Prostate carcinoma

Defining therapeutic objectives and improving overall outcomes


  • Howard I. Scher M.D.

    Corresponding author
    1. Genitourinary Oncology Service, Department of Medicine, Memorial Sloan-Kettering Cancer Center and Department of Medicine, Joan and Sanford Weill Medical College of Cornell University, New York, New York
    • Genitourinary Oncology Service, Department of Medicine, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue H-905, New York, NY 10021
    Search for more papers by this author
    • Fax: (212) 988-0851

    • Dr. Scher has received grant or research support from AstraZeneca, BZL Biologics, Bristol-Myers Squibb, Novartis Pharmaceuticals, and Millennium Pharmaceuticals. He has acted as a consultant for Bristol-Myers Squibb, Conforma Therapeutics, ProQuest Investments, and Wyeth-Ayerst Pharmaceuticals. He is a member of the Speaker's Bureau for AstraZeneca and is a major stock or investment holder in Eli Lilly (stock), Cell Therapeutics (stock), Vidamed (stock), ProQuest (options), Conforma (options), and Genta (options). Dr. Scher has received other financial material support from CaP CURE.


The development and progression of a prostate carcinoma from prediagnosis to death can be characterized as a series of clinical states. The states are milestones that can be used to assess prognosis, define therapeutic objectives, and assess outcomes. The antitumor effects of hormone therapies and cytotoxic agents in patients with prostate carcinoma are placed in context along with the bidrectional tumor-host interactions that contribute to the growth and resistance of osseous lesions. Identifying the factors that contribute to the growth of the disease at different points in the illness has lead to novel, systemic approaches. Proving the benefit of these approaches requires a series of unique trials with unique endpoints relevant to the clinical state of the patients and the specific therapy under evaluation. Cancer 2003;97(3 Suppl):758–71. © 2003 American Cancer Society.

DOI 10.1002/cncr.11151

Prostate carcinoma presents a paradox; a high prevalence in the population, only a proportion of whom have clinically significant disease, and a death rate that remains high. Often, the diagnosis is made in older men with comorbid conditions that pose a greater threat to the individual compared with the malignancy itself. Thus, the need to make a diagnosis is based on the risk that clinically significant disease is present in the gland. When a diagnosis is established, the need for treatment is based on the probability that the disease identified will affect the quality of a patient's life in an adverse way and/or will shorten life expectancy. Unfortunately, for many men, it does: Of 198,500 men who were diagnosed with prostate carcinoma in 2001, 31,200 men succumbed to their disease.1 This demonstrates the heterogeneity of a disease that can range from clinically irrelevant to life-threatening.

This range of prognoses mandates a different approach to the treatment of patients with prostate carcinoma compared with the treatment of patients who have other malignancies. For most malignancies, once a diagnosis is established, the paradigm is to treat patients early and often. The primary therapeutic objective is to eliminate the disease completely, a prerequisite for long-term cure. In contrast, there are situations in the management of patients with prostate carcinoma in which slowing the growth rate of the tumor can be tantamount to a cure, because the nonprostate carcinoma-related mortality exceeds that of the carcinoma itself.2, 3 Even when a recurrence is documented, immediate therapy is not always necessary. Rather, as it is at the time of diagnosis, the need for intervention is based on the tempo of the illness as it unfolds in the individual relative to the risk-reward ratio of the therapy under consideration. Naturally, were it possible to eliminate the disease completely with minimal or no adverse effects on the host, then one certainly would do so. However, once metastases are identified, complete remissions are rare regardless of the therapy employed, particularly when the disease has invaded the skeleton, a common site of metastatic spread.

There are other unique challenges to developing treatments for patients with prostate carcinoma. The first challenge relates to the fact that the primary manifestations of progressive, metastatic disease are an increase in the level of prostate specific antigen (PSA) and/or osseous lesions. Discretely measurable tumor masses occur infrequently. Thus, the primary measures of response simply do not apply. The second challenge is that, regardless of disease extent, a rise in the PSA level is a sign of progression. Unfortunately, the converse is not always true, because declines in the PSA level do not necessarily mean that tumor growth has been slowed4 and changes in the expression of the PSA gene can occur independent of an effect on growth and proliferation. The third challenge relates to the fact that that, for many individuals, the disease follows a protracted course, making it difficult to prove that there has been a change in its natural history or that a treatment has benefited a group of patients within a particular disease state to a degree that would affect standards of care. The last consideration is a mixed blessing and relates to the fact that, as our understanding of the pathogenesis and progression of the disease increases, the number of agents available for testing has increased to the point that there are too many choices to test, and not too few. How to prioritize one approach compared with another remains a significant challenge.

This discussion is focused on several aspects of management for patients with prostate carcinoma: how to define therapeutic objectives in a disease in which the clinical course from diagnosis to death can span a decade or more; how to gauge the efficacy of treatments that are available currently; how to treat prostate carcinoma that has spread to the skeleton, which is a relative sanctuary site; and how to improve outcomes using clinical insights along with new biology.

How to Define Therapeutic Objectives

The beneficial effects of a treatment or approach can be assessed only in the context of defined therapeutic objectives. These objectives will vary, depending on whether the disease is early or late, whether the patient is symptomatic or asymptomatic, and whether the risk of death is imminent or long term. With these considerations in mind, we have described a clinical states model for progression that provides the clinical framework to define objectives and to assess outcomes.5 Figure 1 shows that the states are initial prostate evaluation, no diagnosis of malignant disease; clinically localized disease; rising PSA; clinical metastases noncastrate; and clinical metastases castrate. The figure shows competing causes of death among patients with prostate carcinoma: disease-related causes versus nondisease-related causes. The model was based on points in the natural history that are recognized easily, allowing patients and their physicians to determine where an individual is placed in the clinical spectrum of the illness. In practice, once the patient or his physician has identified the state, a prognostic assessment is made, and therapeutic objectives are defined. The assessment includes an evaluation of which symptoms may or may not be present and, if they are not present, then determining the probability that a patient will develop disease progression that will lead to symptoms or death from disease. It follows that the objectives for each state will differ, as will the endpoint(s) used to assess clinical effects.

Figure 1.

Clinical states model for disease progression. PSA: prostate specific antigen; QOL: quality of life. Modified from Scher HI, Heller G. Clinical states in prostate cancer: towards a dynamic model of disease progression. Urology. 2000;55:323–327.

In this context, it is easier to understand why some progressing malignancies do not require an immediate intervention. In some instances, the patient may not be at risk for developing metastases or symptoms for many years, and the risk of death from nondisease-related causes may greatly exceed that of the disease. Indeed, until a patient has progressive, castration-resistant disease, he is unlikely to die of his cancer. Therapeutic objectives for each state also are listed in Figure 1. They include prevention for patients who do not have a diagnosis but who may be at high risk for the development of the disease. For patients who are diagnosed with localized tumors, the objective is to define which patients can be cured by modalities directed at the prostate alone, which patients have tumors of such low biologic potential that they can be observed safely, and which patients have such aggressive disease that they will require combined modality approaches both to eradicate the tumor locally and to eliminate micrometastases.

For patients who have failed primary therapy and have a rising PSA level and no detectable disease on imaging studies (the so-called state of a rising PSA), the objective is to prevent the development of metastases that can be detected on a scan: which represents a transition to a more lethal point in the illness.3 For patients who have metastatic disease at the time of diagnosis (that is, detectable disease on an imaging study) the objective is to delay progression and to prolong the response to first-line hormonal therapy. Finally, for patients who have progressive disease after undergoing castration, the primary objective is to improve or maintain quality of life as well as to prolong survival.5

How Effective Are the Standard Approaches?

The death rate and the absolute number of deaths from prostate carcinoma have decreased over the past few years.1 Whether this is the result of early detection strategies based on PSA levels and earlier treatment of the disease or on more effective systemic therapy is unclear. For patients with metastases who are destined to die from prostate carcinoma, morbidity and mortality will be reduced only when more effective systemic therapies are identified. However, before we consider new therapies, it is essential that we consider what we already have and the standards that have been established to manage the disease.

For a patient with a clinically localized tumor, the primary treatment options are surgery or radiotherapy, with the latter administered by an external beam approach, implantation of radioactive seeds alone, or a combination of the two. A small percentage of patients are considered for watchful waiting or deferred therapy if the predicted probability of progression to symptoms or death from disease is low relative to the patients' life expectancy. How best to identify such patients and to characterize their tumors are areas of active investigation. Most patients are advised to undergo treatment, in which case, there are two issues to consider. The first consideration is management of the primary tumor or how to eliminate the disease in the primary site with the least morbidity to the patient. For a patient with prostate carcinoma, the latter include maintaining potency, preserving continence, and normal bowel function. The second consideration is an assessment of the risk of metastatic spread. It follows that, for patients who are destined to develop metastases, ultimate cure (defined as the elimination of all tumor cells) will require the integration of a treatment that acts systemically.

Patients with a rising PSA level after surgery or radiation therapy are a unique subgroup. In these patients, the source of the PSA elevation can be either persistent or recurrent disease in the primary site or a systemic recurrence. The distinction is important, because, if the disease is still localized, then additional therapy to the primary site may effect a cure.6, 7 Unfortunately, the imaging techniques currently available are not of sufficiently high specificity or sensitivity to make this distinction with certainty. To address this issue, a number of algorithms that incorporate such factors as the characteristics of the primary tumor, the time to PSA recurrence, and PSA doubling time have been proposed to help predict whether the recurrence is local or systemic.8–10 None are 100% predictive.

In most patients, the rise in PSA level represents a systemic recurrence. Management of these patients is controversial, because a significant proportion of patients have a very low probability of progressing to clinically detectable disease, the development of symptoms, or death from disease. For example, in a series of 1997 patients who underwent radical prostatectomy, of the 300 patients who transitioned to the state of a rising PSA level, the median time to detection of metastatic disease was 8.0 years (range, 1.5–13.0 years).10 It follows that, for a patient who is not destined to develop metastatic disease or symptoms, any therapy is overkill, and toxicities are unacceptable.

What remains controversial is the risk of progression in a given time frame that would lead a patient to accept a given therapy or a physician to recommend it, assuming the predictions are accurate. Most likely, the recommendations would vary for a patient who has a 20% risk of a clinically significant event at 7 years compared with a patient who has a 70% risk at 2 years. Attempting to establish appropriate cut-off values is controversial and beyond the scope of this discussion. These caveats notwithstanding, there is a trend toward the use of androgen-deprivation therapy for patients with PSA levels that continue to rise, if for no other reason than to eliminate the anxiety associated with the knowledge that the disease is progressing.11

There is little controversy in the management of a patient who presents with metastases, who develops metastases after a local failure, or who has symptoms of the disease. For these patients, androgen ablation is the standard.12, 13 This is achieved by blocking one or more points on the hypothalmic-pituitary-gonadal signaling axis by eliminating testicular androgens alone or by eliminating both testicular and adrenal androgens (Fig. 2). The concept of blocking the action of androgens to retard prostate carcinoma growth dates back to the 1940s, when Huggins and colleagues demonstrated that lowering testosterone levels by performing a surgical orchiectomy or administering exogenous estrogens could provide palliation for patients with symptomatic disease.14, 15 Subsequently, a number of medical approaches were introduced, including one of the first targeted drugs, estramustine,16 and the nonsteroidal antiandrogens, which block the binding of testosterone to the androgen receptor.17, 18 Currently, the most widely utilized medical therapy includes a gonadotropin-releasing hormone (GnRH) agonist/antagonist alone or in combination with an antiandrogen. GnRH agonist/antagonists produce an initial surge in luteinizing hormone (LH) along with a rise in testosterone that is followed by down-regulation of the LH receptors in the pituitary, an inhibition of LH release, effecting a chemical castration. These agents were approved on the basis of studies that they showed a therapeutic equivalence to surgical castration19 and an improved safety profile relative to exogenous estrogens. The latter included a significant reduction in cardiovascular events associated with estrogen administration.20 Not surprisingly, most patients prefer the medical approach rather than the surgical approach.21 Because the initial rise in testosterone may exacerbate the disease, GnRH agonists/antagonists typically are administered in combination with an antiandrogen to block the flare.22

Figure 2.

This chart of the hypothalamic-pituitary-gonadal axis illustrates the points of interruption of androgen action. ACTH: adrenocorticotropic hormone; AR: androgen receptor; CRH: corticotropin-releasing hormone; DHEA: dehydroepiandrosterone; DHT: dihydrotestosterone; FSH: follicle-stimulating hormone; GnRH: gonadotropin-releasing hormone; LH: luteinizing hormone; LHRH: luteinizing hormone-releasing hormone; SHBG: sex hormone-binding globulin; T: testosterone.

The combination of an antiandrogen and a GnRH analog has the additional potential to block the effects of adrenal androgens, which can contribute anywhere from 5% to 45% of the residual androgens present in tumors after patients undergo surgical castration alone.23 The question of weather the antitumor effects of a combined or maximal androgen blockade approach were superior to castration alone or GnRH monotherapy occupied the field for many years. Ultimately, several thousand patients were enrolled on trials around the world with seemingly conflicting results. Several meta-analyses have been performed showing that antiandrogens do not add to the antitumor effects of surgical castration.24–27 However, in trials that compared a GnRH analog monotherapy with the combination of a GnRH analog and an antiandrogen, modest superiority was shown for the combination.24 What remains unclear is whether the superiority reflects a true effect on residual adrenal androgens, as postulated, or blockade of the potentially negative effect of the unopposed androgens that result from the initial LH surge.28 Consequently, whether the duration of antiandrogen therapy should be short term (1–3 months) or continuous remains controversial.

Other hormonal approaches that have been investigated recently include high-dose bicalutamide and the GnRH hormone antagonists. High-dose bicalutamide therapy (150 mg) is associated with fewer hot flashes, less of an effect on libido, less muscle wasting, fewer personality changes, and less bone loss.29, 30 Gynecomastia remains a significant problem with this form of therapy. Also under study are the GnRH antagonists, which produce medical castration significantly more rapidly compared with combination therapy and avoid the testosterone surge characteristic of agonist therapy.31 These agents still are investigational in the United States.

In general, agents that lower testosterone levels to a castrate range produce similar antitumor effects. The antiprostate carcinoma effects of castration vary by extent of disease, whereas the methods used to assess treatment effects in an individual depend on the manifestations of the disease in that individual. These manifestations range from a rising PSA level alone to any combination of a rising PSA level with lymph node metastases, bone metastases, and/or visceral disease. Symptoms may or may not be present. In rare instances, radiographic and clinical progression can be documented without a change in PSA level. Not all symptoms are present at any one time. Many of these manifestations cannot be assessed easily using the conventional criteria of response that have purposed to evaluate the antitumor effects of oncolytic therapies in patients with malignant disease.4, 32 Thus, it is best to consider outcomes on the basis of the change in each disease manifestation separately and to report them separately. In patients with metastatic disease that is detectable on an imaging study, 60–70% of patients with abnormal PSA levels will show a normalization in PSA to a value <4 ng/dL after castration, 30–50% of measurable tumor masses will regress by ≥ 50%, and > 60% of patients with symptoms will show palliation, whether they are urinary or osseous in origin. However, in serial bone scans, only 30–40% of scans actually improve, and the majority remain stable.33

These data show the noncurative nature of the disease once it has spread with hormonal therapy alone. The complete elimination of disease in any site is rare. This applies not only to tumors that have metastasized but also to tumors that are limited to the primary site. For example, when hormones are used as neoadjuvant therapy prior to surgery, the proportion of prostates that are rendered tumor free at prostatectomy is <5%,34, 35 demonstrating that cells that are resistant to castration are present even at the time the diagnosis is established.

There is considerable controversy regarding the timing of hormonal treatment. Some advocate androgen ablation prior to primary treatment of a localized tumor, others advocate androgen ablation as an adjuvant to primary therapy, and others are evaluating its role in lieu of primary treatment. In the setting of a rising PSA level, some advise immediate treatment, others defer treatment until metastases are identified on an imaging study, and some wait until the patient has developed symptoms of disease. Data show that hormone therapy can delay the development of metastatic disease, although definitive data showing a survival benefit are lacking.36 The issue centers around the fact that many men with a rising PSA level have a very low risk of progression to the state of clinically detectable disease; as such, their risk of developing symptoms or dying from prostate carcinoma is quite low. For individuals with a low risk of progression, it is difficult to justify potentially toxic therapies.

Data in support of early hormone therapy date back to the early Veterans Administration Cooperative Urological Research Group studies showing, in randomized comparisons, that diethylstilbestrol or orchiectomy could delay the development of metastatic disease in patients with Stage C disease.36 Other studies that often are cited in support of the approach include a trial by Bolla and colleagues in which patients with localized disease were randomized to receive radiotherapy alone or radiation therapy and 3 years of androgen ablation therapy.37 Their results showed a survival benefit, although that trial was criticized for the poor outcome in the control group. In a separate trial, patients with lymph node positive disease at prostatectomy were randomized to receive castration (surgically or medically) or observation.38 The results from that study also showed a benefit in terms of survival, although the confidence intervals around the 5-year and 8-year survival distributions overlapped that for placebo-treated patients (P = 0.02): After a median of 7.1 years of follow-up, 16% of men who received immediate antiandrogen treatment had died, compared with 36% of men in the observation group (P = 0.02). The cause of death was prostate carcinoma in 7% in the immediate therapy group and 33% in the observation group (P < 0.01); whereas, at the time of last follow-up, 77% of men in the immediate therapy group and 18% of men in the observation group had no evidence of disease, including undetectable serum PSA levels (P < 0.001).

In a Medical Research Council study, 938 patients with locally advanced or asymptomatic, metastatic prostate carcinoma were randomized to receive either immediate treatment (orchidectomy or LH-releasing hormone analogue) or the same treatment deferred until an indication occurred. Treatment was commenced for local progression almost as frequently as treatment for metastatic disease. Compared with patients who were treated with deferred therapy, patients who were treated with early therapy were less likely to progress from M0 disease to M1 disease (P < 0.001; two-tailed), were less likely to develop pain (P < 0.001), and were less likely to die of prostate carcinoma.39

More recently, results from the Early Prostate Cancer Program were reported. In the three complimentary trials, patients with localized disease were randomized to receive Casodex (150 mg) or placebo.40, 41 The primary endpoint was objective clinical progression that included detectable disease in soft tissue or the documentation of bone metastases at 2 years. Combining the results of the three trials, the proportion of patients who developed osseous metastases within 2 years was 9% in the bicalutamide group and 13.8% in the placebo group. This represented a hazard ratio reduction of 0.58 (range, 0.51–0.66; P < 0.001). No effect on survival was demonstrated, reflecting the short follow-up and the inclusion of good-risk patients in the treatment program. The reduction in the rate objective progression metastases, which represents a transition to a more lethal form of the illness, provides an additional rationale for early treatment.

Those trials have been criticized because of the small sample sizes,37, 38 the deferral of treatment in the control group for too long,37, 39 short follow-up, and the inclusion of too many patients with good-risk features.41, 42 As such, they do not establish unequivocally that immediate therapy will prolong life. In part, this is because what was considered early at that time (before the widespread application of PSA-based monitoring) is not the same as what is considered early today. In the past, disease progression was determined by detecting disease by physical examination or an imaging study, which often required a significant change in the size of a lesion. Currently, progression is detected much earlier through the use of PSA-based monitoring schemes.

A question that remains is whether the degree of benefit has been proven convincingly, and, if it has, then whether it is sufficient to justify treatment of all patients. Part of the difficulty in making this determination is that the trials on which the evidence is based, with few exceptions, were small. This contrasts with the experience in the management of patients with breast carcinoma, for which the European Breast Cancer Trials Cooperative Group derived treatment recommendations based on the outcomes of approximately 30,000 women who were enrolled in 55 randomized trials that evaluated tamoxifen or no tamoxifen as an adjunct to primary treatment. Those results showed that the reduction in the death rate between control patients and tamoxifen-treated patients was directly proportional to the duration of therapy. Patients who were treated for 1 year, 2 years, and ≥ 5 years enjoyed a reduction in the rates of metastatic recurrence and mortality of 21% and 12%, 29% and 17%, and 47% and 23%, respectively, after 10 years of follow-up. A highly significant trend toward a greater effect with longer treatment (chi-square test, 52.0; 2 P < 0.00001) also was noted. The results observed in patients with prostate carcinoma were similar, although the number of patients enrolled and the length of follow-up were significantly less.43

These data suggest that early therapy can prolong life. It remains unanswered whether deferring treatment until the time that a PSA recurrence is detected, rather than treating at the time of primary therapy, will yield similar long-term outcomes. In this way, patients who have a low risk of recurrence and progression can avoid the symptoms associated with androgen deprivation, including hot flashes, loss of libido, impotence, gynecomastia, weakness, loss of muscle tone, anemia, personality change, gastrointestinal effects, and bone loss. Once again, determining the level of risk, in terms of both absolute probability and time frame, relative to the level of benefit cited would indicate whether the toxicities associated with hormonal therapy would be acceptable in an individual patient.

Once metastases can be detected on physical examination or on an imaging study, clinical death rates are continuous over time. Overall survival rates vary by extent of disease.28, 44, 45 Although the debate continues over which specific hormone therapy is optimal, the central issue is that hormone therapy is not curative.

For patients with tumors that have progressed after castration, a range of options are available. Patients who are on antiandrogen should first undergo therapy withdrawal.46 Other patients are treated with second-line and third-line hormone, including inhibitors of adrenal androgen synthesis (such as ketoconazole, glucocorticoids, and estrogen-based approaches).47 Other patients are considered for cytotoxics, and still other patients are considered for investigational approaches.

Combined mitoxantrone and prednisone is currently approved by the U.S. Food and Drug Administration for the palliation of pain from osseous metastases.48, 49 More recent data suggest that, in fact, prostate carcinoma is responsive to chemotherapy; and, using taxane-based approaches, responses in excess of 50% have been seen when assessed by changes in measurable disease ≥ 70% using the endpoints of a decline > 50% in the PSA level; demonstrable, objective improvements on bone scans; and a response duration of ≥ 6 months.50–53 Phase III trials are ongoing: However, like newly diagnosed patients with disease that has spread to the skeleton and who have received androgen-ablation therapies, the majority of patients with bone metastasis who are treated with cytotoxic agents first develop disease progression in bone. To improve outcomes will require treatments that affect the cells that resist, survive, and ultimately proliferate despite castration.

This issue of resistance to castration can be addressed in part by considering the effects of androgen deprivation on prostate carcinoma growth and proliferation. Androgen deprivation or blockade, as discussed above, results in a decline in PSA levels, regression of measurable tumor masses, followed by a period of clinical quiescence in which the tumor appears static, followed by a rise in PSA level, proliferation of the tumor, and clinically detectable tumor regrowth. These specific phases are difficult to assess histologically in the same patient, because repeated sampling of tumor is not a part of the routine management of the disease. To address this, we and others have worked with the CWR22 human prostate carcinoma xenograft model developed by Pretlow and colleagues.54, 55 These tumors have some features of the human condition, in that they produce PSA and show objective regression after castration that, after a variable period, is followed by a rise in PSA level and an increase in tumor size. In this model, we studied changes in cell cycle regulatory proteins. Early events included a decrease in androgen receptor expression, a decrease in proliferation assessed by Ki-67 staining, and an increase in p53 and p21. Intermediate-to-late events included increases in the cyclin kinase inhibitory proteins p27 and p16. But it was interesting to note that changes associated with apoptosis were not observed. Tumor regrowth occurred after a variable period, as expected, and a histologic examination of regrowing tumors showed increased levels of mdm2 and cyclin D1.56 The results indicated that tumor regression was due to cell cycle arrest rather than apoptosis and that the emergence of androgen independence is associated with a release from the arrest. These results have therapeutic implications (vide infra).

The outcomes of patients who receive standard therapies, whether they are hormonal or cytotoxic, show that a significant proportion of patients respond, although complete responses are rare. Death rates vary by extent of disease but remain continuous over time. Tumor cell sensitivities vary not only by the site of spread (with progression rare in soft tissue sites) but also once the disease has been established in bone and when the first progression occurs in bone.57 Thus, to improve overall outcomes, treatments are needed that will eliminate nonproliferating cells, prevent regrowth, or promote cell death.

Why the Focus on Bone?

The development of clinically detectable disease on an imaging study represents the transition to lethal prostate carcinoma. Bone metastases in particular are the sole site of spread in ≥ 80% of patients who develop clinical metastases.58 Untreated or unchecked, they can produce some of the most feared complications of prostate carcinoma, including pain, immobility, and hematopoietic and spinal cord compromise. Equally important is that, once they are established, metastases are relatively resistant to conventional therapy and rarely can be eliminated completely. Most tumors that spread in the skeleton are osteolytic. In contrast, prostate tumors are osteoblastic. Nevertheless, osteoclast activation remains a factor that contributes to the continued growth of a tumor in bone.

It is interesting to note that the disease spread in bone does not follow discrete anatomic or vascular patterns: It follows the distribution of the bone marrow in an adult.59 This was shown in a analysis of the sites of spread in patients with a small number of abnormalities on bone scans (Fig. 3). Once a tumor cell has gained access to the circulation, the establishment of a metastatic focus in bone involves multiple steps. These steps include adhesion to endothelial cells in the bone marrow and migration through fenestrations in the endothelial cell layer. This may be driven in part by a chemoattractant gradient of bone marrow and stromal-derived growth factors. These factors also may promote the proliferation of tumor cells in the bone marrow space and may contribute to treatment resistance.60 When a metastasis is established, tumor cells and bone marrow-derived cells develop a bidirectional interaction that protects the epithelial cells and promotes tumor cell survival and proliferation.61 Although the disease is osteoblastic radiographically, osteoclast stimulation and activation still are present to a significant degree, showing that the bone remodeling process is not uncoupled; rather, there is a shift in the balance in favor of bone growth. Indeed, markers of increased bone turnover can be detected in the blood and urine of patients with progressing disease.62 It is hypothesized that the resorptive process itself, under the direction of osteoclasts, promotes the release of factors that amplify the metastatic and invasive process.61, 63–65 More recent data suggest that the protease action of PSA may increase the activation of functional signaling molecules adjacent to tumor that further contribute to tumor cell growth and proliferation.64 Figure 4a shows an increase in both osteoblasts and osteoclasts within a bone metastasis. Figure 4b shows the nature of bone formed around the tumor.

Figure 3.

The distribution of 136 prostate carcinoma skeletal metastases in 27 patients with minimal disease. ANT: anterior; POST: posterior. Reprinted with permission from Imbriaco M, Larson SM, Yeung HW, et al. A new parameter for measuring metastatic bone involvement by prostate cancer: the bone scan index. Clin Cancer Res. 1998;4:1765–1772.

Figure 4.

Serial changes in the bone scan index (BSI) over time in patients with progressive prostate carcinoma in bone. (a) Serial bone scans with corresponding BSI measurements over a 13-month interval after the development of metastases. (b) Serial BSI measurements for a group of patients with progressive castration-resistant disease in bone (c) Chart indicating cyclin D1 and Ki-67 expression levels in primary tumors and in metastatic (castrate) tumors. Modified from Sabbatini P, Larson S, Kremer AB, et al. The prognostic significance of extent of disease in bone in patients with androgen-independent prostate cancer. J Clin Oncol. 1999;17:948–957.

An important issue is defining when a metastasis is a metastasis. Assuming that all imaging studies, such as plain films, radionuclide bone scans, positron emission tomography (PET) scans, computerized tomography (CT) scans, and magnetic resonance images (MRIs), are negative for metastatic disease, does the presence of tumor cells on a histologic examination of the bone marrow indicate metastases? Do molecular-based detection techniques, such as polymerase chain reaction (PCR) amplification for PSA or prostate specific membrane message, imply that metastases have taken hold, or does a positive test simply reflect tumor cells that are passing through the blood stream to bone marrow sinusoids in the absence of a metastatic foci?67 Several groups have show that there is high-frequency detection of tumor cells within the bone marrow and using PCR-based techniques. In patients with seemingly localized tumors, ≥ 50% will have cells detectable in the bone marrow that harbor the message for PSA, suggesting that these cells in fact are present.68–70 This has therapeutic implications in defining who is at risk for developing metastatic disease.

The spread of the disease to the bone marrow exposes tumor cells to a rich microenvironment that can promote the growth of the cell. This phase of the illness often is associated with rapid proliferation. Figure 4a shows the serial changes in bone scan index (BSI) over a 13-month period in a patient with progressive disease after castration. The BSI measures the proportion of the adult bone marrow that is involved by tumor based on the normal bone marrow distribution in the adult.59 The BSI, as shown, increased from 0.0% to 7.7%. Figure 4b plots serial changes in the BSI over time for a group of patients with progressive disease. In this cohort, the estimated doubling time in the bone marrow phase of the disease was 43 days.71 This was confirmed by immunohistochemical studies that examined the proportions of tumors overexpressing cyclin D1 at the time of diagnosis compared with tumors at the time metastatic disease developed after castration (12% vs. 68%, respectively), whereas the Ki-67 index was 12% in the primary site compared with 45% in patients with castration-resistant bone metastases.72 Determining whether the high proliferation rate reflects a tumor that is more sensitive to cytotoxics certainly would indicate whether earlier interventions may be useful.

How do we assess treatment effects in a nonmeasurable site, such as bone? Unfortunately, bone scans are not specific for tumors, because these scans measure the secondary effects of the tumor on the skeleton. Consequently, bone healing in response to successful therapy may result in a worsening appearance of a bone scan, even as PSA levels decline and symptoms of disease abate. This pseudoprogression should not be misconstrued as a treatment failure and simply reflects the blastic healing response associated with successful therapy. Typically, a subsequent scan performed at 6 months will show improvement.33 It is also noteworthy that, even if all disease is eliminated in bone, it may take ≥ 2 years for the complete resolution of abnormal imaging findings.

More recently, our group has started looking at PET, which uses a radioactive form of an important biochemical to trace the metabolism of a natural form of that chemical through imaging. Biologically relevant traces under study include 18-fluorodeoxyglucose, 11-C methoinine, choline derivatives and I24 iodine. PET imaging visualizes disease in soft tissue and in bone simultaneously and in a more quantitative manner. This is illustrated in Figure 5, which shows bone scans and PET scans from a patient with metastatic disease in bone who was treated with hormonal therapy. The PSA values obtained at the time of the scans are also included. There was no change in the bone scan, as illustrated, whereas there was a marked decrease in fluorodeoxyglucose accumulation by PET.

Figure 5.

Sequential prostate specific antigen (PSA) levels, bone scans, and fluorodeoxyglucose-positron emission tomography (FDG-PET) scans in a patient who was treated with combined androgen blockage. ANT.: anterior; POST.: posterior. Courtesy of Dr. Steven Larson (Memorial Sloan-Kettering Cancer Center, New York, NY).

To assess the role of PET in patients with metastatic prostate carcinoma, we first evaluated this modality in patients with rising PSA levels and progressive disease in bone and/or soft tissue on CT scans, MRIs, or bone scans. In a lesion-by-lesion analysis of osseous sites visualized on bone scintigraphy and on PET scans, 71% of lesions were seen on both modalities, 23% of lesions were seen on bone scans and not on PET scans, whereas 8% of lesions were seen on PET scans but not on bone scans. It is interesting to note that 97% of the PET negative lesions remained unchanged on bone scans with serial follow-up, whereas all of the PET positive locations ultimately became positive on bone scans. Although that study lacked histologic confirmation of the findings, it suggests that, in formulating treatment decisions, it is important to consider previous as well as current imaging studies. It also supports the presence of a bone marrow phase of the disease, in this case visualized only on a PET scan and not the bone scan, that may prove to be an indication for treatment.73

It is interesting to speculate that a tumor cell in the bone marrow that has not elicited a host response is more sensitive to treatment. If this was the case, and if it could be proven that such cells ultimately lead to established tumor foci, then earlier intervention would not only be justified but may be proven more effective. This, in turn, would change management paradigms in both the noncastrate setting and the castrate setting from a defensive posture (waiting for additional spread or symptoms of disease) to an offensive strategy in which we seek not only to prevent the spread of disease or symptoms but to eliminate the disease entirely, a prerequisite for long-term cure.

Improving Outcomes using Clinical Insights in the New Biology

The clinical behavior of a tumor that has remained localized differs from a tumor that is metastatic. Similarly, the fact that a tumor cell metastases to the skeleton rather than to a lymph node are not accidental but are preprogrammed. Indeed, that bone metastases beget additional bone metastases, and not soft tissue lesions, has been recognized in the laboratory and in the clinic. There also are differences between tumors that are proliferating in a noncastrate environment versus a castrate environment. Each clinical phenotype has a biologic basis and a molecular basis that may be characterized and exploited therapeutically. Improving overall outcomes will require the convergence of several disciplines, including basic and cell biologists, chemists, pathologists, and experienced clinical investigators designing and completing unique trials with unique endpoints based on the state of disease under evaluation.

In this regard, it is essential to consider each clinical trial, whether it is a Phase I trial, a Phase II trial, or a Phase III trial, as part of a sequence of trials for which, if the primary endpoint is satisfied, clinical development continues. Ultimately, to change practice requires the demonstration in a prospective, randomized, Phase III trial that a new therapy has demonstrated clinical benefit for a patient population relative to the previously available standard. An improvement in survival is one measure, but not the only measure, of clinical benefit.

To date, no systemic therapy has been shown definitively to prolong the life of patients with prostate carcinoma, despite the fact there are treatments available that produce high response rates. Response, as measured by a change in PSA level, an objective measure of tumor regression, or relief of symptoms, is an important outcome in Phase II investigations, although it may not be a surrogate of clinical benefit or for a change in the natural history of the disease. To establish surrogacy requires a prospective, randomized comparison in which a positive clinical effect is observed, and the clinical effect is eliminated when the effect on the surrogate is factored in the result.74 However, before conducting large, comparative trials with definitive clinical endpoints, outcomes are needed to determine whether there is sufficient efficacy (Phase II) and to what degree, to justify a large-scale, Phase III study.75 The overestimation of benefits of a treatment leads to trials that may fail to detect important differences because sample sizes were too small.

In many diseases, the results for patients with late-stage disease are extrapolated to patients with early-stage disease. For patients with prostate carcinoma, the equivalent is to use results from castrate patients with metastatic disease for patients with localized disease. This may or may not be appropriate, depending on the agent under investigation. Many groups are evaluating novel cytotoxic therapies in this setting, and the results show that agents that target the cytoskeleton indeed do have significant antitumor effects. Other approaches are targeted toward the specific mechanisms that are associated with progression intrinsic to the tumor, such as endothelin antagonists;76, 77 mechanisms that emanate from the host, such as antiangiogenic approaches; and approaches directed toward the interaction of the tumor and the host (bone-seeking radiopharmaceuticals and cold bisphosphonates). The targets of relevance for tumors that represent one clinical state may not be relevant to another clinical state (Fig. 6).77 Consider two well-described targets, HER-2 and BCL-2. The former is an oncogenic signaling protein associated with proliferation78 that can activate the androgen receptor independent of ligand,79 whereas BCL-2 has antiapoptotic effects.80 In both targets, the proportions of tumors that express the protein increase significantly when comparing tumors from castrate patients with metastatic lesions from patients with progressive, castrate, metastatic disease.81, 82 It follows that a positive finding in a patient with late-stage disease would not be relevant to a patient with early-stage disease. Other alterations occur in p53 and the androgen receptor.83–88 These changes underscore the importance of the disease state when typing tumors to grade treatment selection and when testing biologic agents. In contrast, the frequency of expression of prostate specific membrane antigen89 is relatively constant across all clinical states, although the intensity of staining is greater in tumors from castrate patients with metastatic disease.

Figure 6.

Proportion of tumors that expressed prostate specific membrane antigen (PSMA), HER-2, and BCL-2 within and across different states of the disease. PSA: prostate specific antigen.

An additional consideration that is receiving increasing recognition is the importance of the timing of one intervention relative to another. This lead us to consider a functional time course classification of points of therapeutic attack (Fig. 7). The traditional approach is to evaluate Treatment A (whether it is hormone therapy or chemotherapy), assess the outcomes, and change to Treatment B when Treatment A no longer works (Fig. 7, Point 1), as noted above. This is the approach used with most cytotoxic agents. An alternative may be to develop a therapy that prevents the reemergence of proliferating, resistant cells (Fig. 7, Point 4) or a therapy that is designed to increase the apoptotic rate of a nonproliferating, stable, dormant tumor (Fig. 7, Point 3). Putative differentiating agents, such as the histone deacetylase inhibitors90, 91 or phenylbutyrate,92 are examples under study in the clinic. In addition, several groups are beginning to evaluate combinations of a hormone or cytotoxic agent with a second agent designed to increase apoptotic rates (Fig. 7, Point 2). Combinations of a hormonal agent or a cytotoxic drug along with an antisense to BCL-2 is one example,93, 94 and the same combination of ansamycin drug that produces the degradation of key signaling molecules is another.95 To eliminate nonproliferating cells, a number of vaccination approaches are under study.

Figure 7.

A time course classification for combining different types of approaches with androgen ablation based on the changes in size of established CWR22 xenografts. CAST: castration; SAC: time of tumor sacrifice. Modified from Agus DB, Cordon-Cardo C, Fox W, et al. Alterations of cell cycle regulators in prostate cancer: response to androgen withdrawal and development of androgen independence. J Natl Cancer Inst. 1999;91:1869–1876.

The clinical evaluation of these approaches requires unique trials with clinically relevant endpoints. In this regard, it has been shown that two bone-seeking radiopharmaceuticals, Metastron96, 97 and Quadramet,98 reduce the pain of skeletal metastases despite the lack of a survival benefit. It also has been shown that a systemic chemotherapy regimen of mitoxantrone and prednisone relieves pain and palliates symptoms of disease.48, 49 It has also been demonstrated that the cold bisphosphonate, zoledronate, palliates symptoms and reduces the frequency of significant skeletal events, such as new pain, the need for radiation therapy, and microfractures.99 Finally, in a unique design, investigators at the University of Texas M. D. Anderson Cancer Center evaluated a systemic chemotherapy regimen followed by a combination of a Metastron and doxorubicin in a bone consolidation approach.100 Their results showed both a reduction in skeletal events and the suggestion of a survival benefit. This approach is now undergoing a large-scale Phase III evaluation.

The outlook is changing. More therapies are available than in years past. Not all can or will be tested in the clinic, which mandates the development of strict criteria to determine the therapies (if any) that should be brought to trial; and, when there is a solid foundation on which to build. Through careful trial design, questions will be addressed expediently to show progress so that outcomes can be improved for more patients with the disease.

Clinical states help patients and physicians therapeutic objectives that are essential for improving the overall outcomes of patients with prostate carcinoma. The standard approaches that are available currently, including hormone therapy as well as cytotoxic therapy, may produce short-term benefits and provide building blocks on which to improve outcomes. The unique approaches that will be required to affect tumors that have metastasized to bone and the development of these approaches will be facilitated through novel diagnostic techniques as well as molecular profiling.