Originally published as JCO Early Release 10.1200/JCO.2009.23.8063 on August 31 2009

This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Google Scholar Articles by Efstathiou, E. Articles by Logothetis, C. J. PubMed PubMed Citation Articles by Efstathiou, E. Articles by Logothetis, C. J. Related Articles Related Article Social Bookmarking                            What's this?

Informative Clinical Investigation: A Demanding Taskmaster

Clinic of Therapeutics, Alexandra Hospital, University of Athens, Athens, Greece, and David H. Koch Center for Applied Research of Genitourinary Cancers, The University of Texas M. D. Anderson Cancer Center, Houston, TX

Jeri Kim, Christopher J. Logothetis

Division of Cancer Medicine, Department of Genitourinary Medical Oncology, and David H. Koch Center for Applied Research of Genitourinary Cancers, The University of Texas M. D. Anderson Cancer Center, Houston, TX

In this issue of Journal of Clinical Oncology, Antonarakis et al¹ report the results of a neoadjuvant randomized, double-blind trial in selected patients with localized prostate cancer (Gleason score [GS] > 7, prostate-specific antigen [PSA] 15 ng/mL, clinical stage T2b, or any combination with > 45% risk of capsular penetration). Before prostatectomy, patients were treated for 4 to 6 weeks with placebo or with neoadjuvant celecoxib (400 mg orally twice a day), a selective cyclooxygenase (COX) -2 inhibitor. The investigators tested the drug's ability to achieve adequate concentration in the prostate and to modulate COX-2, its metabolites (prostaglandins), and biomarkers relevant to critical downstream signaling pathways implicated in carcinogenesis. The authors conclude that the short course of celecoxib in high-risk clinically localized prostate cancer does not modulate the biomarkers in a meaningful manner, and they caution against its further development. The relevance of these observations and their broader significance illustrate the opportunities and challenges preoperative models present in the study of prostate cancer.

The objective of the Antonarakis et al¹ study was to determine if celecoxib can reduce intratumoral prostaglandin levels, modulate critical downstream COX-1/COX-2 signaling, and alter the tumor phenotype. In their findings, however, the investigators observed no differences in the concentration of prostaglandin, proliferation (measured as Ki-67 expression) or apoptosis (p27^kip1, p21^waf1), or angiogenesis (factor VIII), despite demonstrating an increased concentration of celecoxib (albeit 50x lower than plasma concentrations) in the prostates of the treated group but not the control group. Their results do not support the epidemiologic observations of a reduction in prostate cancer risk and a modest delay in cancer progression reflected in decreased PSA velocity at 3 and 6 months, which was observed in men who experienced biochemical relapses after radiation or surgery.¹ Studies evaluating potential adjunctive use of celecoxib with radiation in localized prostate cancer or in combination with hormone therapy or chemotherapy in advanced disease are underway.¹ Understanding the discordance between these investigators' conclusions and the reported results from clinical trials that suggest that celecoxib favorably alters the course of prostate carcinogenesis and progression will guide its further development for prostate cancer prevention and therapy.

Several assumptions implicit in the design of their study may account for this discordance. These include the assumptions that the response to short-term COX-2 inhibition in localized prostate cancer mirrors the effect of longer therapy in the spectrum of biology from premalignant to advanced disease states, that the total prostate prostaglandin concentration will change as a consequence of therapy despite the known heterogeneity of individual prostate cancer foci, that the differences in expression of biomarkers will be detectable despite the heterogeneity of multifocal prostate cancers, and that the tissue prostaglandin concentration reflects the activity of COX-2. The validity of the latter two assumptions is of particular concern because of the heterogeneity of COX-2 mRNA expression in cancer and in the adjacent "normal" tissue they observed. Furthermore, prostaglandin concentration may be modulated differently within specific cancer foci, modulation that would not be detected in measurement of overall prostate prostaglandin concentration. In addition, COX-2 inhibition may modulate alternative eicosanoid pathways (eg, the lipoxygenase pathway), whose metabolites were not measured.

Important differences in biology and practical considerations lead us to the conclusion that specific trials should be designed to address the challenges of favorable-risk cancers (low volume), whereas separate designs are required for poor-risk (high volume) prostate cancers. Each of these disease states has both practical and conceptual implications on the design, execution, and interpretation of preoperative trial results. The study of favorable risk cancer (GS 7, PSA < 10 ng/mL, cT1c/2) is challenged by limited access to adequate tumor. To overcome this, we pursue an ex vivo biopsy strategy, simulating the sextant prostate biopsy scheme, which provides the opportunity to obtain tissue from prostatectomy specimens—even though this is a blinded biopsy, tumor can be obtained in approximately 70% of cases. These ex vivo biopsy samples can be used for laser capture microdissection coupled with cDNA microarray study for global gene-expression analysis.² The heterogeneous nature of prostate cancer demands that the cancer and control specimens be characterized in detail by individual foci. In a preprostatectomy multicenter phase II, randomized, placebo-controlled, double-blind study of finasteride, which is underway in patients with favorable-risk prostate cancer and designed to determine the frequency of the predetermined discriminating molecular signature (Gleason grade 3 v 4), the end points of the trial will be based on the marker expression in tissue microarrays constructed using the dominant (largest) tumor focus of the prostatectomy specimens (MDA03-1-03). Although having a uniform selection of tumor focus for marker studies affords reproducibility of the clinical trial design, we must be aware that this effort to assure uniformity is based on the untested assumption that the dominant tumor focus is representative of the clinically relevant biology of the entire cancer. This assumption may introduce an unappreciated bias in interpretation of the tissue marker outcomes.

In contrast, the neoadjuvant high-risk model approach, which is used to study cancers with lethal potential, provides abundant tissue, but interpretation is complicated by the heterogeneity of the treated primary tumor and no established criteria for comparison with control (due in part to morphologic changes). Therefore, accounting for and then understanding disease state–specific effects of drugs may be very informative. We distinguish between two types of presurgical studies, the preoperative model (eg, thalidomide),³ and neoadjuvant trials.⁴ The intention of the tissue-based preoperative models is to gain insight into molecular events in response to the genotoxic challenge of a candidate therapeutic agent, whereas the purpose of the neoadjuvant study is to observe the effects on the cancer, such as cytoreduction which is used to estimate the efficacy of the candidate therapeutic.³

Several strategies have been used to overcome the potential misleading effect of assumptions implicit in preoperative trial design. First, the high-risk category has sufficient cancer when compared with low-volume disease; however, a method for isolating similar tumor foci for comparison between treatment and control groups has yet to be determined. This is particularly relevant because response to a candidate therapeutic agent is likely to be different in specific cancer foci, given the known biologic heterogeneity of individual foci of cancer. Selecting the cancer focus or foci appropriate for molecular study should not only account for the heterogeneity but also be appropriate for the hypothesis to be tested. A case in point is the drug thalidomide, which has no significant cytotoxicity in prostate cancer in vitro yet has clinical efficacy especially in combination with a cytotoxic agent.³ We and others have attributed this effect to the tumor microenvironment. In the thalidomide study,³ we tested the concept that it exerts its antitumor activity by initially modulating the microenvironment and, indirectly, the epithelial compartment. Had the trial been designed with a longer treatment duration, we would have seen increased cytoreduction. On the other hand, this would have obscured the observation favoring the sequential modulation of the microenvironment and the epithelial compartment. We therefore would be unable to provide evidence supporting the hypothesis that initial modulation of the microenvironment accounts for the therapy benefit of thalidomide.⁵

Investigators have embraced the concept of the preoperative model in prostate cancer in hopes of overcoming the knowledge gaps that limit our ability to make rapid progress in treatment of the most common cancer among men in the United States. The report of Antonarakis and colleagues further establishes feasibility but also illustrates challenges. Although their findings are interesting, in our view, they are insufficient to dissuade further development of celecoxib for use in prostate cancer.

The investigators' results do illustrate the potential of preoperative trials in prostate cancer, but attention to detail is necessary. Preoperative models can be used in prostate cancer to gain insight into underlying biology when there is clinical evidence of efficacy in metastatic disease, to confirm the relevance of findings in experimental model systems to screen candidate therapeutic agents, and to identify candidate predictive markers. Investigators interested in the development of new therapies for prostate cancer have resorted to early clinical application with informative proof-of-principle studies to overcome the lack of model systems that portray human prostate cancer with fidelity. This approach can be used to gain confidence in the relevance of experimental findings to human prostate cancer, as a basis for rational combinatorial strategies, and as a discovery platform for hypothesis-generating associations across disease states. Interest and enthusiasm for this approach is justified; however, experience illustrates the need for a well-developed infrastructure and clearly framed testable hypotheses to avoid potentially misleading conclusions.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.

AUTHOR CONTRIBUTIONS

Conception and design: Eleni Efstathiou, Jeri Kim, Christopher J. Logothetis

Manuscript writing: Eleni Efstathiou, Jeri Kim, Christopher J. Logothetis

Final approval of manuscript: Christopher J. Logothetis

NOTES

See accompanying article on page 4986

REFERENCES

1. Antonarakis ES, Heath EI, Walczak JR, et al: Phase II, randomized, placebo-controlled trial of neoadjuvant celecoxib in men with clinically localized prostate cancer: Evaluation of drug-specific biomarkers. J Clin Oncol 27:4987–4994, 2009.

2. Tsavachidou D, McDonnell TJ, Wen S, et al: Selenium and vitamin E: Cell type- and intervention-specific tissue effects in prostate cancer. J Natl Cancer Inst 101:306–320, 2009.[Abstract/Free Full Text]

3. Efstathiou E, Troncoso P, Wen S, et al: Initial modulation of the tumor microenvironment accounts for thalidomide activity in prostate cancer. Clin Cancer Res 13:1224–1231, 2007.[Abstract/Free Full Text]

4. Febbo PG, Richie JP, George DJ, et al: Neoadjuvant docetaxel before radical prostatectomy in patients with high-risk localized prostate cancer. Clin Cancer Res 11:5233–5240, 2005.[Abstract/Free Full Text]

5. Dahut WL, Gulley JL, Arlen PM, et al: Randomized phase II trial of docetaxel plus thalidomide in androgen-independent prostate cancer. J Clin Oncol 22:2532–2539, 2004.[Abstract/Free Full Text]

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

Related Article

• Phase II, Randomized, Placebo-Controlled Trial of Neoadjuvant Celecoxib in Men With Clinically Localized Prostate Cancer: Evaluation of Drug-Specific Biomarkers
  Emmanuel S. Antonarakis, Elisabeth I. Heath, Janet R. Walczak, William G. Nelson, Helen Fedor, Angelo M. De Marzo, Marianna L. Zahurak, Steven Piantadosi, Andrew J. Dannenberg, Robin T. Gurganus, Sharyn D. Baker, Howard L. Parnes, Theodore L. DeWeese, Alan W. Partin, and Michael A. Carducci
  JCO 2009 27: 4986-4993 [Abstract] [Full Text]

This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Google Scholar Articles by Efstathiou, E. Articles by Logothetis, C. J. PubMed PubMed Citation Articles by Efstathiou, E. Articles by Logothetis, C. J. Related Articles Related Article Social Bookmarking                            What's this?