Does P-glycoprotein predict response to chemotherapy, and if so, is there a reliable way to detect it?

Abstract

The development of resistance to cancer chemotherapeutic drugs remains a major obstacle in the treatment of cancer patients. Medical oncologists are all too familiar with the phenomenon of tumor recurrence following an initial clinical response to chemotherapeutic drugs. Determining mechanisms of drug resistance is critical to the development of rational therapeutic strategies to overcome or prevent drug resistance. One mechanism of resistance which has been well characterized is that of multidrug resistance (MDR) (1). Tumor cells that have the MDR phenotype are generally resistant to such natural products as the vinca alkaloids, anthracyclines, and podophylotoxins. This form of drug resistance has consistently been associated with increased expression of a 170-kilodalton molecular weight plasma membrane protein termed the P-glycoprotein (2). P-glycoprotein functions as an energy-dependent drug efflux pump that reduces intracellular drug accumulation, thereby conferring resistance to many different drugs. P-glycoprotein is encoded by a small family of closely related genes, and of the two genes in humans, only the mdrl gene (also known as PGfYl) causes the MDR phenotype (3). In order to evaluate fully the role of P-glycoprotein in clinical drug resistance, it is important to develop assays that are specific for the mdrl protein and yet sensitive enough to detect the minute concentrations of protein present in small tissue specimens. An added confounding factor is the fact that P-glycoprotein is expressed in certain normal tissues; thus assays not only must be specific and sensitive, but also must be able to discriminate between normal and neoplastic tissue. Initial studies analyzing the role of P-grycoprotein in human tumor specimens utilized biochemical analyses of bulk tumor tissue. While these initial analyses were considered to be specific for P-glycoprotein, they were relatively insensitive and did not discriminate between normal and neoplastic tissue. In this issue of the Journal, Verrelle and colleagues (4) analyze the expression of P-glycoprotein in locally advanced breast cancer, using an immunohistochemical means of detection. These investigators used the monoclonal antibody C494, which is considered specific for the mdrl protein (5). This study of 20 patients found a surprisingly high incidence of P-grycoprotein expression in tumor cells from patients with primary, locally advanced breast carcinoma. Eighty-five percent of the patients expressed P-glycoprotein in at least some of their tumor cells. Using a semiquantitative scaling system that considered both the number of positive cells and the intensity of staining, they demonstrated a significant association between response to chemotherapy and time to treatment failure (ie, disease progression or recurrence) in seven patients who had a high degree of staining in the majority of tumor cells. While this semiquantitative approach found a subset of patients with a poor prognosis, a number of factors must be considered in order to prudently interpret their data. One primary concern is the small number of patients in this study. Performing a subset analysis on a total of 20 patients is generally fraught with a number of pitfalls. For example, it is difficult to achieve equivalence between two small patient subsets regarding other pertinent prognostic factors (such as proliferative capacity of the tumor) that might influence therapeutic outcome. Obviously, more patients will need to be analyzed prospectively in future studies to determine the role of P-glycoprotein in predicting relapse following chemotherapy. A second concern is the very high incidence of P-glycoprotein expression detected in these patients. Not only does this result differ from the relatively low incidence of P-glycoprotein expression in breast cancers described by other investigators, who used different antibodies (6,7) or other methods such as the detection of RNA (8,9), but it also raises the issue of the signal-to-background ratio in attempting to develop a quantitative immunohistochemical assay. If one analyzes the tumor specimens as being either positive or negative for P-glycoprotein expression, then only three patients are in the negative category and P-glycoprotein is rendered useless as a prognostic factor. Nonetheless, Verrelle and colleagues have demonstrated that patients in whom the majority of tumor cells strongly express P-glycoprotein do less well following treatment with chemotherapy than patients in whom P-glycoprotein detection in tumor cells is weak or absent. Attention to the refinement of immunohistochemical techniques in order to increase the signal-to-background ratio will be critical to achieving quantitation of P-glycoprotein in future studies. A similar study of P-glycoprotein in locally advanced breast cancer was recently reported by Ro et al (10). Using the C-219 monoclonal antibody and different fixative techniques, these investigators observed the frequent expression of P-glycoprotein in tumors following pre-operative chemotherapy. Such expression was significantly associated with a poor response to chemotherapy. Taken together, these two studies of P-glycoprotein expression in locally advanced breast cancer (4,10) suggest the prospect of predicting therapy failures by pretreatment testing for P-glycoprotein expression. Positive results would provide a rationale for alternative chemotherapy or for the circumvention of MDR by using chemosensitizers such as verapamil (11).