Rising PSA After Local Therapy Failure: Immediate vs Deferred Treatment

Introduction

The “prostate-specific antigen (PSA)–era” (1988 to present) has dramatically altered the epidemiology of prostate cancer in the United States and in many other industrialized countries.[1] Although the prevalence of prostate cancer has fallen somewhat since its peak in the early 1990s, the American Cancer Society still estimates that approximately 179,000 new cases will be diagnosed in 1999.[2]

An unprecedented stage migration has accompanied this large shift in incidence. The Surveillance, Epidemiology and End Results (SEER) Program of the National Cancer Institute noted a 52% decline in the rate of distant metastatic (stage D) prostate cancer between 1990 and 1994.[3] At the same time, the rate of diagnosis of localized disease skyrocketed. Our Department of Defense Center for Prostate Disease Research database at Walter Reed Army Medical Center (WRAMC) found that the incidence of localized prostate cancer (stages A and B) increased from approximately 50% of cases in 1988 to more than 75% of cases by 1996.[1]

Along with this change in stage distribution has come a change in treatment patterns. The SEER program found that rates of radical prostatectomy rose from 17.4 cases per 100,000 in 1988 to 54.6 cases per 100,000 in 1992.[3] By 1992, 36.6% of patients with locoregional disease underwent radical prostatectomy, and 32.3% received radiation therapy.

Furthermore, there has been a shift in the age-adjusted rates of these treatments. Most notably, there was a three- to fourfold rise in the rate of radical prostatectomy in men 45 to 59 years old, and a two- to threefold rise in men 60 to 69 years of age.[3] Rates of radiation therapy also increased one- to twofold in 45- to 79-year-old men.

In the late 1990s, clinicians are now seeing the effects of the boom in the diagnosis and localized treatment of prostate cancer of the early 1990s. A large number of generally younger men who were treated for clinically localized prostate cancer have experienced a recurrence of their disease. Figure 1 illustrates the problem clinicians are facing.

With more than 50,000 men per year developing a PSA-only recurrence (indicated only by an elevated PSA level, as will be discussed in the next section), it is obvious that this is a key issue for urologists, radiation oncologists, medical oncologists, and, perhaps most importantly, the patient and his family.

Assessment of PSA-Only Recurrence After Prostatectomy

The PSA level at which to define treatment failure after radical prostatectomy varies in the literature. Some series have used any detectable level; others, a single value > 0.4 or 0.5 ng/mL; and still others, two consecutive values ³ 0.2 ng/mL. At our hospital, employing the Abbott IMx assay, we use the criterion of two values ³ 0.2 ng/mL, or any single value ³ 0.5 ng/mL.

In clinical practice, it generally is quite obvious when radical prostatectomy patients develop a PSA-only recurrence because their PSA becomes detectable and continues to rise. The use of an ultrasensitive PSA assay may result in the identification of relapsing patients 1 to 2 years earlier than can be achieved with a conventional assay.

The timing of the rise in PSA level after surgery also is important. Patients whose PSA never falls to an undetectable level in the postoperative period generally are considered to have systemic disease. However, some of these men who do not attain an undetectable PSA after surgery do respond to salvage radiation to the prostatic bed. This suggests that systemic disease is not universal in this setting. Likewise, a PSA level that rises rapidly during the postoperative period may be indicative of metastatic disease. Patients whose PSA level remains undetectable for long periods (1 to 4 years) and then gradually rises are considered to have local disease recurrence.

 After Radiation Therapy

Until recently, the definition of PSA-only recurrence after radiation therapy was widely debated. In 1997, the American Society for Therapeutic Radiology and Oncology (ASTRO) convened a consensus panel to determine guidelines for PSA-only recurrence (biochemical failure) after radiation therapy.[4] The panel agreed on the following four guidelines:

1. Biochemical failure is not a justification per se to initiate additional treatment. It is not equivalent to clinical failure. Biochemical failure is, however, an appropriate early end point for clinical trials.

2. Three consecutive increases in PSA level provide a reasonable definition of biochemical failure after radiation therapy. For clinical trials, the date of failure should be the midpoint between the postirradiation nadir PSA level and the first of three consecutive rises. (The use of three, rather than two, consecutive values reduces the likelihood of falsely characterizing a “bouncing” PSA level as a biochemical failure. This phenomenon results when sequential determinations of PSA level show one or two rises, followed by a fall and a subsequent failure to rise again.)

3.  As yet, no definition of PSA-only recurrence has been shown to be a surrogate for clinical progression or survival.

4. Nadir PSA level is a strong prognostic factor, but no absolute level is a valid cut-off point for identifying successful and unsuccessful treatments. Nadir PSA level is similar in prognostic value to pretreatment prognostic variables.

Prediction of PSA-Only Recurrence

Because early adjuvant therapy may be beneficial to patients with localized disease in whom treatment is destined to fail, many studies have evaluated a variety of prognostic variables in an attempt to identify individuals who are at high risk of disease recurrence after surgery. The following variables have shown a significant correlation with PSA-only recurrence[5-15]: pretreatment PSA level; prostatic acid phosphatase level; prostatectomy specimen Gleason sum; pathologic stage; tumor volume; endorectal coil magnetic resonance imaging results; DNA ploidy; race; and, more recently, molecular biomarkers, such as p53, bcl-2, and Ki-67. Recently, some investigators have combined prognostic variables into models or equations that can be used to predict the likelihood of recurrence.

Johns Hopkins Model

Partin et al[10] at The Johns Hopkins Hospital were the first group to develop a simple biostatistical model equation that categorized post–radical prostatectomy patients into three risk groups (low, intermediate, and high risk) based on their likelihood of serologic failure. Many preoperative and pathologic variables were analyzed. However, after multivariate regression analysis, only three variables were included in the final model to select adequately for risk stratification after surgery. Sigmoidal transformation of PSA level (defined in equation in middle column), prostatectomy Gleason sum, and specimen confinement (margin status) were incorporated into an equation that calculated the relative risk of recurrence (R_w) as: R_w = (0.061 × PSA_ST) + (0.54 × postoperative Gleason sum) + (1.87 × specimen confinement)

Specimen confined = organ confined or extracapsular extension with negative margins.ecimen confined = organ confined or extracapsular extension with negative margins.

Patients were then stratified into three risk groups depending on their calculated value of Rw. This model employed traditional variables that are assessed at most institutions, making this form of risk assessment a practical clinical tool that can be used in decisions concerning adjuvant therapy. The model allows those patients at high risk for recurrence to be identified shortly after surgery, while their tumor burden is theoretically at a minimum.

Department of Defense Center for Prostate Disease Research Models

Our research group at the Department of Defense Center for Prostate Disease Research at WRAMC and the Uniformed Services University also has been working to develop prognostic models that will predict PSA-only recurrence after radical prostatectomy. Although the Johns Hopkins model[10] has provided a great start, it was developed for patients with clinical stage B2 (T2b or T2c) and may not be applicable to the majority of patients who are being treated in the late 1990s.

Using data from 378 patients of all clinical stages at WRAMC and data from 91 patients in a separate hospital validation cohort, we performed similar modeling using traditional prognostic factors.[15] The prognostic variables that significantly correlated with disease recurrence were incorporated into a model equation that calculates the relative risk of recurrence (R_r) as: R_r = exp [(0.51 × race) + (0.12 × PSAst) + (0.25 ×postop Gleason sum) + (0.89 × organ confinement)]

Race was defined as “1” if the patient was African-American or “0” if he was Caucasian or another race. Sigmoidal transformation of PSA (PSAst) was calculated using the equation:

PSA_st = 10 / (1+e⁶.^{8704-(0.9815 PSA level)})

The postoperative Gleason sum was defined as a continuous integer value (range, 2 to 10). Organ confinement was defined as “0” if the tumor was organ-confined (no extraprostatic extension) or “1” if the tumor was non–organ-confined (extraprostatic extension and/or positive margins).

Table 1 shows the 3- and 5-year Kaplan-Meier disease-free survival rates for the three risk groups for both the WRAMC model cohort of 378 patients (top panel) and the validation cohort of 91 independent patients (bottom panel). We have placed this traditional model equation into our urologic clinic local area computer network. Each clinician can enter patients’ race, PSA level, Gleason sum, and pathologic stage into the Microsoft Excel program postoperatively, and the program will automatically calculate the R_r and show the recurrence information that appears in Table 1. We can print this information and use it as an aid for counseling patients with regard to adjuvant therapies. This model is now available for use on our World Wide Web home page (www.cpdr.org).

In addition to this recurrence model using traditional prognostic factors, we developed a model to predict recurrence after radical prostatectomy using traditional clinical and pathologic variables combined with the results of molecular biomarker assays (p53 and bcl-2 immunohistochemistry of radical prostatectomy specimens).[16]

Our models are initial attempts to combine prognostic factors and take advantage of advances in hospital-based desktop computers so as to improve patient care. However, large, multicenter, prospective studies are needed to fine-tune existing prognostic models and to develop similar models for patients who receive external-beam radiotherapy or brachytherapy.