James L. Gulley, Director, Clinical Immunotherapy Group, Laboratory of Tumor Immunology and Biology, Center for Cancer Research, National Cancer Institute, 10 Center Drive, 8B07 MSC 1750, Bethesda, MD 20892, USA. e-mail: email@example.com
Authors from the National Cancer Institute present a study which assessed the time to androgen independence and overall survival for patients with metastatic prostate cancer. They reported a longer than expected survival with androgen-independent prostate cancer, and make original observations on the time to metastatic disease for patients with no evidence of radiographic disease.
The new wave of chemotherapeutic agents for hormone-resistant prostate cancer has generated a new interest in producing more. Dr Cora Sternberg from Rome presents information on satraplatin, a third-generation orally available platinum analogue that shares more structural similarities with cisplatin. There are several other papers in this section on prostate bladder and renal cancer, and finally a paper from London on penile cancer, asking what surgical resection margins are required to achieve oncological control, and finding that the traditional 2 cm margin may be unnecessary, with excision margins of only a few millimetres needed to give excellent control.
To determine the natural history of patients with prostate cancer who start initial androgen-deprivation therapy (ADT) for biochemical failure with no radiographic evidence of disease (D0) or with radiographic metastatic disease (D2), as the history is either not well-defined or is changing, and such data are critical for guiding therapy after prostate cancer recurrence.
PATIENTS AND METHODS
We retrospectively assessed the time to androgen-independence (AI), defined as the first sustained rise in prostate-specific antigen (PSA) level on ADT, time to metastatic disease and overall survival for 80 patients with metastatic prostate cancer in clinical trials at the National Cancer Institute.
ADT was initiated after metastatic disease in 37 patients and before metastatic disease in 43 patients; in these 43 patients, the median time to developing metastatic disease on ADT was 37.8 months. The median time to AI from the initiation of ADT was 19.3 and 13.1 months in D0 and D2 patients, respectively. The median overall survival from the start of ADT was 89.0 and 63.0 months, and the median overall survival from the time of AI was 63.1 and 44.2 months in D0 and D2 patients, respectively. These 80 patients, which included 43 who had no metastatic disease when starting ADT, had a median survival of 54.8 months after AI prostate cancer.
We describe the natural history of AI prostate cancer in D0 patients who eventually developed metastasis, and in D2 patients. The results suggest a longer than expected survival with AI prostate cancer, and to our knowledge this is the first study to report the time to metastatic disease for D0 patients from ADT and from AI. These results can be used to help design clinical trials in patients with D0 prostate cancer.
Androgen-deprivation therapy (ADT) is the most effective means of treating advanced prostate cancer and for reversing a rising PSA level in patients with no radiographic evidence of disease (D0). However, prostate cancer typically becomes resistant to androgen ablation and progresses despite castrate levels of testosterone. Two large clinical trials comparing ADT with combined androgen blockade have shown that patients with metastatic prostate cancer who are treated with ADT typically progress after 14–20 months [1,2]. The overall median survival after ADT in these two trials of men with D2 (radiographic metastatic) disease is 28–36 months. Clinical trials of men with androgen-independent (AI) prostate cancer generally show that the median survival is 12–18 months. However, these values are generally from phase II and III chemotherapy trials of men with advanced disease, and enrolment generally is not at the first increase in PSA level after ADT [3–6].
The survival from the first sustained increase in PSA level after starting ADT is not well characterized. A recent study suggested that the median survival for these patients might be up to 40–68 months . That study was the first to describe survival after ADT from the sustained increase in PSA level; we are not aware of a study that specifically defines the natural history of D0 patients from the first sustained increase in PSA level after ADT. Therefore, we sought to delineate the natural history of AI prostate cancer in patients enrolled in two clinical trials at the National Cancer Institute (NCI). A major caveat of the present analysis in D0 patients is that while not all of such men necessarily develop metastasis, all of the present patients did so.
PATIENTS AND METHODS
We retrospectively assessed data from sampling times before enrolment of 80 patients on two chemotherapy trials at the NCI. Therefore, the present analysis is based on data before these patients were treated on these trials at the NCI. Patients had extensive PSA documentation and histories of treatment with ADT. In the first study patients were enrolled on a trial of ketoconazole with or without alendronate (K&A). In the second study patients received docetaxel with or without thalidomide (T&T). Both trials were approved by the NCI internal review board and were in accord with the Declaration of Helsinki. All patients had metastatic disease to soft tissue or bony disease and were progressing on ADT.
The patients were divided into two cohorts; the first, referred to as D2, comprised patients treated with ADT after being diagnosed with metastatic disease; the second, referred to as D0, comprised patients who were treated with ADT, who had an increasing PSA level but no radiographic evidence of metastatic disease. AI was defined as the first increase in PSA level in a series from the PSA nadir after starting ADT (either medical or surgical castration). We characterized the D0 and D2 cohorts for Gleason score and type of ADT, and examined the time to AI from starting ADT, and from the PSA nadir date, and survival from the start of ADT for both these cohorts. In the D0 cohort we also assessed the time to first metastatic disease after starting ADT and survival from the date of diagnosis.
The probability of survival, or of developing metastases, or AI as a function of time, was determined using the Kaplan-Meier method. The significance of the difference between pairs of Kaplan-Meier curves was determined using the Mantel-Haenszel technique; all P values are two-tailed.
The median age of patients on the K&A trial and T&T trial was 68.8 years and 66.0 years, respectively, and for patients on both trials was 67.2 years; the median (range) age at diagnosis in the D0 and D2 cohorts was 61 (45–73) and 62 (39–73) years (Table 1). Of the 80 patients, 76 (95%) had metastatic disease to the bone, as shown by uptake of isotope on a bone scan, and 46 (58%) had soft-tissue disease, as confirmed by CT or MRI; 42 (53%) had both soft-tissue and bone disease.
Table 1. The distribution of age at diagnosis, the Gleason scores, form of ADT and survival and time to AI of D0 and D2 cohorts
N (%) of patients per cohort
Age at diagnosis, years
Form of ADT
GnRH agonist alone
GnRH agonist and androgen antagonist
Orchidectomy and androgen antagonist
Survival and time to AI, months per patient cohort (P)
AI from ADT
AI from PSA nadir
Survival from ADT
Survival from AI
Overall survival from diagnosis
Of the 80 patients, 43 (D0 cohort) received ADT before developing radiographic evidence of disease; 22 of them had a radical prostatectomy (RP) as their initial local therapy, while 15 received radiation therapy, one received brachytherapy and five had no local therapy. The 37 patients in the D2 cohort had metastatic disease at the time of ADT; 11 (30%) had soft-tissue metastatic disease before starting ADT and 31 had bone metastatic disease. The median Gleason score for the D0 and D2 patients was 8 and 7, respectively. The distribution of Gleason scores for these two subsets of patients are also detailed in Table 1.
Historical hormonal treatment data were collected for both patient cohorts. Men were treated with a GnRH agonist alone, GnRH agonist and an androgen-receptor antagonist, orchidectomy alone or orchidectomy plus an androgen antagonist; these data are also given in Table 1.
For patients in the D0 cohort the median time to AI from starting ADT was 19.3 months, and to AI from the PSA nadir was 7.3 months; the median time from starting ADT to first metastatic disease was 37.8 months. Importantly, not all D0 patients develop metastatic disease, and the present patients represent a subset of such patients who all developed metastasis. The median survival from starting ADT was 89.0 months and median survival from AI 63.1 months. The selection bias for D0 patients who developed metastatic disease probably also affected the survival data. The median overall survival from diagnosis was 111.6 months. Figure 1a shows the Kaplan-Meier curves of the probabilities of AI, first metastatic disease and survival as a function of time from the start of ADT for the D0 cohort.
For patients in the D2 cohort the median time to AI from starting ADT was 13.1 months, and to AI from the PSA nadir was 7.5 months; the median survival duration from starting ADT was 63.0 months and the median survival from AI was 44.2 months. The median overall survival from diagnosis was 94.2 months; Fig. 1b shows the Kaplan-Meier curves of the probabilities of AI and survival as a function of time from starting ADT for the D2 cohort. Table 1 also summarizes the data for the D0 and D2 cohorts.
As expected, the time from ADT to AI and survival, and the overall survival, were shorter in the D2 than the D0 cohort; in the combined population the median time to AI was 15.6 months and the median survival from AI was 54.8 months.
Importantly, the D0 cohort in the present study was a very select group. Pound et al. described the natural history of patients with an increase in PSA level after RP, finding that, of 304 men with an increase in PSA level after RP, the actuarial time to metastasis was 8 years. Only a third (103 men) of their patients developed metastatic disease within the study period; most had ADT only after being diagnosed with metastatic disease. In contrast, the present D0 cohort represents a selected group, all of whom eventually developed metastasis and started ADT before having metastatic disease.
Perhaps the largest studies assessing the natural history of D2 patients after ADT are the two large randomized trials comparing medical or surgical castration with maximum androgen blockade [1,2]. The four groups of patients on these trials had a median time to progression of 13.9–20.4 months and an overall survival after starting ADT of 28.3–35.6 months. However, progression after starting ADT in these studies was defined radiographically and the increase in PSA level was not assessed.
Oefelein et al. first described the survival of patients with an increasing PSA level after ADT and reported a median survival of 40 and 68 months for patients with and without skeletal metastasis, respectively. A second retrospective study by Shulman et al. of 160 patients with AI prostate cancer with and without metastatic disease who were treated with ADT found a median survival of 46 months from AI. These patients were stratified into low, intermediate and high risk based on PSA doubling time, time to PSA recurrence and PSA nadir.
The results for survival with AI in the present combined cohort are similar to that from biochemical failure in the heterogeneous group of patients reported by Shulman et al. Also, survival from AI in the present D0 and D2 cohorts was similar to that reported by Oefelein et al.. However, the survival of patients in the larger prospective studies comparing ADT with combined androgen blockade is significantly lower than in the present study, and than reported by Oefelein et al. and Shulman et al. These differences may be due partly to a difference in performance status, a bias toward patients who have tried more treatment regimens or more effective chemotherapy regimens in recent years. It is also possible that, because PSA values are followed more closely for actively treated patients, a change is made sooner from a treatment whose benefit has been exhausted to one that may be of benefit. Importantly, any bias of the present data from treatment on K&A or T&T might influence survival, but not other endpoints such as the time from ADT to AI.
Although the difference was not statistically significant, D2 patients had a shorter time to an increase in PSA level than D0 patients. This probably reflects the presence of more malignant clones in D2 patients and the consequently higher likelihood of having AI clones. We think that, given a larger sample, the difference in time to AI from ADT between the cohorts might be statistically significant, as might survival from AI prostate cancer and overall survival from diagnosis.
It is important to be able to anticipate the lag time from biochemical failure to radiographic evidence of disease. We report that, for patients with an increase in PSA level as the only evidence of disease, the median time to metastatic disease from starting ADT was 37.8 months in patients followed clinically. Described another way, the median time to a sustained increase in PSA level from ADT in D0 patients was 19 months and that from an increase in PSA to metastatic disease was another 19 months. One caveat is that all of the patients in the present trials developed metastases from stage D0. Not all D0 patients necessarily develop metastasis , and thus the present D0 cohort is not necessarily representative of such patients in general. This probably biases the present results toward D0 patients who have a more aggressive clinical course. Indeed, evidence from a placebo arm of a study evaluating time to metastatic disease in D0 patients suggests that at 2 years from study enrolment two-thirds of men were free of metastasis . A second caveat is that this is a retrospective study in patients with a good performance status (Eastern Cooperative Oncology Group 0–2) and well enough to travel to the NCI at the time of enrolment in a single-institution trial. To our knowledge, this is the first report to show the time to first metastatic disease from ADT in D0 patients. This has implications both for the time to symptomatic disease and for anticipating the time to needing palliative therapy, and for designing new clinical trials. Also, given the morbidity of ADT , the risks and benefits of ADT must be weighed carefully in D0 patients. The higher Gleason score profile in the present D0 cohort compared with the D2 cohort might suggest that practitioners tend to give ADT to D0 patients with more aggressive disease. Factors that were predictive of a higher likelihood of metastatic recurrence in the study by Pound et al. were a Gleason score of 8–10, PSA recurrence within 2 years of RP and a PSA doubling time of < 10 months, although only the Gleason score and PSA doubling time remained predictive with a longer follow-up
The present findings shed new light on the biology and natural history of AI prostate cancer and in considering secondary systemic therapy after the failure of ADT. In view of the timeline of endpoints of biochemical failure, first metastatic disease and survival after ADT, it is not known what the secondary treatment should be after AI prostate cancer. One could consider the study by Fujikawa et al., in which only those men with AI prostate cancer who initially responded well to ADT responded to flutamide as a secondary therapy. Men who did not regain a normal PSA level with ADT had a PSA response to flutamide that was significantly worse than men whose PSA became normal with ADT. Given the recent results showing a survival benefit with chemotherapy, this should be given more consideration for secondary systemic therapy to men with AI prostate cancer who had a suboptimal response to ADT [13,14].
The study by Fujikawa et al. might also indicate a difference in the biology of AI prostate cancer in men who initially respond well to ADT compared with those who did not. It is clear that growth factors have a role in activating the androgen receptor in AI prostate cancer . However, it has been shown that a low level of androgen-receptor ligand is also required , and might be sufficient with castrate levels of testosterone. Patients who initially respond well to ADT and tend to respond well to antiandrogen(s) better fit the picture of this biology. Other factors might operate for patients who do not respond to testosterone-lowering therapy or antiandrogens. The biological difference between these types of patients might have implications for future novel therapies that target the androgen receptor [17,18]. While the present study helps to characterize the natural history of AI prostate cancer, further study is required to elucidate the biological basis behind individual variations in the course of disease and in the response to therapy.
In conclusion, we describe the natural history of AI prostate cancer in D0 patients who eventually developed metastasis and D2 patients on two clinical trials at the NCI. In the D0 cohort the median times from ADT to AI, to first metastatic disease, and the median survival duration, were 19.3, 37.8 and 89.0 months, respectively. The results in both cohorts confirm the longer than expected survival with AI prostate cancer, and to our knowledge are the first to show the time to metastatic disease from the first increase in PSA level after ADT. These results have important therapeutic and prognostic implications for advanced prostate cancer and for the design of clinical trials in this setting.
This research was supported in part by the Intramural Research Program of the NIH, and the National Cancer Institute.