06 ng/mL and 78% (95% CI, 51–91%) for patients with a PSA level > 0.06 ng/mL (P = 0.18) (n = 136).
Testosterone levels may occasionally remain low even after the cessation of LHRH agonist therapy, and this may in turn maintain PSA levels at low levels during the follow-up period. Therefore, patients in the advanced disease group who were treated with LHRH agonists had a serum testosterone drawn by their physicians if they reported erectile dysfunction more than 9 months after BT. Postoperative testosterone levels averaged 400 ng/dL and no individual value was below the lower limit of normal. These data indicate that erectile dysfunction was not due to the continued suppression of testosterone from previous LHRH treatment.
Patients with normal erectile function are unlikely to have castrated levels of testosterone. Nevertheless, because we did not have known testosterone levels available for all patients treated with LHRH agonists, we sought to determine whether there was a difference in the proportion of patients having undetectable PSA in the screening interval between those with available testosterone measurements and those without. Using the cutoff value of 0.20 ng/mL, the respective percentages of patients with undetectable PSA during the screening interval for those with available testosterone measurements and those without were 87% and 79%, respectively (P = 0.39). Similarly, for those with a PSA level under the cutoff value of ≤ 0.06 ng/mL, the corresponding percentages were 71% and 58%, respectively (P = 0.25). These data indicate that for both the 0.06 ng/mL and 0.20 ng/mL definitions, the proportion of patients that fell into the undetectable PSA group during the screening interval did not significantly differ between those with an available testosterone measurement and those without. This is further evidence that suppressed testosterone was not responsible for an undetectable PSA level during the screening interval. In addition, these data also suggest that prior LHRH therapy did not keep the PSA level artificially low during the follow-up period.
Several authors have investigated whether the PSA nadir could be used to predict subsequent failure after radiotherapy for the treatment of localized prostate carcinoma, the idea being that if we can find an early predictor of failure, salvage treatment may be instituted earlier. In addition, the efficacy of novel radiotherapeutic approaches could be evaluated with greater rapidity, thus accelerating the pace of scientific progress.
Lee et al.4 reported on 364 men with T1-T3 localized prostate carcinoma who received definitive EBRT without prior adjuvant hormonal therapy. Biochemical failure was defined as a PSA level of ≥ 1.5 ng/mL and 2 consecutive serum PSA elevations. The 5-year overall biochemical DFS rate was 56%. The PSA nadir was found to be an independent predictor of DFS. When examining specific PSA nadir values, the DFS rate at 3 years for a PSA nadir of 0–0.99 ng/mL was reported to be 99%, versus 49% for a PSA nadir of 1.0–1.99 ng/mL and 16% for a PSA nadir of ≥ 2.0 ng/mL. The authors concluded that patients who achieve a PSA nadir of < 1.0 ng/mL after EBRT have a favorable biochemical DFS rate.
Crook et al.15 performed a similar study investigating 207 men treated with EBRT for localized prostate carcinoma who were followed prospectively with systematic transrectal ultrasound-guided biopsies of the prostate and PSA measurements. Biochemical failure was defined as a PSA level of ≥ 2 ng/mL and > 1 ng/mL over the PSA nadir. The PSA nadir was found to be predictive of biochemical, local, and distant failure on univariate analysis. Multivariate analysis revealed only the PSA nadir to be predictive of local failure. The median PSA nadir for patients who were free of any type of failure was ≤ 0.5 ng/mL, and was achieved at a mean time of 20 months. Therefore, the authors concluded that a PSA nadir of ≤ 0.5 ng/mL was associated with a very advanced probability of disease-free status, whereas the risk of failure appeared to rise progressively with PSA nadirs > 0.5 ng/mL.
Critz et al.5, 13, 14, 21–25 have conducted numerous studies in an attempt to define the PSA nadir value that defines freedom from disease for men treated with combined BT and EBRT for localized prostate carcinoma. One study22 reported on 660 men with T1-2N0 prostate carcinoma who were treated with an iodine-125 (125I) BT implant followed by radiotherapy. Recurrence was defined as a PSA rising above whatever nadir was achieved. Approximately 81% of patients had a PSA nadir of ≤ 0.5 ng/mL. These patients had 5-year and 10-year DFS rates of 93% and 83%, respectively. For those patients with a PSA nadir of 0.6–1.0 ng/mL, the 5-year DFS rate decreased to 26%. All men with a PSA nadir > 1.0 ng/mL ultimately failed treatment after 5 years of follow-up. The authors concluded that a PSA nadir of ≤ 0.5 ng/mL should be achieved for possible cure after radiotherapy for localized prostate carcinoma. Subsequent reports21, 23, 24, 29 concluded that the PSA nadir is the fundamental measurement that determines possible cure after radiotherapy and that a PSA nadir of ≤ 0.5 ng/mL is a reasonable definition of freedom from disease. A more recent article from 2002 by Critz et al.14 suggested using a PSA cutoff value of ≤ 0.2 ng/mL as the standard definition of freedom from prostate carcinoma after treatment with BT.
Unfortunately, we believe that despite its promise, the “PSA nadir” approach is not applicable in patients undergoing radiotherapy and receiving neoadjuvant or adjuvant hormonal therapies. Because the use of androgen ablation has been shown to have an impact on the success of radiotherapeutic treatment,16, 17, 30 it is being used increasingly and therefore fewer patients can benefit from the use of the PSA nadir value.
To address this issue, we attempted to develop a new predictive marker in patients treated with neoadjuvant androgen ablation after BT of the prostate. This new marker was defined as an undetectable PSA level during the 6–12-month period after treatment. Two definitions of “undetectable PSA” were examined (≤ 0.20 ng/mL and ≤ 0.06 ng/mL) to determine whether they could be used to predict DFS. We found that both definitions could be used to predict a statistically significant difference in DFS. A PSA level ≤ 0.20 ng/mL appeared to be a better discriminator between those patients who failed treatment and those who remained free of disease. Those patients with a PSA level ≤ 0.20 ng/mL at 6–12 months from the time of 103Pd implantation had a DFS at 48 months after implantation of 96%, compared with 80% for those patients with a PSA level > 0.20 ng/mL at 6–12 months after implantation. Although we were able to examine patients with early-stage disease separately, our work was limited by the fact that we did not have enough patients in the advanced disease group to analyze them independently of the early-stage disease group. Therefore, a larger cohort would be necessary to make firm conclusions regarding this specific group.
Because to our knowledge antiandrogens and 5-alpha reductase inhibitors do not have a suppressive effect on serum testosterone, patients with early-stage disease should not have their follow-up PSA values affected by prior androgen ablation. However, it has been reported by Nejat et al. that treatment with LHRH agonists can result in the continued suppression of testosterone, even after the medication is withdrawn.31 In the study by Nejat et al., the median age of the patients was 71 years, compared with 67 years in the current study, and the median time to normalization of testosterone was 7 months. It is important to note that short-term androgen deprivation, which is what was used in the current study, was found to be associated with minimal testosterone suppression after the discontinuation of LHRH therapy.
Nevertheless, we felt compelled to mitigate the effect of hormonal therapy as a cause for a low PSA level during the screening interval among the patients in the advanced disease group. We addressed this effort using two approaches. First, we reviewed the serum testosterone levels in the advanced disease group to exclude the effect of adjuvant hormonal therapy on PSA levels. In this group, patients had their serum testosterone level measured by their physicians if they reported erectile dysfunction in the 9–12 months after BT. Postoperative testosterone levels were available in 23 patients, with an average value of 400 ng/dL. Therefore, it is unlikely that an undetectable PSA level at 6–12 months after interstitial BT and neoadjuvant hormonal therapy in the group of patients with advanced disease is due to residual testosterone suppression in this patient cohort. Furthermore, in the patients with advanced disease for whom the testosterone levels were not available, we would expect these patients to report symptoms of erectile dysfunction when questioned if they really had castration levels of testosterone.
Second, to exclude suppressed testosterone as an explanation for the undetectable PSA level during the screening interval, we examined the rates at which patients were classified into the undetectable PSA group. We found that the rates of classification into the undetectable PSA subgroup did not differ between those patients with and those without testosterone measurements using either PSA cutoff value, thereby failing to support the notion that suppressed testosterone was related to undetectable PSA during the screening interval. Although we believe these analyses have adequately addressed the issue of testosterone suppression, we understand that the gold standard approach would have been to have testosterone levels measured in each patient every time the PSA level was measured. Unfortunately, this is not possible outside of a clinical trial because the testosterone measurement would not have been a medically necessary diagnostic test in patients not reporting erectile dysfunction or other side effects of testosterone suppression at least 6 months after LHRH therapy.
An undetectable PSA level measured during the 6–12 months after neoadjuvant and adjuvant androgen ablation and BT appears to be a useful predictor of a reduced risk of biochemical failure after treatment. The 0.20 ng/mL rather than the 0.06 ng/mL definition of an undetectable PSA level appears to be a better discriminator between those patients who fail treatment and those who remain free of disease. Based on these results, the testing of these concepts appears warranted in larger prospective studies not only in patients treated with neoadjuvant androgen ablation and BT but in those patients treated with such neoadjuvant treatments and EBRT alone.
The authors wish to thank Drs. Jay Y. Gillenwater and Marguerite C. Lippert for contributing patients to the study and Drs. Tyvin A. Rich and Bernard Schneider for their assistance with the brachytherapy implant procedure.