Prostate-Specific Antigen as a Marker of Disease Activity in Prostate Cancer: Part 1
Prostate-Specific Antigen as a Marker of Disease Activity in Prostate Cancer: Part 1
ABSTRACT: Despite the impact of prostate-specific antigen (PSA) testing on the detection and management of prostate cancer, controversy about its usefulness as a marker of disease activity continues. This review, based on a recent roundtable discussion, examines whether PSA measurements can be used rationally in several clinical settings. Following radical prostatectomy and radiation therapy, prediction of survival by PSA level is most reliable in high-risk patients. PSA doubling time after radiation therapy is the strongest predictor of biochemical failure. PSA measurements have been associated with inconsistent results following hormonal treatment; reduced PSA levels may result from antiandrogen treatment, which decreases expression of the PSA gene, and therefore, the level of PSA production. In the setting of primary and secondary cancer prevention, PSA is important in risk stratification when selecting patients for studies. Part 1 of this two-part article, which concludes in the September issue, focuses on the physiology of PSA, its measurement and use in clinical practice, and its predictive value following radical prostatectomy and radiation therapy. [ONCOLOGY 16:1024-1051, 2002]
Prostate-specific antigen (PSA) testing has changed early detection and management of prostate cancer dramatically since its introduction into clinical practice in the early 1980s. Nevertheless, its usefulness as a marker of disease activity and its correlation with survival remain controversial because, in some disease settings, definitive data are lacking for its rational use. This review defines the clinical settings in which PSA can be rationally used as an indicator of disease activity and therapeutic effectiveness.
Because the natural history of prostate cancer can be long—20 to 50 years in some cases—it is impractical to use survival as the only test of a therapy’s usefulness in clinical trials. This is especially true of chemopreventive agents that are active at the earliest stages of a disease that may not be detectable for as long as 20 years. Even the longest-running trial of a potential chemopreventive therapy—the Selenium and Vitamin E Cancer Prevention Trial (SELECT)—is slated for only 12 years.
Using a marker of disease activity as an end point to demonstrate therapeutic efficacy is imperative for trials at various stages of the disease process. Because patients are encouraged to participate in early-detection programs for some cancers at age 40, the time from early detection to clinical diagnosis of recurrent disease may be as long as 10 years, and the time from diagnosis to death may be another 20 to 30 years after definitive therapy—timelines that would pose practical impossibilities for clinical trials.
In the Johns Hopkins series reported by Patrick C. Walsh, MD, and colleagues, the time from surgery to biochemical recurrence as measured by PSA ranged from 2 to 16 years, and from PSA recurrence to metastases averaged 8 years; the median time from metastases to death was 5 years (Figure 1). Chemoprevention aside, these intervals are too long when planning clinical trials to evaluate the effectiveness of therapies for prostate cancer. There is a need for markers as surrogate end points, and PSA has the potential to fill such a role.
The purpose of this review is to define the physiology and clinical use of PSA, and to determine whether PSA measurements can be used rationally as a marker of disease activity in the following clinical settings (Table 1):
• post-radical prostatectomy
• after radiation therapy for local disease
• during hormonal and other drug therapies
• in primary and secondary chemoprevention.
Part 1 of this two-part review, which concludes in the September issue, looks at the physiology of PSA and how its measurement can predict time to disease recurrence and, in a limited setting, survival of patients following radical prostatectomy and radiation therapy.
PSA is a serine protease in the kallikrein family of proteases; it is also called human kallikrein 3 (hK3). Produced in high concentrations by prostatic epithelium, PSA is secreted mainly into seminal fluid, where it dissolves the gel that forms after ejaculation by digesting the major gel-forming proteins, thereby resulting in increased sperm motility.
PSA is not a traditional tumor marker that is produced only by malignant tissue. Normal prostate tissue as well as hyperplastic and neoplastic tissue express PSA and, in fact, often produce more PSA protein than malignant prostate tissue. In prostate cancer, however, the architecture and polarization of epithelial cells are deranged, disrupting normal secretory pathways and causing PSA to "leak" or to be actively secreted into extracellular space and escape into the circulation. As a result, PSA is found in the serum in concentrations 30 times higher per unit weight of cancerous tissue than of normal tissue, and 10 times higher per unit weight of cancerous tissue than of the epithelial tissue in benign prostatic hyperplasia (BPH).[3,6,7]
This is the basis for the use of PSA levels in the detection of prostate cancer. However, there is much overlap between PSA values in BPH and prostate cancer, and various refinements have been made to improve the diagnostic value of this test. Some of these modifications will be considered later.
It is possible that PSA itself plays a modulating role in prostate cancer. Recent studies have suggested that men with high plasma levels of insulin growth factor (IGF)-I may have an increased risk of developing prostate cancer. IGF-I in the circulation is complexed to IGF-binding proteins, which also influence the biological activity of IGF. The major IGF binding protein is IGFBP-3, and it has been shown that PSA digests IGFBP-3. This may result in higher-concentrations of IGF-I and, consequently, stimulation of cancer growth.[10,11] On the other hand, PSA may suppress tumor growth by inducing formation of angiostatin.
PSA is also expressed in far lower concentrations by tissues other than prostate epithelium, including the periurethral glands, endometrium, normal breast tissue, breast tumors, breast milk, adrenal neoplasms, parotid gland, and renal cell carcinomas. Nevertheless, in the sera of women, concentrations detected by ultrasensitive assays are rarely higher than 0.5 ng/mL and usually range from 0 to 0.2 ng/mL.[13,14]
Much of the PSA released into the circulation forms complexes with protease inhibitors in plasma. Between 50% and 90% (typically 70% to 85%) of the total assayed PSA in the circulation is complexed to alpha-1-chymotrypsin. Trace amounts complex with an alpha-1-protease inhibitor and alpha-2-macroglobulin[16,17]; the remaining serum PSA is unbound or free. The proportion of complexed PSA delivered by cancerous tissue is higher than that of BPH tissue, possibly because PSA reaching the bloodstream directly can easily form complexes, whereas PSA that reaches circulation through extracellular space is susceptible to extensive proteolysis (it is then said to be "nicked" or "clipped") and is less likely to be bonded by endogenous protease inhibitors.
For this reason, the percentage of free PSA has been used to help increase the sensitivity of cancer detection when PSA is in the normal range (4.0 ng/mL or less) and the specificity when PSA is in the "gray zone" (4.1 to 10.0 ng/mL).[4,8] In this context, use of the ratio of "free PSA to total PSA" has increased the specificity of total PSA levels up to 20 ng/mL without undue loss of sensitivity.
Prostate-specific proteins were identified by several groups in the 1970s. In 1979, Wang and colleagues, who purified the antigen from prostatic tissue and demonstrated its relationship to prostate cancer, were the first to call the protein PSA.
Soon after PSA assays were developed, total serum PSA was shown to detect residual disease after radical prostatectomy and recurrence of tumor on long-term follow-up. In light of such evidence, PSA assays received approval from the US Food and Drug Administration (FDA) for monitoring therapy, and by 1988, PSA became widely used as a marker for prostate cancer. In the context of monitoring, the patient provides his own reference for the assay, but in the detection of prostate cancer, interassay variability became a concern.
The first widely used assay in the United States, Tandem-R (Hybritech), established a reference range for PSA—4.0 ng/mL or less—which was found in 97% of 207 apparently healthy men aged 40 years and older. This and subsequent studies confirming the validity of this reference range[21,22] provided the basis for the commonly used cutoff of 4.0 ng/mL as the normal total PSA in men age 40 years and older.
Most other tests were interpreted on the basis of this reference range, but not all assay methods measure the same PSA concentration, and each assay has its own reference range. Even when reference ranges are similar, tests differ greatly in their upper limits for BPH. The only two assays available in the United States until 1991—the monoclonal Tandem-R and the polyclonal Pros-Check (Yang Laboratories)—differed considerably, sometimes by a factor of two. Since that time, most assays have been thoroughly evaluated and modified to use reference samples as a standard.
Fine Tuning the Accuracy of PSA Tests
At present, most commercial PSA tests are sensitive monoclonal immunoassays that measure PSA-alpha-1-chymotrypsin and free PSA (total PSA). The affinity of the antibodies used in these assays against the different forms of PSA varies, sometimes resulting in a nonequimolar response to free and complexed forms of PSA. Some assays may preferentially detect more free PSA forms and thus overreport percent free PSA levels in patients with BPH, which may limit interpretation. Only assays that have an equimolar reaction with free PSA and PSA complexes show similar measurements. An assay that does not detect all forms of PSA on an equimolar basis is most accurate when it corresponds with a calibration standard.[4,23]
Recognizing these problems, many manufacturers of nonequimolar assays have modified their assays to produce equimolar responses to the complexed and free forms of PSA. In addition, through an international effort, the Expert Committee on Biological Standardization of the World Health Organization recently established standard reference preparations to validate and calibrate PSA assays.[25,26] One reference preparation is a standard containing 1 mg of free PSA, and the other is a preparation containing 1 mg of total PSA, including 0.1 mg free PSA and 0.9 mg PSA complexed with alpha-1-chymotrypsin. This 90:10 proportion is typical of the sera in men with prostate cancer. Applying this 90:10 calibrator to nine different PSA assays reduced the coefficients of variation among them from 28.3% to 9.5%. This standard has been used to calibrate the last five PSA assays approved by the FDA, and most manufacturers now use these standards.
Especially after radical prostatectomy, but also after other definitive treatments such as radiation or hormonal therapy, PSA levels have been found to correlate with disease activity. In early series from several institutions, patients with no detectable PSA postprostatectomy had no evidence of residual disease, whereas detectable postoperative PSA concentrations correlated with subsequent local recurrence or distant metastases—sometimes occurring as long as 1 to 3 years later. Similar findings were made after radiation and hormonal therapy.[13,27]
The limits of PSA detectability have been extended by ultrasensitive PSA assays that can accurately measure serum PSA levels as low as 0.01 to 0.001 ng/mL.[8,28] Ultrasensitive PSA assays have made it more difficult to distinguish between serum PSA concentrations produced by prostate tissue (including regrowth of cancerous prostate cells) and those produced by other tissues. But long-term follow-up of radical prostatectomy patients has indicated that postsurgical PSA levels as low as 0.01 to 0.07 ng/mL may predict recurrence.
In a recent study using PSA thresholds of 0.2, 0.3, and 0.4 ng/mL, the risks of a continued rise in PSA in the next 3 years were 49%, 62%, and 72%, respectively, leading the investigators to conclude that PSA levels of 0.4 ng/mL or greater best defined a recurrence prompting further therapy. Other large series have used 0.2 ng/mL to indicate disease recurrence. These studies have firmly established the clinical utility of PSA in the postoperative setting, and as these and other studies mature with longer follow-up, it is likely that clinical progression will be seen in virtually all patients experiencing any PSA recurrence.
Other PSA-Based Parameters
Additional measures that improve the predictive value of PSA after definitive therapy are the time to PSA recurrence and the rate at which PSA rises. Pound et al found that the timing of PSA recurrence after surgery, along with preoperative PSA level, accurately predicted disease-free survival and the pattern of recurrence—ie, local vs distant disease recurrence.
The concepts of PSA velocity and PSA doubling time have been used to characterize the rate at which PSA rises. PSA velocity requires at least three measurements over a 2-year period, or at least 12 to 18 months apart, to characterize the change in PSA level per unit of time. First introduced to improve the ability of PSA to detect prostate cancer, velocity differences were significant between men with cancer and men with BPH up to 9 years before diagnosis (Figure 2).
• PSA Doubling Time—When PSA becomes detectable after radical prostatectomy, it tends to increase exponentially, so that a plot of log PSA values against time is linear. PSA doubling time is calculated by dividing the slope of this plot into the natural log of 2 (0.693).[2,33] A doubling time of less than 6 months was found to suggest metastatic disease. Patel et al found doubling time to be a better predictor of risk of recurrence and time to clinical recurrence after radical prostatectomy than preoperative PSA, specimen Gleason sum, or pathologic stage.
In a large radical prostatectomy series (1,997 men) at The Johns Hopkins Hospital, the timing and natural history of disease progression of men with an elevated PSA after therapy were studied. The time to PSA elevation, Gleason score, and PSA doubling time were significant predictors of the probability of and time to development of metastases. A doubling time of 10 months provided the most statistically significant predictor of time to distant disease progression after an isolated elevation in PSA (Figure 3).