Commentary (Droller): Prostate-Specific Antigen as a Marker of Disease Activity in Prostate Cancer

September 1, 2002
Michael J. Droller, MD

Oncology, ONCOLOGY Vol 16 No 9, Volume 16, Issue 9

This second installment on prostate specific antigen (PSA) as a marker of disease activity and cancer cell viability in prostate cancer focuses on its role in monitoring the effects of a variety of therapies at different stages of the disease. In addition, the authors propose guidelines for studying the efficacy of new treatments in this setting.

This second installment on prostate specific antigen (PSA) as a marker of disease activity and cancer cell viability in prostate cancer focuses on its role in monitoring the effects of a variety of therapies at different stages of the disease. In addition, the authors propose guidelines for studying the efficacy of new treatments in this setting.

Processes Linked to PSA Changes

Because PSA is produced by normal prostatic epithelium and hyperplastic tissue as well as by prostate cancer cells, PSA elevations may occur as a result of inflammation ("leakage" into the extracellular spaces with subsequent escape into the circulation), benign prostatic hyperplasia ("bystander" leakage on the basis of volume of cells and their activity), and prostate cancer (derangement of normal glandular architecture with disruption of normal secretory pathways into the seminal fluid). This is important in assessing the meaning of PSA changes when initially considering a possible diagnosis of prostate cancer, and when considering primary nonsurgical treatment failure.

In the latter case, the use of PSA as a marker for disease and the effect of various therapies reflects both cell physiology and the particular characteristics of a prostate cancer cell-the assumption being that the machinery responsible for PSA production works in tandem with mechanisms that underlie the viability and activity of the cancer cell, its potential for progression, and its response to therapy. This becomes even clearer when prostate cancer has metastasized.

These assumptions belie the complexity of the correlations that may become manifest. For example, in hormone-sensitive prostate cancer, PSA may be inconsistent as a marker of disease activity and a measure of treatment efficacy in terms of effect solely on cell viability. Since antiandrogens reduce expression of the PSA gene, a dissociation between a fall in PSA level and its implications for cell cytotoxicity and consequent clinical efficacy may occur.

The best correlation may be seen when tumor cells are proliferating (despite castrate levels of testosterone) such that PSA levels in hormone-treated patients may need to be evaluated along with other outcome measures. Some investigators, therefore, have suggested that PSA may be of greater value as a measure of cancer cell treatment outcomes in hormone-insensitive prostate cancer. The increasing use of drugs that affect PSA independent of their effect on prostate cancer cells underscores the relevance of this suggestion.

PSA as a Surrogate of Survival

Investigators have also proposed that treatment responses manifesting as decreases in PSA of at least 50% correlate with improved survival. The durability of response, initial PSA level (at start of treatment), accuracy of the measurement indicating a 50% decrease in PSA level, and the effect of other variables on PSA production and its detection need to be considered in validating this correlation. Moreover, if PSA is used as the basis for therapeutic efficacy, it is important to consider the phase of a patient’s disease, how the stage and extent of disease are measured, and the manner in which a particular treatment may influence PSA production in association with the inhibition of tumor progression.

Phase II/III Transition Trials

On the basis of these considerations, as well as a concern that studies of therapeutic efficacy in advanced prostate cancer (or second-line treatment) are inadequate, the authors suggest that phase II trials be made more comparable to phase III trials, testing larger numbers of patients and demonstrating more substantive clinical benefit. Such so-called "phase II/III transition" trials would then be used to justify larger randomized (ie, true phase III) trials. Furthermore, they propose that disease phase at the time of investigational treatment should be specified and more precisely characterized, so that the effect of treatment can be assessed in the context of the patient’s prognosis at that particular time.

The use of a more precise "checkpoint" might be more valuable in assessing whether a new intervention actually prolongs survival or provides other benefits. When studies are not done in this way, the possible benefits of treatment may be obscured by variables that affect PSA production but are not relevant in validating true therapeutic efficacy.

Consideration of Disease Phase

It seems reasonable to assume that different therapies may affect different aspects of a disease process. Thus, consequent individualization of treatment may depend on recognition of a specific phase of cancer in a particular individual as well as the intrinsic biology of the disease, its potential course, the mechanisms at work during various phases of the disease, and the possibility that a particular treatment should be targeted to a specific mechanism in the pathogenesis of the disease. If efficacy is assessed outside of an appropriate (ie, possibly responsive) context, it may be unpredictable (at best) or prevented (at worst).

This view reflects an increasing appreciation of how complex the overall cancer process is, and how specific cancers and phases of cancer may relate to the individual. It is especially relevant in prostate cancer: Selected end points (traditionally, survival, and "disease-free" survival) may be influenced by the phase at which the disease is diagnosed, the sensitivity with which we are able to detect disease recurrence or persistence at a particular phase, and the means by which we can more accurately assess the influence of lead-time bias in interpreting treatment results.

If treatment efficacy is assessed by an effect on PSA level at each phase of the disease, the precise use of PSA as a surrogate of life expectancy becomes even more critical. The physiologic effect of a specific treatment on PSA in association with either early disease, disease recurrence, or metastatic disease (and disease progression in each of these phases) needs to be characterized and then correlated with disease-free and actual survival. Individualization of treatment and design of studies that take such individualization into account may then be more readily accomplished. The authors propose that PSA changes in such instances in either androgen-dependent or androgen-independent disease can be used in screening potentially active agents, permitting an assessment of efficacy regardless of whether the agent is cytotoxic or cytostatic.

Efficacy in Chemopreventive Approaches

Similar considerations may pertain to the role of PSA in gauging the efficacy of chemopreventive approaches. Here, too, the mechanisms by which various agents act on disease progression-whether or not in association with the mechanisms of PSA production-are important to understand, so that the relationship between the two and interpretation of treatment efficacy can be validated.

The authors question whether PSA can be a valid end point in assessing chemopreventive efficacy even before a prostate cancer can be detected clinically. Because such agents may affect PSA production by benign or hypertrophied tissue, the effect of treatment on PSA production by the cancer itself may be overwhelmed. If the cancer is not detected until fairly well along in its course (its development having been masked by an independent effect of treatment on "benign" PSA production), the issue becomes one of variability in cancer activity and its clinical expression.

In addressing this, the authors suggest that chemoprevention should be studied in patients at high risk for disease (eg, those with prostatic intraepithelial neoplasia on biopsy or those with strong genetic risk). However, the characterization of PSA as affected by treatment vs its expression by the cancer and the variability of each have not progressed sufficiently to permit confidence in its interpretation.

Furthermore, the very heterogeneity of cancers would appear to preclude the use of PSA as a reflection of efficacy in preventive treatment, because physiologic activity (vis--vis PSA production) and bioactivity (vis--vis cancer biologic potential as detected and expressed at its earliest phase) have not yet been sufficiently characterized.


In summary, this review provides an important perspective on the role of PSA in the assessment of various treatments and their specific efficacies in the course of various phases of prostate cancer. It seems appropriate to incorporate these suggestions into new study designs, and to further attempt to characterize the course of disease, define the role of PSA (or other markers) in reflecting pathogenesis, and interpret the efficacy of treatments in the context of pathogenesis and biologic potential for each phase of the disease.