SEARCH

SEARCH BY CITATION

Keywords:

  • castration-resistant prostate cancer;
  • metastasis;
  • visceral;
  • lymph node;
  • bone;
  • liver;
  • lung;
  • patterns

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

Metastatic castration-resistant prostate cancer (mCRPC) most commonly metastasizes to the bone, and less commonly to nonosseous sites (eg, lymph nodes, liver, lung). With new therapies extending survival in mCRPC, it was hypothesized that the pattern of metastases is changing over time. The pattern of metastatic disease was evaluated in men with mCRPC, as reported in baseline characteristics of prospective clinical trials over 2 decades.

METHODS

This study identified all phase 2 and 3 therapeutic studies in men with mCRPC in PubMed and American Society of Clinical Oncology abstracts from 1990 to 2012. Studies were excluded if they did not report demographic data and sites of metastasis, or excluded patients with a specific site of metastatic disease (except brain). For each type of metastasis, weighted least squares linear regression models were used to evaluate temporal trends.

RESULTS

A total of 290 eligible studies (270 phase 2 studies and 20 phase 3 studies) involving 19,110 patients were identified. Between 1990 and 2012, the rate of nonosseous metastasis increased significantly at 1.6% per year (P < .0001), whereas the rate of osseous metastasis decreased at 0.5% per year (P < .0001). The rate of lymph node metastasis increased at 1.4% per year (P < .0001), but the rate of liver and lung metastasis remained relatively stable.

CONCLUSIONS

A notable change was found in the pattern of metastasis in patients with mCRPC. Because these evolving patterns may have important implications in treatment selection and prognosis, it is crucial that future clinical trials of patients with mCRPC define patients with a uniform reporting of nonosseous metastasis. Cancer 2014;120:833–839. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Prostate cancer is the most common nonskin cancer in men, with an estimated 1 in 6 lifetime risk.[1] Although better screening and treatments have resulted in improved outcomes, a subset of patients eventually develops metastatic disease. Androgen deprivation is highly effective in most patients, but eventually resistance develops, leading to a lethal phenotype known as metastatic castration-resistant prostate cancer (mCRPC). Historically, mCRPC has been recognized as a disease to manifest primarily in the skeletal system, resulting in complications such as bone pain and fractures.[2, 3] Unlike many other cancers, sites of disease such as lung, liver, and soft tissue have traditionally been less commonly encountered clinically.[4]

With a better understanding of prostate cancer pathogenesis and biology, a renaissance in drug development has occurred in the treatment of mCRPC.[5] In recent years, at least 6 therapies with distinctively different mechanisms of action have been approved by the US Food and Drug Administration (FDA) for the treatment of mCRPC after demonstrating a survival benefit in randomized phase 3 studies. Improvement in the control of other solid tumors, such as breast cancer, has altered the natural history of the disease with the emergence of new patterns of metastasis.[6] We hypothesized that a similar phenomenon has occurred in prostate cancer.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Search Strategy

We searched PubMed and the American Society of Clinical Oncology (ASCO) abstracts for the following terms: “castration-resistant prostate cancer,” “castrate-resistant prostate cancer,” “hormone-refractory prostate cancer,” “hormone-resistant prostate cancer,” and “androgen-independent prostate cancer.”

Systematic Study

Studies were restricted to phase 2 and phase 3 clinical trials published in English between January 1990 and June 2012. Eligible trials studied patients with CRPC, at least a subset of whom had metastatic disease. Extent of metastatic disease was captured as reported in the baseline characteristics of each study. If a study included nonmetastatic patients, the exact number of metastatic patients must have been reported. For the purposes of this study, metastatic was defined as stage IV by American Joint Commission on Cancer (AJCC) criteria. In addition, each study must have reported the number of patients with at least one specific site of metastasis.

Studies were deemed ineligible if they required or excluded a site of metastasis, with the exception of studies excluding brain metastases. Reviews, meta-analyses, retrospective studies, and duplicate studies were excluded. Due to the information required for the analysis, trials obtained from the ASCO search were included only if the full study details were available in a slide presentation. For each study, demographic data and study-reported sites of bone and nonbone metastatic disease (lymph node, visceral, soft tissue, liver) were recorded by one author. A subset of the data was assessed for quality control by a second author.

Statistical Analysis

For data analysis, studies were classified into “chemo-naive” and “prior-chemo” categories, based on whether patients had received prior cytotoxic chemotherapy. Specific types of chemotherapy were often unreported, thus precluding further analysis based on type of cytotoxic treatment. Data for bone, nonbone, lymph node, liver, and lung metastases were analyzed over 3 different time periods: 1990 through 1999, 2000 through 2012, and 1990 through 2012, because the number of approved therapies has increased significantly in the past decade. A separate analysis was performed for each type of metastasis.

In each analysis, the unit of observation was a study, the dependent variable was the percent of patients in the study with a specific metastatic site, and the independent variables were the calendar year of the study publication and the median age of subjects in the study. In the few instances in which median age was not available, mean age was used if available. In the regression analyses, each observation was weighted inversely proportional to the variance of the dependent variable, which means that for 2 studies with the same percentage of patients with a given site of metastasis, the study with the larger number of patients would be weighted proportionately more than the study with the smaller number of patients. In each analysis, regression lines were estimated with separate slopes for the prior chemotherapy and nonchemotherapy groups, and the slopes were tested for equality (ie, parallel lines). If the slopes were not significantly different from each other, a common slope was estimated for both groups, resulting in parallel lines with different intercepts. The difference in intercepts provides an estimate of the difference between chemo-naive and prior-chemo incidences when the lines are parallel.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The search results and reasons for manuscript and abstract inclusion and exclusion are detailed in Figure 1. Among 821 PubMed publications and 1153 ASCO abstracts, 290 studies reported data on patterns of nonbone metastatic disease and chemotherapy use and were included in the analysis. The baseline characteristics of patients enrolled on these studies are shown in Table 1.

image

Figure 1. Study flowchart shows the selection of eligible trials.

Download figure to PowerPoint

Table 1. Characteristics of the Study Population
CharacteristicNo. of Studies (%)
Year 
1990-199972 (24.8%)
2000-2012218 (75.2%)
Phase 
2270 (93.1%)
320 (6.9%)
Studies including patients with prior chemotherapy treatment 
Yes116 (40.0%)
No174 (60.0%)
Studies that reported specific sites of metastasis 
Bone180 (62.1%)
Nonbone172 (59.3%)
Lymph node155 (53.4%)
Liver90 (31.0%)
Lung79 (27.2%)

The rates of nonbone and lymph node metastasis significantly increased over the study periods (Figs. 2A and 3). The rate of increase in nonbone metastases was 1.6% (P < .0001) per year between the years 1990 and 2012; with 0.4% per year (P = .80) between 1990 and 1999 and 2.7% per year (P < .0001) from 2000 to 2012 (Table 2). The rate of increase in lymph node metastases was 1.4% per year between 1990-2012 (P < .0001), mainly attributed to an increase of 2.5% (P < .0001) per year between 2000 and 2012 (P < .0001). The incidences of nonbone (7.5%, P = .0039) and lymph node (8.9%, P = .001) metastases were significantly higher in the group of patients that had received prior chemotherapy compared to the chemo-naive group between 1990 and 2012 (Table 3).

image

Figure 2. Rates of metastasis, for the period 1990 to 2012, are shown for (A) nonosseous and (B) osseous disease.

Download figure to PowerPoint

image

Figure 3. Rates of metastasis, for the period 1990 to 2012, are shown for specific sites: (A) lymph nodes, (B) liver, and (C) lungs. Abbreviations: Mets, metastases; pts, patients.

Download figure to PowerPoint

Table 2. Patterns of Metastatic Castration-Resistant Prostate Cancer Metastasis, 1990-2012
Site of MetastasisAnnual Rate of Change Per Year (1990-2012)
Nonbone+1.6% (P < .0001)
1990-19992000-2012
+0.4% (P = .80)+2.7% (P < .0001)
Lymph node+1.4% (P < .0001)
1990-19992000-2012
−0.8% (P = .46)+2.5% (P < .0001)
LiverChemo-naive: −0.3% (P = .0193)
Prior-chemo: +0.2% (P = .2531)
1990-19992000-2012
+0.4% (P = .47) Chemo-naive: −0.25% (P = .1338)
 Prior-chemo: +0.37% (P = .0929)
Lung+0.1% (P = .41)
1990-19992000-2012
−0.7% (P = .31)Chemo-naive: −0.17% (P = .3954)
 Prior-chemo: +0.67% (P = .0532)
Bone−0.5% (P < .0001)
1990-19992000-2012
−0.2% (P = .73)−0.6% (P = .0012)
Table 3. Difference in Incidence of Metastasis, Prior-Chemo Versus Chemo-Naive
Site of MetastasisDifference in Incidence of Metastasis (Prior-Chemo vs Chemo-Naive)
  1. NE indicates “not evaluable.”

Nonbone+7.5% (P = .0039)
1990-19992000-2012
−0.3% (P = .96)+7.1% (P = .0121)
Lymph node+8.9% (P = .001)
1990-19992000-2012
+8.7% (P = .0452)+5.6% (P = .07)
LiverNE (Lines Intersecting)
1990-19992000-2012
−1.0% (P = .71)NE (Lines Intersecting)
Lung−0.04% (P = .97)
1990-19992000-2012
+1.6% (P = .54)NE (Lines Intersecting)
Bone+0.06% (p=0.96)
1990-19992000-2012
−1.8% (P = .26)+0.3% (P = .81)

The rates of liver and lung metastasis were relatively stable between 1990 and 2012 (Fig. 3). However, the trend for liver metastasis was significantly different based on prior chemotherapy treatment; unlike the prior chemotherapy group, there was a significant decrease in the chemotherapy-naive group (−0.3% per year, P = .0193). This trend was not seen for lung metastasis during the study period; the difference between slopes was not significant, nor was the difference between prior chemotherapy and chemotherapy-naive groups (−0.04%, P = .97).

Bone metastases decreased at a rate of 0.5% per year between 1990 and 2012 (P < .0001), and most significantly between 2000 and 2012 (−0.6% per year, P = .0012). The incidences of bone metastasis between chemotherapy-naive and prior-chemotherapy groups for 1990-2012 were not significantly different (0.06%, P = .96).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The skeletal system is the most common metastatic site for prostate cancer.[7] Although the mechanisms of metastatic spread in prostate cancer are not well understood,[8] ongoing efforts have led to a better understanding of the underlying biology for this phenomenon.[2, 3, 9-11] With availability of better treatments for mCRPC, patients are living longer and have more therapeutic options. In addition, the presence of nonbone metastases in men with mCRPC is now becoming more routine. This led to a hypothesis that the treated history of prostate cancer may be changing, with altered patterns of progression at soft tissue sites not historically seen.

To this extent, we sought to define the pattern of metastasis in patients with mCRPC. We found a significant increase in the rate of nonbone metastasis in phase 2 and 3 clinical trials of men with mCRPC. This increase was even greater after 2000, coinciding with a period of time of increasing availability of approved therapies, and the incidence of nonbone metastases was significantly higher in studies with patients who had prior chemotherapy. Interestingly, this trend can be mostly attributed to an increase in lymph node metastasis, because the rates of liver and lung metastasis remained relatively stable over the study period. One hypothesis is that selective pressure from increasing availability of novel therapies has altered the natural history of mCRPC. One such example is the increasing observation of the “neuroendocrine-phenotype,” a form of CRPC associated with aggressive soft tissue involvement, and is characterized by significant overexpression and gene amplification of AURKA and MYCN.[12] A similar trend was not seen with liver and lung metastasis, perhaps because those with large visceral disease burden are less likely to be candidates of clinical trials.

It is also interesting that the rate of bone metastasis has decreased significantly over the study period, in both prior-chemo and chemo-naive groups. One possible explanation for this is the development of bone-directed therapies. The widespread use of the FDA-approved bisphosphonate derivative zoledronic acid, and more recently RANK ligand inhibitor denosumab in men with metastatic osseous disease, may have directly contributed to this finding. Another possibility is that increased use of better radiographic imaging over the study period (eg, positron emission tomography/computed tomography) may have decreased false positive identification of osseous metastasis.

A major finding from the current analysis was the lack of standardized definitions for reporting baseline extent of disease in patients with metastatic CRPC. For example, 26.8% (106 of 396 studies eligible for analysis) failed to report the specific distribution of metastatic sites at baseline. Vague but overlapping nomenclature for metastatic sites was also used, including “non-osseous metastasis,” “soft-tissue metastasis,” “visceral metastasis,” “measurable disease,” “lymph node disease,” and others. A better understanding of the evolving natural history of prostate cancer will require detailed clinical phenotyping of patients, using standard definitions akin to those that have been adopted to define prostate cancer progression.[13]

There are several limitations to this study. First, the study data were collected from published phase 2 and 3 clinical studies between 1990 and 2012. Without individual patient data, analysis of potential confounding factors such as functional status and prior treatments was not possible. In addition, we were unable to control for changes in imaging technology and practice over the study period. Advances in scanning technology could have resulted in increased detection of metastasis in more recent years. However, whereas these advances occurred mainly in the first decade of our study, we still saw significant increases in nonosseous metastases between 2000 and 2012.

A second limitation to this study is that the study population involved patients with mCRPC enrolled on clinical trials, which may not be truly representative of all patients with prostate cancer. In fact, it is understood that patients who enroll in clinical trials tend to be “healthier” and “more fit” to receive treatment. Third, due to limitations set forth by therapeutic clinical trials (eg, organ function), analysis of specific endpoints may be affected. For instance, most studies exclude patients with liver dysfunction and poor functional status, which is often associated with higher disease burden. These factors may result in an underrepresentation of patients with liver or lung metastasis in such a study population. Also, analysis of brain metastasis is not possible because most trials excluded patients with central nervous system involvement. Fourth, there is a difference between the year that a clinical study was initiated (and conducted) versus when it was published. Thus, results observed in this study may reflect the trend of metastasis for earlier study periods (eg, prior to 1990), but not for the later years.

Conclusions

The changing landscape of therapy is temporally associated with a notable change in the pattern of metastasis in patients with mCRPC. The evolving patterns of metastatic disease are an important and understudied phenomenon that has important implications in treatment selection and prognosis of this lethal disease. It is crucial that future clinical trials of patients with mCRPC define patients with a uniform reporting of nonosseous metastasis. This will ensure that future studies yield truly accurate results in assessing the pipeline of new therapies that is now entering clinical testing.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES