HER2 in Prostate Cancer--a Viable Target or Innocent Bystander?

The promise of molecular medicine is that clinical outcomes will be improved by directing therapy toward the mechanisms and targets associated with the growth of an individual patient's tumor. In practice, the targets are not static and change as the disease progresses (1–3). For prostate cancer, the framework to address this dynamic is provided by a model that considers the disease as a series of states from diagnosis to death (Fig. 1). Therapeutic aims and trial hypotheses are defined by determining the clinical and biologic factors associated with tumor growth at the state a given patient resides. Factors determined to be present on the majority of cancers within a state can be targeted without profiling an individual patient's tumor. Those present at a lower frequency, or that change over time, would require individual profiling so that those patients most likely to benefit from a specific approach are treated (3). The paradigm for therapy directed to a biologic determinant on a solid tumor is exemplified by trastuzumab (Herceptin™; Genentech, South San Francisco, CA), an antibody that targets the extracellular domain of the HER2 gene product (Her-2), in advanced breast cancer. Trastuzumab is believed to induce growth arrest by decreasing the expression of activated Her-2 receptors on the cell surface. Antibody-dependent cell-mediated cytotoxicity may also contribute to tumor response. Antitumor effects correspond with the level of expression: Tumors that do not express Her-2 or do so at low levels are insensitive to trastuzumab (4). In clinical trials, trastuzumab was active as a single agent and was shown to prolong life when it was added to two standard chemotherapy regimens in patients with advanced disease (5). The Her-2 target was credentialed by the demonstration that clinical benefit was restricted to patients with tumors that overexpressed the HER2 gene product (6), which led to the approval by the U.S. Food and Drug Administration (FDA) of a diagnostic test for expression that is now considered to be part of the “routine” management of breast cancer (7,8). The cost and complexity of developing drugs often mandate that clinical testing focus on one disease at a time. It is assumed that a favorable result in one disease will be applicable to patients with other tumor types. It is now recognized in breast cancer that amplification of the HER2 gene occurs in 25% of cases leading to the overexpression of Her-2 at the message and protein levels (9). Profiling of Her-2 expression provides prognostic information, is treatment predictive, and can be used to select therapy. Indeed, Her-2 is more than a just a marker, because active signaling through this determinant contributes to the pathogenesis and clinical aggressiveness of the disease (5,10). But does Her-2 have the same role in prostate cancer? Is it present on tumors representing different states of the disease? Is a given Her-2 expression profile, prognostic, treatment predictive, or of no clinical significance? In this issue of the Journal, the report by Signoretti et al. (11) adds to the growing, albeit contradictory and confusing, literature evaluating Her-2 expression in prostate cancer. Why the confusion? One review (12) cited “positive” results in 0%–100% of samples using immunohistochemistry, and others (12–14) have cited a range of outcomes with fluorescent in situ hybridization (FISH). Part of the disparity is due to differences in tissue selection and preparation, the reagents, the techniques applied, and the definitions of “positive.” It is within this context that the report by Signoretti et al. (11) is important because it includes an analysis of Her-2 at the DNA, RNA, and protein levels and evaluates tumors representing different clinical states (Fig. 1). Two FDA-approved diagnostic tests were used: the Her-2 immunohistochemistry assay by Dako (Dako Corp., Carpinteria, CA) and a FISH assay for amplification (Ventana Medical Systems, Inc., Tucson, AZ), as well as other assays. The expression of the androgen receptor, prostate-specific antigen (PSA), and Ki-67 was also determined. Three cohorts of patients were studied representing two disease states: 1) localized tumors that were untreated or that had been exposed to 3 months of androgen deprivation and 2) progressing castrate metastatic lesions from bone (Fig. 1). The results show how Her-2 expression levels vary across the clinical spectrum of prostate cancer and how different reagents and outcome measures can lead to dramatically different conclusions when analyzing the same samples. These findings have been suggested by others (14–18) but are clarified here. Using an absolute scoring system and defining a positive result as 10% of cells positive with an expression level of 2+ or 3+, 25% (17 of 67) of untreated primary tumors, 59% (20 of 34) of localized tumors after neoadjuvant hormone therapy, and 78% (14 of 18) of castrate metastatic tumors overexpressed the Her-2 protein (11). Our group obtained similar results; 20% (nine of 45) of localized untreated tumors, 68% (23 of 34) of localized tumors after androgen deprivation (P = .0001), and 80% (16 of 20) of castrate metastatic lesions were “positive” in a retrospective analysis (19), and 6% of untreated primary tumors and 42% of castrate metastatic samples were positive in a prospective analysis of patients being considered for treatment with trastuzumab (20). Of nine cases where the untreated primary cancer was negative for Her-2 and a castrate metastatic lesion was available, three (33%) were Her-2 positive (20). These results are similar to those seen with LAPC4 prostate cancer xenografts in which a consistent increase in Her-2 protein levels was observed with progression to androgen independence (21), and forced overexpression of Her-2 in the LNCaP tumor model was observed to be sufficient to confer androgen-independent growth in vitro to androgen-dependent cells (22). However, the immunohistochemistry and RNA expression data do not provide insights into the functional significance of the findings. Signoretti et al. (11) also used a relative scoring system that compared the intensity of staining on tumor cells relative to basal cells by optical imaging. With the use of the relative scoring system, 51% of untreated patients were “high” expressors versus 25% with the use of the absolute scoring method. Which is “correct”? The relative scoring system assumes that the staining of basal cells does not change following androgen deprivation. This may be difficult to assess because the amount of cytoplasm in a basal cell is small and the morphology of the basal cell changes after androgen deprivation (23). This also makes grading problematic. An additional difficulty of the analysis is the validation with tumors that were preselected to show overexpression, which tends to overestimate sensitivity and specificity. Ultimately, the significance of the findings and whether 25% or 51% is the proportion of tumors that are “abnormal” require an external validation of the “outcome” measure with a clinically significant end point, such as time to recurrence, death from disease, or sensitivity to specific therapeutic intervention. When the results are tabulated, it may turn out that different definitions of “abnormal” predict for different outcomes. The lack of an association between Her-2 expression and PSA recurrence may reflect the sample size studied. What is the role of Her-2 in prostate cancer progression? Current dogma in breast cancer suggests that amplification at the DNA level is most predictive of outcomes (5). With the exception of one study that showed amplification in 42% of cases (14), amplification has been decidedly uncommon in prostate cancer, regardless of the technique employed (11,13,19,24). The staining patterns in breast cancer and prostate cancer are different. The question remains whether the increase in expression is functionally significant or a bystander effect in prostate cancer. Nevertheless, the finding of different levels of expression at different points in the disease suggests a functional role. The authors (11) suggest that the increase in expression in localized tumors following androgen deprivation may contribute to cell survival in the absence of proliferation, while the increase in the proportion of tumors that overexpress Her-2 in castrate metastatic disease may contribute both to survival and to the increased proliferation index of these lesions. The finding of a coincident expression of the androgen receptor and downstream target genes may reflect ligand-independent activation of the receptor and suggests a role for targeting the androgen receptor as well. Additional questions are whether Her-2 that is activated by ligand-driven heterodimerization with other HER family members contributes to prostate cancer progression and whether inhibition of the Her-2 component of the complex arrests the growth of tumors driven by ligand-activated tyrosine kinases such as the epidermal growth factor receptor, which has also been shown to be overexpressed in advanced prostate cancer (25,26). The report by Signoretti et al. (11) addresses one marker, Her-2. The capabilities of genetic profiling now permit the simultaneous evaluation of thousands of determinants. How will we reduce these determinants to practice? We will do so with a careful, systematic approach that includes an adequate number of patient samples representing different stages in the disease, testing procedures that are reproducible and with standardized reagents, and with reports that include the details of the scoring system, the sensitivity, and specificity of the assay (27). Most important is that any outcome measure must be validated with the use of clinically significant events, such as time to recurrence, death from disease, or sensitivity to a specific compound. A clinical trial in prostate cancer with trastuzumab alone has been performed, and no objective responses (PSA declines or regression of measurable disease) were observed in patients with progressive castrate metastatic disease (20). This does not exclude a potential benefit for other therapies directed toward the Her-2 signaling cascade, such as ansamycins which produce a rapid reduction in the level of Her-2 expression in cell lines that overexpress Her-2 (28), antisense approaches (29), and novel antibodies directed toward the dimerization of Her-2 and conjoiners (30). Schematic of clinical states model of prostate cancer. Arrows represent the cancer states from the cohorts of patients studied in Signoretti et al. (11) The figure was modified from reference (3) with permission.